WEBVTT
00:00.570 --> 00:03.511
[SPEAKER_01]: Everyone can spot if something is made with AI.
00:03.531 --> 00:07.512
[SPEAKER_01]: Like it just like look at two or three websites, you know they are AI generated.
00:07.652 --> 00:10.253
[SPEAKER_01]: And like the same is with copy, the same is with imagery.
00:10.613 --> 00:17.676
[SPEAKER_01]: So I feel that in the long term brands that are selling genuine experiences are the ones that are gonna win.
00:20.057 --> 00:26.999
[SPEAKER_00]: You're listening to the keep optimizing podcast to increase your traffic, improve your conversion rates and grow your profits.
00:29.960 --> 00:35.125
[SPEAKER_02]: Hello and welcome on Chloe Thomas the host of this e-commerce marketing focus podcast.
00:35.245 --> 00:43.372
[SPEAKER_02]: If you're not familiar well at keep optimising we release five episode mini series at a time that dive into one e-commerce marketing topic.
00:43.492 --> 00:51.840
[SPEAKER_02]: It could be a channel like email or SEO or Facebook ads or it could be a big topic like acquisition retention or analytics.
00:52.320 --> 01:06.887
[SPEAKER_02]: In each of the episodes, new ones released on Wednesdays, I get to interview a different expert to explore the latest device on making it work for you and then after each five episodes, mini series, we take a week off to give you time to take action on what you've learnt.
01:07.567 --> 01:17.613
[SPEAKER_02]: Well, we have covered all sorts of angles on conversion rate improvements in this latest series, but one glaring emission has been a dive into how AI can help you.
01:18.053 --> 01:20.154
[SPEAKER_02]: So that's what we're going to be covering in this episode.
01:20.654 --> 01:24.876
[SPEAKER_02]: We're going to be talking about where you should use AI, where you shouldn't, how to
01:26.017 --> 01:32.985
[SPEAKER_02]: segment your your CROW activity to get the best out of AI and lots of other really interesting bits and pieces.
01:33.145 --> 01:40.192
[SPEAKER_02]: If you've been struggling to work out how to get AI to deliver for your e-commerce business and you're going to really enjoy this upcoming conversation.
01:40.573 --> 01:45.077
[SPEAKER_02]: And remember, there won't be an episode next week because we're be between series.
01:45.838 --> 01:51.381
[SPEAKER_02]: So you, that will give you time to catch up on any of this that you've missed, share with your team and take action.
01:51.841 --> 02:01.185
[SPEAKER_02]: Make sure you list them right at the end of the episode to find out what our next series is, but also because my guest shares some really great insight during the insider tips round as well.
02:03.605 --> 02:07.948
[SPEAKER_02]: In this episode, I'm chatting with conversion rates expert Alena Costover.
02:08.348 --> 02:18.495
[SPEAKER_02]: Alena is head of marketing at UXify, an AI native digital experience platform that helps websites to achieve flawless user experience and double engagement.
02:18.875 --> 02:19.696
[SPEAKER_02]: Hello, Alena.
02:20.216 --> 02:20.877
[SPEAKER_02]: Hi, Chloe.
02:21.377 --> 02:22.538
[SPEAKER_02]: Great to have you here.
02:22.658 --> 02:25.720
[SPEAKER_02]: How did you get into the world of conversion rates?
02:26.600 --> 02:36.913
[SPEAKER_01]: Well, in all honesty, I feel that the conversion rate optimization world found me, I've been heading marketing like small marketing teams for the past 10 years.
02:37.814 --> 02:43.300
[SPEAKER_01]: And in my role, I was responsible for the site creation, like the...
02:44.101 --> 03:12.948
[SPEAKER_01]: making sure that the site is optimized and converting, so as such, I had to learn all the tricks around conversion rate optimization, and then on the other hand, being part of companies that are selling to e-commerce, I had the privilege to talk with e-commerce owners, marketers, and agencies that are solving exactly that problem and optimizing for better conversion and see what works,
03:14.128 --> 03:19.073
[SPEAKER_01]: So this is how my past actually crossed with convertible rate optimisation.
03:20.114 --> 03:20.755
[SPEAKER_02]: Excellent.
03:20.815 --> 03:34.288
[SPEAKER_02]: Well look, we're going to get deep into a topic which we haven't really got very into weirdly across this series, which is how AI can help us improve our conversion rates, which I suspect there's quite a lot it can do.
03:34.848 --> 03:37.050
[SPEAKER_02]: know, I know you are deep in what it can do.
03:37.130 --> 03:44.815
[SPEAKER_02]: So what's the best way to deploy AI into use AI to help us improve the conversion rates on our e-commerce site?
03:46.405 --> 03:51.686
[SPEAKER_01]: I think the best way is to start small and always start by measuring things.
03:52.226 --> 03:57.607
[SPEAKER_01]: So let's give an example for site speeds, so how fast your site could be.
03:58.147 --> 04:02.628
[SPEAKER_01]: Like the most natural way you do is measure your performance.
04:02.648 --> 04:09.069
[SPEAKER_01]: You can go to like Google page speed insights and actually check if you have problem or not.
04:09.489 --> 04:13.450
[SPEAKER_01]: And once you know that there is a problem, then there is a lot of you can do.
04:13.830 --> 04:23.198
[SPEAKER_01]: you can see what the problems are and I typically or divide the problems of can they break your site or will they make your money?
04:23.838 --> 04:29.063
[SPEAKER_01]: If the problems are very critical and they can break your site, then I would say
04:29.843 --> 04:44.175
[SPEAKER_01]: Use an AI assist at a approach where AI can help you and enhance what you're doing, but if the opportunity is clear and it can make you more money and there's no risk of breaking your website, then I'll say go with the AI solution.
04:44.976 --> 04:55.104
[SPEAKER_01]: And maybe another point on that, activities that usually require a lot of data and a lot of analysis of data and quick analysis of data,
04:57.326 --> 05:08.444
[SPEAKER_01]: elements that you can give AI to optimize because it's just faster in thinking and cognition and like understanding like the patterns and reacting in real life.
05:09.238 --> 05:11.839
[SPEAKER_02]: I love how you divide that up into different bits.
05:11.919 --> 05:21.040
[SPEAKER_02]: We've got the measurement part, obviously, fairly, I think most people have now used AI to create some kind of dashboard or other.
05:21.400 --> 05:23.121
[SPEAKER_02]: So we know we're going to be doing that.
05:23.141 --> 05:25.041
[SPEAKER_02]: So we're paying better attention to the numbers.
05:25.461 --> 05:34.803
[SPEAKER_02]: And then the things we find out of that, are you doing the just let AI fix it, or are you doing the AI assisted root by going,
05:36.464 --> 05:44.749
[SPEAKER_02]: What's the risk here, basically, is what we're saying, is that how much are we talking about playing around with button colors, or are we talking about rewiring the check out?
05:45.409 --> 05:51.193
[SPEAKER_02]: And like clearly one of those, you want to do the assisted and one of those, you know, you can give over to it.
05:51.773 --> 05:57.317
[SPEAKER_02]: And does this mean we end up with kind of like a multi-part?
05:58.444 --> 06:09.493
[SPEAKER_02]: way of handling RAI, where we've got things which are more manual, where we're working with AI or not AI to fix, to measure, to amend and so forth.
06:09.813 --> 06:14.457
[SPEAKER_02]: And then we've got stuff that's just carrying on, crunching numbers, improving things.
06:14.497 --> 06:19.882
[SPEAKER_02]: It's kind of like almost like a two, two runway version of AI now.
06:20.722 --> 06:23.024
[SPEAKER_02]: There's no, not, not the AI of commercial optimization.
06:24.399 --> 06:46.696
[SPEAKER_01]: I would say is, at least from my point of view, and why I think that is a website, like there are a lot of things in the website, you can really live it to be self-optimized and have them be fully autonomous, side-speed, removing friction, personalization, optimizing images, things like that.
06:47.736 --> 06:57.300
[SPEAKER_01]: very few things can go wrong and usually like the knowledge that you have to acquire to do that yourself manually properly, it's quite vast.
06:57.781 --> 07:08.805
[SPEAKER_01]: So it's going to take you either a lot of time or a lot of resources to solve those problems and those problems can and are currently very easily fixed with just letting AI do the job.
07:09.405 --> 07:12.347
[SPEAKER_01]: But then on the other hand, there are other items which
07:13.167 --> 07:28.173
[SPEAKER_01]: Some people made my disagree with me, but I feel that website microcopy, website branding, and how your website is looking, those still have to sit with real person.
07:28.213 --> 07:32.315
[SPEAKER_01]: Like, obviously it can be AA assisted, but it's ultimately your brand.
07:32.415 --> 07:35.837
[SPEAKER_01]: Like, it's a bit counterintuitive for me to,
07:36.825 --> 07:45.236
[SPEAKER_01]: give away that privilege to show your personality in your brandy front of your audience and just let AI do it for you.
07:45.676 --> 07:46.998
[SPEAKER_01]: It can work, obviously.
07:47.399 --> 07:52.225
[SPEAKER_01]: But I find it a short-term win instead of long-term,
07:53.853 --> 07:54.073
[SPEAKER_01]: with.
07:54.594 --> 08:00.220
[SPEAKER_02]: Well, yeah, and I think that makes so much sense, not least because brand is where you want to stand out.
08:00.320 --> 08:03.243
[SPEAKER_02]: And AI works by looking at the averages of everything.
08:03.403 --> 08:03.903
[SPEAKER_02]: Exactly.
08:03.923 --> 08:11.511
[SPEAKER_02]: Nothing is like, you just end up, if we get let AI control all our brands, we'd all end up beige in a fancy way.
08:12.492 --> 08:19.816
[SPEAKER_01]: And I feel that even nowadays, like, purchasers are getting even more critical.
08:19.836 --> 08:23.898
[SPEAKER_01]: Like, everyone can spot if something is made with AI.
08:23.938 --> 08:26.539
[SPEAKER_01]: Like, you just look at two or three websites.
08:26.599 --> 08:28.480
[SPEAKER_01]: You know they are AI generated.
08:28.620 --> 08:31.942
[SPEAKER_01]: And the same is with copy, the same is with image imagery.
08:32.302 --> 08:36.204
[SPEAKER_01]: So I feel that nowadays consumers are getting much more critical.
08:36.704 --> 09:01.363
[SPEAKER_01]: and like brands like, I don't know if you have heard of them like, hot-gore pickles or their girls that are just selling like pickles like as basic as that, but just the brand and the narrative that they build around it, it's so genuine and relatable that that's what sells and they feel that ultimately in in the long term brands that are selling genuine experiences, are the ones that are going to win.
09:02.224 --> 09:03.325
[SPEAKER_01]: So again,
09:06.007 --> 09:09.808
[SPEAKER_01]: A lot of people say, like, if nothing comes in, nothing is going to come out.
09:09.868 --> 09:23.891
[SPEAKER_01]: So you still have to have, like, you still need a person on the driver's seat to navigate and like, like, help AI kind of navigate your brand to the best successful point.
09:24.251 --> 09:25.551
[SPEAKER_02]: Yeah, that makes so much sense.
09:25.951 --> 09:29.012
[SPEAKER_02]: But on the, on the automated side of it, the bit where we would
09:33.928 --> 09:38.013
[SPEAKER_02]: It's kind of got two of its own runways to stick with my terrible runway analogy.
09:38.073 --> 09:39.956
[SPEAKER_02]: Well, we'll just go with it.
09:40.236 --> 09:48.647
[SPEAKER_02]: So we've got the learning and improving side of it, where it's looking through the data to find new things to fix and new things to improve.
09:48.707 --> 09:50.229
[SPEAKER_02]: But it's also got the...
09:51.130 --> 10:14.737
[SPEAKER_02]: Well that's very obvious, side of things of image optimization and these things that we should be doing all the time but because new skews come in and we fail to optimize the images or and a Rodney used to do that and Rodney left and no one took it up or whatever, you know you end up with these things that have got left that are just blindingly obvious which the AI can can hold can deal with what you would you say that that's true there's kind of like the
10:15.840 --> 10:23.361
[SPEAKER_02]: The hygiene runway of doing all the stuff that should be that we know should be done and then there's the optimization runway for the automated activity.
10:23.949 --> 10:26.151
[SPEAKER_01]: Yeah, definitely.
10:26.171 --> 10:35.979
[SPEAKER_02]: And how do we get, I mean, the first of those, the hygiene one, pretty straightforward, relatively easy to do.
10:36.019 --> 10:40.984
[SPEAKER_02]: The second one, how do we go about getting AI to do that well for us?
10:41.004 --> 10:46.028
[SPEAKER_02]: There's an awful lot of data it needs, information it needs training, particularly it needs in some level of oversight.
10:46.693 --> 10:46.873
[SPEAKER_01]: Yeah.
10:47.633 --> 10:51.555
[SPEAKER_01]: So I can give you an example of how we do it.
10:51.755 --> 10:52.295
[SPEAKER_01]: Please do.
10:52.395 --> 10:55.096
[SPEAKER_01]: It's, I find it very fascinating.
10:55.156 --> 11:05.940
[SPEAKER_01]: So one of our agents, it's optimizing the user journey by predicting where users are going to click next and then optimizing like preloading that page in advance.
11:06.540 --> 11:12.623
[SPEAKER_01]: So the way we train the model to understand that behavior is first you map out
11:13.443 --> 11:20.465
[SPEAKER_01]: typical user behavior on the website, so it's like a very generic like where you how users typically browse on the website.
11:21.145 --> 11:26.726
[SPEAKER_01]: Then we look at historic data how users on that particular website are behaving.
11:27.386 --> 11:35.288
[SPEAKER_01]: Then we analyze you as a user right now what you what are you doing and what have you done before like which pages did you visit.
11:36.008 --> 11:43.577
[SPEAKER_01]: And then the fourth layer of cognition like understanding what's happening is understanding where are you exactly right now on the side.
11:43.657 --> 11:52.647
[SPEAKER_01]: So having all those four points, like kind of you put different weights on them and like kind of analyze what could happen next.
11:53.268 --> 11:53.649
[SPEAKER_01]: And then,
11:54.970 --> 12:10.374
[SPEAKER_01]: I think it's all about testing, running it over and over again, see where the model actually made a wrong prediction and analyzing what happened and kind of having this ongoing loop of like, oh, I made, I said something, I predicted something correctly.
12:10.794 --> 12:20.597
[SPEAKER_01]: I know that next time most probably this is going to happen again, or hey, I predicted something in a wrong way, let's analyze why and kind of improve that for my next run.
12:23.418 --> 12:33.365
[SPEAKER_01]: multi-layered data analysis, paired with an ongoing improvement look that we saw work works best.
12:33.685 --> 12:50.697
[SPEAKER_01]: And ultimately, it's always a person that goes and analyzes like different sessions and different journeys to see if something may be got wrong, or so you always have a human ultimately in the back and making doing the final QA.
12:53.584 --> 13:01.548
[SPEAKER_02]: It's like we have to think, as we get on this automated route, we have to think about conversion optimization and a completely different way to how we've previously thought about it.
13:01.568 --> 13:08.351
[SPEAKER_02]: You know, a normal, you know, traditional conversion optimization would be, here's a problem, here's a hypothesis.
13:09.872 --> 13:27.350
[SPEAKER_02]: Let's test the hypothesis, let's review it very linear piece where we tackle one thing at a time, whereas with the AI automation, it feels like, you know, the agents in that, it feels like more where we're training them to fix a problem on the fly.
13:27.831 --> 13:29.572
[SPEAKER_02]: If that makes it so, it's not like a
13:30.373 --> 13:31.794
[SPEAKER_02]: Right, March, we're testing this.
13:31.995 --> 13:34.617
[SPEAKER_02]: April will test this in a naval test this.
13:35.037 --> 13:42.644
[SPEAKER_02]: But it's more, right, let's optimize one for preloading pages to speed up site speed.
13:42.664 --> 13:44.166
[SPEAKER_02]: Let's optimize one for something else.
13:44.206 --> 13:44.886
[SPEAKER_02]: And one for something else.
13:44.906 --> 13:47.789
[SPEAKER_02]: And we're to completely different.
13:48.990 --> 13:57.201
[SPEAKER_02]: building process, a management process to the old linear static, one thing at a time may be testing.
13:57.321 --> 14:01.166
[SPEAKER_02]: It feels like to me, is that a mindset shift it's worth making?
14:01.985 --> 14:22.565
[SPEAKER_01]: I think so, in my opinion, letting AI agents do CRL is open up a lot of opportunities also for smaller brands, because what you said, like, you know, you have the A-B test, like you have to have a minimal sample size for each of the groups to actually have meaningful results.
14:22.926 --> 14:25.008
[SPEAKER_01]: So, usually very,
14:25.488 --> 14:33.509
[SPEAKER_01]: proper AB tests were reserved for companies of like certain volume in purchases and certain volume of page views.
14:34.250 --> 14:37.730
[SPEAKER_01]: And with AI, this is no longer a limitation.
14:37.790 --> 14:45.832
[SPEAKER_01]: Like you can have a side that only has, I don't know, maybe 10,000 sessions every month, and still have that website
14:53.093 --> 15:04.277
[SPEAKER_01]: AI is fixing the problems as they arise, instead of anticipating of what could have been wrong, or what could be wrong, or what could be improved.
15:04.917 --> 15:07.958
[SPEAKER_01]: So it's for me just a different mindset.
15:08.920 --> 15:11.842
[SPEAKER_02]: And I suppose then our A.B.
15:11.942 --> 15:25.153
[SPEAKER_02]: testing energy is left for the really big brand stuff where we're getting involved with and the things which may have not made it onto the table historically because we were focused on more of the stuff that we're now handing over to the agents.
15:25.173 --> 15:29.917
[SPEAKER_02]: With that B, is that what you're seeing happen to businesses who are starting down this pathway?
15:30.562 --> 15:34.363
[SPEAKER_01]: For sure, I feel that exactly what you said.
15:34.483 --> 15:39.105
[SPEAKER_01]: Like the small tasks and the more mundane tasks are now being handed over to agents.
15:39.825 --> 15:47.087
[SPEAKER_01]: And then agencies or like the site optimization teams can actually concentrate on those bigger chunks.
15:47.727 --> 15:56.730
[SPEAKER_01]: And like go much deeper also in the creative part, not just on like the technical like the knowledge that they also made like with the button should be read
16:00.231 --> 16:00.955
[SPEAKER_01]: Let AI do.
16:01.946 --> 16:06.127
[SPEAKER_02]: Yeah, and it's, it's all for us that we all keep using the same button example.
16:06.167 --> 16:12.088
[SPEAKER_02]: And we all know it's like so not the most important thing to test, but it's such an easy way of getting across the concept.
16:12.568 --> 16:15.249
[SPEAKER_02]: I think we've, we've both already been guilty of it once each so far.
16:15.289 --> 16:18.209
[SPEAKER_02]: So we'll see how the score card looks like at the end.
16:18.829 --> 16:22.010
[SPEAKER_02]: Going back to the, the automated AI runways.
16:22.590 --> 16:26.151
[SPEAKER_02]: It feels like this type of conversion
16:32.152 --> 16:43.196
[SPEAKER_01]: Yes, for sure, like I was just lately doing an analysis on because the reality is not that many tools right now are doing a gente conversion rate optimization.
16:43.536 --> 16:44.156
[SPEAKER_01]: It's just a few.
16:44.617 --> 16:46.877
[SPEAKER_01]: And I was looking at most of them.
16:47.298 --> 16:47.418
[SPEAKER_01]: And
16:48.418 --> 16:57.982
[SPEAKER_01]: actually what I found out is that the majority are doing optimization in terms of personalization, search optimization, catalog optimization.
16:58.002 --> 17:13.028
[SPEAKER_01]: So I think that's the kind of the core thing that's currently out there and then like here and in there you're going to find some solutions optimizing for user experience or for optimizing a beta testing like that.
17:13.108 --> 17:23.853
[SPEAKER_01]: was very interesting for me, like using AI to generate all your hypothesis that you can then go with your zero agency or team and go ahead and test.
17:24.493 --> 17:32.237
[SPEAKER_02]: I was like an unbiased set of AB tests, not the boss wants this one tested, so we better do it at times in our AI.
17:32.817 --> 17:33.137
[SPEAKER_02]: Yeah.
17:34.138 --> 17:50.106
[SPEAKER_02]: So on that basis then, I guess we're potentially in the next few years moving to a place where personalisation and conversion rate optimisation in that sense become very much into link because they're both feeding off or should be feeding off very similar data sets.
17:50.586 --> 18:07.881
[SPEAKER_02]: The conversion rate agent is going to do a better job if they've got sales data and they know that that person's coming from Facebook ads and that they've ordered five times before and last time they came in via email or whatever it might be, which is all that the personalization thrives off as well.
18:08.862 --> 18:10.944
[SPEAKER_02]: I want, I'm going to go deep now.
18:11.204 --> 18:13.407
[SPEAKER_02]: I do wonder if we end up with a different category.
18:13.687 --> 18:18.732
[SPEAKER_02]: You know, if we start talking about these things we do with different terminology in the next few years, it's a big questionality.
18:18.812 --> 18:21.134
[SPEAKER_02]: We're welcome just to kind of go really, guys.
18:22.707 --> 18:30.253
[SPEAKER_01]: I think, I mean, what we're seeing even nowadays is that terminology changes a lot.
18:30.333 --> 18:40.180
[SPEAKER_01]: Like with AI, like the thought of having something like a genetic conversion rate optimization, like this didn't exist a year ago, right?
18:40.240 --> 18:44.223
[SPEAKER_01]: Like it's completing your term that's like being formed right now.
18:44.904 --> 18:48.846
[SPEAKER_01]: So I want to be surprised if the way
18:49.787 --> 19:05.337
[SPEAKER_01]: conversion rate optimization, personalization, but also recommendations like upsells and all those things actually change not only in terms of terminology, but also how they work with each other because ultimately where I see the industry going is.
19:06.377 --> 19:29.059
[SPEAKER_01]: data being like the ultimate king and like data not being like, you know, siloed into different platforms but having it sit on one place because ultimately this is how you actually get value and like this is how you can optimize processes can improve conversion like offer more personalization and like everything else that follows.
19:29.739 --> 19:47.653
[SPEAKER_02]: I think the person who invents the perfect hub for all of business's data securely with the right access, with the training and everything built into it, is once the winner of that comes out, I think it's going to really push things forward, but the moment it feels like it's all very siloed.
19:48.514 --> 19:51.256
[SPEAKER_01]: Yeah, that's true, but what I'm seeing is
19:53.068 --> 20:19.581
[SPEAKER_01]: our attempt to consolidate that with almost every brand or especially like most of the SaaS services offering their own MCPs, so kind of feeding the information to the large language models and I feel that I kind of that MCP game and like being able to call different tools within the LLM of your choice bit, like cloth or chatjupity is gonna be ultimately the future,
20:21.282 --> 20:36.850
[SPEAKER_01]: What other tools tried to achieve, maybe five, ten years ago, I think Signal was similar to, but I think the why they failed was that because people were able to chat with their data and I kind of chat with the information that was available.
20:36.870 --> 20:43.074
[SPEAKER_01]: We always had to do all those complex steps and processes and now everything is just conversational.
20:43.434 --> 20:46.736
[SPEAKER_01]: So I think that's a big change right now.
20:47.356 --> 21:08.047
[SPEAKER_02]: Yeah, and to be having all the data fed into something that fundamentally understands the data, you know, it's been forever trying to field match and make it all, you know, oh yeah, I was trying to do the most simple things because it thinks that's something different than what it is, whereas with the LLMs, they have a certain amount of intelligence already, of course.
21:08.748 --> 21:14.591
[SPEAKER_02]: Anything else the audience needs to know about AI and conversion rate management before
21:18.043 --> 21:21.524
[SPEAKER_01]: I would say that they shouldn't fear from it.
21:22.364 --> 21:25.425
[SPEAKER_01]: And my advice is always start with something small.
21:25.505 --> 21:29.166
[SPEAKER_01]: Some start with something that ideally won't break your website.
21:30.346 --> 21:43.009
[SPEAKER_01]: Like, be it removing friction points or starting with some personalization and just interacting with AI agents really helps kind of
21:43.947 --> 22:02.972
[SPEAKER_01]: decrease that fear of like oh my god that's like that's so bad that I cannot control it because that's not that's not the reality you always have guardrails you always have set rules you can always guard a agents to not push anything before you actually go and check it so just start
22:07.360 --> 22:08.181
[SPEAKER_02]: great advice.
22:08.943 --> 22:10.886
[SPEAKER_02]: It's been great picking your brains about all of this.
22:11.346 --> 22:22.604
[SPEAKER_02]: Listers, remember we're going to stay tuned right to the end so you don't miss out on Elana's insider tips on conversion rates and my suggestion for more free resources to help you improve things even further in your businesses.
22:24.703 --> 22:28.406
[SPEAKER_02]: Okay, and then as so far, we've gone deep into how can AI help?
22:28.906 --> 22:32.149
[SPEAKER_02]: Now you get to ask your insider knowledge about the whole of conversion rate.
22:32.549 --> 22:38.174
[SPEAKER_02]: So for the following questions, you're going to be anything to do with that, which of course does include everything we've already been talking about.
22:38.314 --> 22:39.194
[SPEAKER_02]: And then are you ready?
22:39.835 --> 22:40.055
[SPEAKER_02]: Yes.
22:40.656 --> 22:40.996
[SPEAKER_02]: Awesome.
22:41.056 --> 22:42.597
[SPEAKER_02]: Let's start with new advice.
22:42.997 --> 22:48.422
[SPEAKER_02]: If we've inspired someone to take their first step, what do they need to know to give themselves the best chance of success?
22:49.342 --> 23:01.229
[SPEAKER_01]: I'm going to repeat myself by saying that the one thing that you really need to do is start by measuring a problem and from there on it's really easy to find the right tools that can help you optimize it.
23:02.950 --> 23:04.011
[SPEAKER_01]: So start with that.
23:04.391 --> 23:10.755
[SPEAKER_02]: Yeah, that one thing I don't think is ever going to change in conversion rates is yeah, we're start with the numbers.
23:11.575 --> 23:14.717
[SPEAKER_02]: Now once you've started, of course, you've got to keep optimizing.
23:14.757 --> 23:17.139
[SPEAKER_02]: So what's your favorite way to improve performance?
23:18.426 --> 23:32.921
[SPEAKER_01]: So one thing I really I think people forget a lot of times, but I feel it's so crucial is to do regular checkups on their stores and especially on their apps and what's left from the apps they deleted.
23:33.401 --> 23:37.025
[SPEAKER_01]: So my advice to you is like every six months go ahead.
23:37.405 --> 23:40.666
[SPEAKER_01]: check all the apps that you have in your Shopify store.
23:41.107 --> 23:42.867
[SPEAKER_01]: Make sure if you're using them.
23:43.527 --> 23:45.188
[SPEAKER_01]: Check if they're actually making your money.
23:45.248 --> 23:47.789
[SPEAKER_01]: If they're giving you the value of your thought, they're going to give you.
23:48.309 --> 23:50.350
[SPEAKER_01]: And which are on just delete them.
23:50.870 --> 23:56.253
[SPEAKER_01]: Like make sure that you delete it also like the small increments of code they usually leave behind.
23:57.273 --> 24:05.477
[SPEAKER_01]: what I'm saying that is because that's something people almost always forget, but this leads to problematic points like friction.
24:06.157 --> 24:09.859
[SPEAKER_01]: And the friction happens in the most unfortunate moments.
24:10.059 --> 24:18.783
[SPEAKER_01]: So when someone tries to click on to card button and this button doesn't work because you have so many apps that actually block that process.
24:19.844 --> 24:21.925
[SPEAKER_01]: That's very unfortunate.
24:22.145 --> 24:24.646
[SPEAKER_01]: So make sure to keep your
24:27.027 --> 24:29.949
[SPEAKER_01]: Thank you, Laker.
24:30.530 --> 24:31.511
[SPEAKER_02]: Brilliant advice.
24:32.071 --> 24:37.075
[SPEAKER_02]: If someone listing wants to learn more, is there a cheap or free resource you would recommend?
24:37.922 --> 24:38.122
[SPEAKER_01]: Yeah.
24:38.723 --> 24:48.749
[SPEAKER_01]: So whenever you want to learn more about conversion rate optimization, there are a lot of great resources in terms of podcasts and shows this being one of them.
24:49.389 --> 24:54.513
[SPEAKER_01]: I always say that chatjipiti or occult are always also a great start.
24:55.613 --> 25:01.357
[SPEAKER_01]: One thing I really like and like a resource that I don't see being talked enough in the industry
25:07.641 --> 25:18.888
[SPEAKER_01]: Yeah, by Sara Karamelagos every month she analyzes all Shopify teams and she measures what's their effect on user experience.
25:19.048 --> 25:26.152
[SPEAKER_01]: So if they make your site slower, if they call some friction points or if they call some instability.
25:26.172 --> 25:32.076
[SPEAKER_01]: So like if things are moving up and down your store and I think that's the first place.
25:35.778 --> 25:44.348
[SPEAKER_01]: just to make sure that the theme they use is actually optimized for good site and user, good speed and user experience.
25:44.828 --> 25:50.795
[SPEAKER_02]: What an awesome site and yet one which I suspect many of the things that I'm not sure I want to look.
25:51.975 --> 26:00.098
[SPEAKER_02]: You know, if you're a few years into using the same theme and you went on that website and discovered it was, it was not a good one, you would be like, oh, no.
26:00.398 --> 26:02.839
[SPEAKER_02]: Like, approach it your own risk, everybody that one.
26:03.099 --> 26:04.600
[SPEAKER_02]: So that was theme vitals.
26:05.240 --> 26:10.682
[SPEAKER_02]: And finally, crystal ball time, what's coming up in the next six to 12 months that we should be getting ready for?
26:11.853 --> 26:16.637
[SPEAKER_01]: I think that more AI is just going to enter this sphere.
26:17.598 --> 26:24.924
[SPEAKER_01]: As we just talked about a genetic conversion rate optimization, I'm foreseen that much more companies are going to jump on that.
26:25.264 --> 26:28.467
[SPEAKER_01]: We are going to see even more optimizations happening.
26:28.527 --> 26:31.270
[SPEAKER_01]: Like, I'll automate it and autonomous way.
26:32.130 --> 26:48.653
[SPEAKER_01]: Additionally, I feel that sidekick is going to get even more powerful, so I'm very excited to see how it starts to talk more with the apps that are currently in the app store, and maybe also some apps that are outside of the app store.
26:48.693 --> 26:58.875
[SPEAKER_01]: We'll see about that, but I just the ease of building a website and using AS every point of like building an optimizing a website that's going to
27:00.884 --> 27:09.411
[SPEAKER_02]: Now, I think you're spot on the, um, we are very nearly into the show though, so could you please at the list as now where they can find you and UXify?
27:10.292 --> 27:14.095
[SPEAKER_01]: Yes, uh, you can always visit our website, UXify.com.
27:14.596 --> 27:18.219
[SPEAKER_01]: We are also very active on LinkedIn, so make sure to give us a follow.
27:18.539 --> 27:26.646
[SPEAKER_01]: And yeah, if you ever want to talk more about conversion rate optimization and Daniela Gentiquey, we are here.
27:27.545 --> 27:38.284
[SPEAKER_02]: Awesome, there we go everybody and I know thank you so much for coming on and talking about this I think you'll have made AI and conversion rates a lot more understandable for everyone listening so we really appreciate it.
27:38.304 --> 27:39.105
[SPEAKER_02]: Thank you for being here.
27:39.827 --> 27:41.069
[SPEAKER_01]: Thank you for having me, Chloe.
27:43.032 --> 27:56.761
[SPEAKER_02]: So there you go, a clear guide to what's happening with AI and conversion rail to myization at the moment, I think you give it in those kind of three runways I was talking about, it's probably the best, my head, at least it's the best way to make it make sense.
27:56.781 --> 27:57.482
[SPEAKER_02]: You've got the...
27:58.302 --> 28:07.088
[SPEAKER_02]: The obvious stuff we should be doing day to day anyway, like image, optimization, fixing broken things, et cetera, et cetera, et cetera, that you get the AI handling so it just happens.
28:07.909 --> 28:26.202
[SPEAKER_02]: Then you've got the things that you can train AI agents to work on on the fly as people are actually experiencing the website, like preloading pages and all sorts of other clever clever things that just they're not the risky things that could destroy your site or damage your business, but they are the things which
28:26.902 --> 28:33.086
[SPEAKER_02]: are worth kind of ongoing testing, optimizing, improving, but with a small layer of humanet oversight especially to get them up and running.
28:33.847 --> 28:45.434
[SPEAKER_02]: And then that third runway, which is where you and your team, with some help from AI, where you need it, tackle the big stuff, the branding, the wording, all those big, big things.
28:45.695 --> 28:51.819
[SPEAKER_02]: And of course, that whole automated piece then starts to massively overlap in my head with the world of personalisation.
28:52.479 --> 28:57.905
[SPEAKER_02]: You can get links to everything we discussed, the transcripts are notes and more at keepoptimizing.com.
28:58.385 --> 29:03.471
[SPEAKER_02]: You can also use our direct episode short links that's keepopt.com for such the episode number.
29:04.091 --> 29:07.175
[SPEAKER_02]: And when you get to the website, you can add yourself to our email list.
29:07.655 --> 29:11.156
[SPEAKER_02]: Thank you for tuning in to this episode of the Keep Optimizing Podcast.
29:11.796 --> 29:16.158
[SPEAKER_02]: All five of our episodes revisiting the conversion rates throughout our now live.
29:16.558 --> 29:22.520
[SPEAKER_02]: Think of it as you'll get me up to date on conversion rates fast, masterclass, so please do have a listen to them all.
29:23.300 --> 29:32.583
[SPEAKER_02]: Either scroll up your podcast feed and you'll find them or use the short link, keep up.com forward slash CRO and our next cast masterclass topic.
29:32.883 --> 29:33.363
[SPEAKER_02]: Ready for this?
29:33.803 --> 29:36.564
[SPEAKER_00]: We are doing Google Ads.
29:37.234 --> 29:41.337
[SPEAKER_02]: Yes, we're going back to the world of PMAX and Google ads.
29:41.897 --> 29:45.739
[SPEAKER_02]: And we are going to be revisiting all of that in a couple of weeks.
29:46.059 --> 29:51.283
[SPEAKER_02]: You can get that catch that thread so far even at keepopt.com for such Google ads.
29:51.343 --> 29:54.665
[SPEAKER_02]: Just do that and you'll go straight to the right page on the website or scorn on your feed.
29:55.385 --> 30:02.891
[SPEAKER_02]: make sure you tune in in two Wednesday's time to catch the start of our dive into all that's new that you need to know about Google Ads.
30:03.391 --> 30:05.833
[SPEAKER_02]: So yes there is no episode next week.
30:06.173 --> 30:17.202
[SPEAKER_02]: That's going to give you time to catch up on any of the CRO episodes that you missed and re-list into any that you need to or share them with your team and then put into action what you've learned, which is the most important thing to do.
30:17.722 --> 30:23.124
[SPEAKER_02]: Make sure you subscribe to us or following us on your favourite podcast player so you don't miss out when we come back in two weeks.
30:23.564 --> 30:35.209
[SPEAKER_02]: And if you know someone who's particularly interested in improving their conversion rates or their Google ads, please tell them we're covering it all as this show exists to help you and them improve your marketing.
30:35.909 --> 30:40.831
[SPEAKER_02]: Have a great week and make sure you listen to the next episode so I can help you to keep optimizing your marketing.
30:44.035 --> 30:46.040
[SPEAKER_00]: Access everything, keep optimizing.
30:46.722 --> 30:50.290
[SPEAKER_00]: At keepoptimizing.com, that's within us, not as in.
00:00.570 --> 00:03.511
[SPEAKER_01]: Everyone can spot if something is made with AI.
00:03.531 --> 00:07.512
[SPEAKER_01]: Like it just like look at two or three websites, you know they are AI generated.
00:07.652 --> 00:10.253
[SPEAKER_01]: And like the same is with copy, the same is with imagery.
00:10.613 --> 00:17.676
[SPEAKER_01]: So I feel that in the long term brands that are selling genuine experiences are the ones that are gonna win.
00:20.057 --> 00:26.999
[SPEAKER_00]: You're listening to the keep optimizing podcast to increase your traffic, improve your conversion rates and grow your profits.
00:29.960 --> 00:35.125
[SPEAKER_02]: Hello and welcome on Chloe Thomas the host of this e-commerce marketing focus podcast.
00:35.245 --> 00:43.372
[SPEAKER_02]: If you're not familiar well at keep optimising we release five episode mini series at a time that dive into one e-commerce marketing topic.
00:43.492 --> 00:51.840
[SPEAKER_02]: It could be a channel like email or SEO or Facebook ads or it could be a big topic like acquisition retention or analytics.
00:52.320 --> 01:06.887
[SPEAKER_02]: In each of the episodes, new ones released on Wednesdays, I get to interview a different expert to explore the latest device on making it work for you and then after each five episodes, mini series, we take a week off to give you time to take action on what you've learnt.
01:07.567 --> 01:17.613
[SPEAKER_02]: Well, we have covered all sorts of angles on conversion rate improvements in this latest series, but one glaring emission has been a dive into how AI can help you.
01:18.053 --> 01:20.154
[SPEAKER_02]: So that's what we're going to be covering in this episode.
01:20.654 --> 01:24.876
[SPEAKER_02]: We're going to be talking about where you should use AI, where you shouldn't, how to
01:26.017 --> 01:32.985
[SPEAKER_02]: segment your your CROW activity to get the best out of AI and lots of other really interesting bits and pieces.
01:33.145 --> 01:40.192
[SPEAKER_02]: If you've been struggling to work out how to get AI to deliver for your e-commerce business and you're going to really enjoy this upcoming conversation.
01:40.573 --> 01:45.077
[SPEAKER_02]: And remember, there won't be an episode next week because we're be between series.
01:45.838 --> 01:51.381
[SPEAKER_02]: So you, that will give you time to catch up on any of this that you've missed, share with your team and take action.
01:51.841 --> 02:01.185
[SPEAKER_02]: Make sure you list them right at the end of the episode to find out what our next series is, but also because my guest shares some really great insight during the insider tips round as well.
02:03.605 --> 02:07.948
[SPEAKER_02]: In this episode, I'm chatting with conversion rates expert Alena Costover.
02:08.348 --> 02:18.495
[SPEAKER_02]: Alena is head of marketing at UXify, an AI native digital experience platform that helps websites to achieve flawless user experience and double engagement.
02:18.875 --> 02:19.696
[SPEAKER_02]: Hello, Alena.
02:20.216 --> 02:20.877
[SPEAKER_02]: Hi, Chloe.
02:21.377 --> 02:22.538
[SPEAKER_02]: Great to have you here.
02:22.658 --> 02:25.720
[SPEAKER_02]: How did you get into the world of conversion rates?
02:26.600 --> 02:36.913
[SPEAKER_01]: Well, in all honesty, I feel that the conversion rate optimization world found me, I've been heading marketing like small marketing teams for the past 10 years.
02:37.814 --> 02:43.300
[SPEAKER_01]: And in my role, I was responsible for the site creation, like the...
02:44.101 --> 03:12.948
[SPEAKER_01]: making sure that the site is optimized and converting, so as such, I had to learn all the tricks around conversion rate optimization, and then on the other hand, being part of companies that are selling to e-commerce, I had the privilege to talk with e-commerce owners, marketers, and agencies that are solving exactly that problem and optimizing for better conversion and see what works,
03:14.128 --> 03:19.073
[SPEAKER_01]: So this is how my past actually crossed with convertible rate optimisation.
03:20.114 --> 03:20.755
[SPEAKER_02]: Excellent.
03:20.815 --> 03:34.288
[SPEAKER_02]: Well look, we're going to get deep into a topic which we haven't really got very into weirdly across this series, which is how AI can help us improve our conversion rates, which I suspect there's quite a lot it can do.
03:34.848 --> 03:37.050
[SPEAKER_02]: know, I know you are deep in what it can do.
03:37.130 --> 03:44.815
[SPEAKER_02]: So what's the best way to deploy AI into use AI to help us improve the conversion rates on our e-commerce site?
03:46.405 --> 03:51.686
[SPEAKER_01]: I think the best way is to start small and always start by measuring things.
03:52.226 --> 03:57.607
[SPEAKER_01]: So let's give an example for site speeds, so how fast your site could be.
03:58.147 --> 04:02.628
[SPEAKER_01]: Like the most natural way you do is measure your performance.
04:02.648 --> 04:09.069
[SPEAKER_01]: You can go to like Google page speed insights and actually check if you have problem or not.
04:09.489 --> 04:13.450
[SPEAKER_01]: And once you know that there is a problem, then there is a lot of you can do.
04:13.830 --> 04:23.198
[SPEAKER_01]: you can see what the problems are and I typically or divide the problems of can they break your site or will they make your money?
04:23.838 --> 04:29.063
[SPEAKER_01]: If the problems are very critical and they can break your site, then I would say
04:29.843 --> 04:44.175
[SPEAKER_01]: Use an AI assist at a approach where AI can help you and enhance what you're doing, but if the opportunity is clear and it can make you more money and there's no risk of breaking your website, then I'll say go with the AI solution.
04:44.976 --> 04:55.104
[SPEAKER_01]: And maybe another point on that, activities that usually require a lot of data and a lot of analysis of data and quick analysis of data,
04:57.326 --> 05:08.444
[SPEAKER_01]: elements that you can give AI to optimize because it's just faster in thinking and cognition and like understanding like the patterns and reacting in real life.
05:09.238 --> 05:11.839
[SPEAKER_02]: I love how you divide that up into different bits.
05:11.919 --> 05:21.040
[SPEAKER_02]: We've got the measurement part, obviously, fairly, I think most people have now used AI to create some kind of dashboard or other.
05:21.400 --> 05:23.121
[SPEAKER_02]: So we know we're going to be doing that.
05:23.141 --> 05:25.041
[SPEAKER_02]: So we're paying better attention to the numbers.
05:25.461 --> 05:34.803
[SPEAKER_02]: And then the things we find out of that, are you doing the just let AI fix it, or are you doing the AI assisted root by going,
05:36.464 --> 05:44.749
[SPEAKER_02]: What's the risk here, basically, is what we're saying, is that how much are we talking about playing around with button colors, or are we talking about rewiring the check out?
05:45.409 --> 05:51.193
[SPEAKER_02]: And like clearly one of those, you want to do the assisted and one of those, you know, you can give over to it.
05:51.773 --> 05:57.317
[SPEAKER_02]: And does this mean we end up with kind of like a multi-part?
05:58.444 --> 06:09.493
[SPEAKER_02]: way of handling RAI, where we've got things which are more manual, where we're working with AI or not AI to fix, to measure, to amend and so forth.
06:09.813 --> 06:14.457
[SPEAKER_02]: And then we've got stuff that's just carrying on, crunching numbers, improving things.
06:14.497 --> 06:19.882
[SPEAKER_02]: It's kind of like almost like a two, two runway version of AI now.
06:20.722 --> 06:23.024
[SPEAKER_02]: There's no, not, not the AI of commercial optimization.
06:24.399 --> 06:46.696
[SPEAKER_01]: I would say is, at least from my point of view, and why I think that is a website, like there are a lot of things in the website, you can really live it to be self-optimized and have them be fully autonomous, side-speed, removing friction, personalization, optimizing images, things like that.
06:47.736 --> 06:57.300
[SPEAKER_01]: very few things can go wrong and usually like the knowledge that you have to acquire to do that yourself manually properly, it's quite vast.
06:57.781 --> 07:08.805
[SPEAKER_01]: So it's going to take you either a lot of time or a lot of resources to solve those problems and those problems can and are currently very easily fixed with just letting AI do the job.
07:09.405 --> 07:12.347
[SPEAKER_01]: But then on the other hand, there are other items which
07:13.167 --> 07:28.173
[SPEAKER_01]: Some people made my disagree with me, but I feel that website microcopy, website branding, and how your website is looking, those still have to sit with real person.
07:28.213 --> 07:32.315
[SPEAKER_01]: Like, obviously it can be AA assisted, but it's ultimately your brand.
07:32.415 --> 07:35.837
[SPEAKER_01]: Like, it's a bit counterintuitive for me to,
07:36.825 --> 07:45.236
[SPEAKER_01]: give away that privilege to show your personality in your brandy front of your audience and just let AI do it for you.
07:45.676 --> 07:46.998
[SPEAKER_01]: It can work, obviously.
07:47.399 --> 07:52.225
[SPEAKER_01]: But I find it a short-term win instead of long-term,
07:53.853 --> 07:54.073
[SPEAKER_01]: with.
07:54.594 --> 08:00.220
[SPEAKER_02]: Well, yeah, and I think that makes so much sense, not least because brand is where you want to stand out.
08:00.320 --> 08:03.243
[SPEAKER_02]: And AI works by looking at the averages of everything.
08:03.403 --> 08:03.903
[SPEAKER_02]: Exactly.
08:03.923 --> 08:11.511
[SPEAKER_02]: Nothing is like, you just end up, if we get let AI control all our brands, we'd all end up beige in a fancy way.
08:12.492 --> 08:19.816
[SPEAKER_01]: And I feel that even nowadays, like, purchasers are getting even more critical.
08:19.836 --> 08:23.898
[SPEAKER_01]: Like, everyone can spot if something is made with AI.
08:23.938 --> 08:26.539
[SPEAKER_01]: Like, you just look at two or three websites.
08:26.599 --> 08:28.480
[SPEAKER_01]: You know they are AI generated.
08:28.620 --> 08:31.942
[SPEAKER_01]: And the same is with copy, the same is with image imagery.
08:32.302 --> 08:36.204
[SPEAKER_01]: So I feel that nowadays consumers are getting much more critical.
08:36.704 --> 09:01.363
[SPEAKER_01]: and like brands like, I don't know if you have heard of them like, hot-gore pickles or their girls that are just selling like pickles like as basic as that, but just the brand and the narrative that they build around it, it's so genuine and relatable that that's what sells and they feel that ultimately in in the long term brands that are selling genuine experiences, are the ones that are going to win.
09:02.224 --> 09:03.325
[SPEAKER_01]: So again,
09:06.007 --> 09:09.808
[SPEAKER_01]: A lot of people say, like, if nothing comes in, nothing is going to come out.
09:09.868 --> 09:23.891
[SPEAKER_01]: So you still have to have, like, you still need a person on the driver's seat to navigate and like, like, help AI kind of navigate your brand to the best successful point.
09:24.251 --> 09:25.551
[SPEAKER_02]: Yeah, that makes so much sense.
09:25.951 --> 09:29.012
[SPEAKER_02]: But on the, on the automated side of it, the bit where we would
09:33.928 --> 09:38.013
[SPEAKER_02]: It's kind of got two of its own runways to stick with my terrible runway analogy.
09:38.073 --> 09:39.956
[SPEAKER_02]: Well, we'll just go with it.
09:40.236 --> 09:48.647
[SPEAKER_02]: So we've got the learning and improving side of it, where it's looking through the data to find new things to fix and new things to improve.
09:48.707 --> 09:50.229
[SPEAKER_02]: But it's also got the...
09:51.130 --> 10:14.737
[SPEAKER_02]: Well that's very obvious, side of things of image optimization and these things that we should be doing all the time but because new skews come in and we fail to optimize the images or and a Rodney used to do that and Rodney left and no one took it up or whatever, you know you end up with these things that have got left that are just blindingly obvious which the AI can can hold can deal with what you would you say that that's true there's kind of like the
10:15.840 --> 10:23.361
[SPEAKER_02]: The hygiene runway of doing all the stuff that should be that we know should be done and then there's the optimization runway for the automated activity.
10:23.949 --> 10:26.151
[SPEAKER_01]: Yeah, definitely.
10:26.171 --> 10:35.979
[SPEAKER_02]: And how do we get, I mean, the first of those, the hygiene one, pretty straightforward, relatively easy to do.
10:36.019 --> 10:40.984
[SPEAKER_02]: The second one, how do we go about getting AI to do that well for us?
10:41.004 --> 10:46.028
[SPEAKER_02]: There's an awful lot of data it needs, information it needs training, particularly it needs in some level of oversight.
10:46.693 --> 10:46.873
[SPEAKER_01]: Yeah.
10:47.633 --> 10:51.555
[SPEAKER_01]: So I can give you an example of how we do it.
10:51.755 --> 10:52.295
[SPEAKER_01]: Please do.
10:52.395 --> 10:55.096
[SPEAKER_01]: It's, I find it very fascinating.
10:55.156 --> 11:05.940
[SPEAKER_01]: So one of our agents, it's optimizing the user journey by predicting where users are going to click next and then optimizing like preloading that page in advance.
11:06.540 --> 11:12.623
[SPEAKER_01]: So the way we train the model to understand that behavior is first you map out
11:13.443 --> 11:20.465
[SPEAKER_01]: typical user behavior on the website, so it's like a very generic like where you how users typically browse on the website.
11:21.145 --> 11:26.726
[SPEAKER_01]: Then we look at historic data how users on that particular website are behaving.
11:27.386 --> 11:35.288
[SPEAKER_01]: Then we analyze you as a user right now what you what are you doing and what have you done before like which pages did you visit.
11:36.008 --> 11:43.577
[SPEAKER_01]: And then the fourth layer of cognition like understanding what's happening is understanding where are you exactly right now on the side.
11:43.657 --> 11:52.647
[SPEAKER_01]: So having all those four points, like kind of you put different weights on them and like kind of analyze what could happen next.
11:53.268 --> 11:53.649
[SPEAKER_01]: And then,
11:54.970 --> 12:10.374
[SPEAKER_01]: I think it's all about testing, running it over and over again, see where the model actually made a wrong prediction and analyzing what happened and kind of having this ongoing loop of like, oh, I made, I said something, I predicted something correctly.
12:10.794 --> 12:20.597
[SPEAKER_01]: I know that next time most probably this is going to happen again, or hey, I predicted something in a wrong way, let's analyze why and kind of improve that for my next run.
12:23.418 --> 12:33.365
[SPEAKER_01]: multi-layered data analysis, paired with an ongoing improvement look that we saw work works best.
12:33.685 --> 12:50.697
[SPEAKER_01]: And ultimately, it's always a person that goes and analyzes like different sessions and different journeys to see if something may be got wrong, or so you always have a human ultimately in the back and making doing the final QA.
12:53.584 --> 13:01.548
[SPEAKER_02]: It's like we have to think, as we get on this automated route, we have to think about conversion optimization and a completely different way to how we've previously thought about it.
13:01.568 --> 13:08.351
[SPEAKER_02]: You know, a normal, you know, traditional conversion optimization would be, here's a problem, here's a hypothesis.
13:09.872 --> 13:27.350
[SPEAKER_02]: Let's test the hypothesis, let's review it very linear piece where we tackle one thing at a time, whereas with the AI automation, it feels like, you know, the agents in that, it feels like more where we're training them to fix a problem on the fly.
13:27.831 --> 13:29.572
[SPEAKER_02]: If that makes it so, it's not like a
13:30.373 --> 13:31.794
[SPEAKER_02]: Right, March, we're testing this.
13:31.995 --> 13:34.617
[SPEAKER_02]: April will test this in a naval test this.
13:35.037 --> 13:42.644
[SPEAKER_02]: But it's more, right, let's optimize one for preloading pages to speed up site speed.
13:42.664 --> 13:44.166
[SPEAKER_02]: Let's optimize one for something else.
13:44.206 --> 13:44.886
[SPEAKER_02]: And one for something else.
13:44.906 --> 13:47.789
[SPEAKER_02]: And we're to completely different.
13:48.990 --> 13:57.201
[SPEAKER_02]: building process, a management process to the old linear static, one thing at a time may be testing.
13:57.321 --> 14:01.166
[SPEAKER_02]: It feels like to me, is that a mindset shift it's worth making?
14:01.985 --> 14:22.565
[SPEAKER_01]: I think so, in my opinion, letting AI agents do CRL is open up a lot of opportunities also for smaller brands, because what you said, like, you know, you have the A-B test, like you have to have a minimal sample size for each of the groups to actually have meaningful results.
14:22.926 --> 14:25.008
[SPEAKER_01]: So, usually very,
14:25.488 --> 14:33.509
[SPEAKER_01]: proper AB tests were reserved for companies of like certain volume in purchases and certain volume of page views.
14:34.250 --> 14:37.730
[SPEAKER_01]: And with AI, this is no longer a limitation.
14:37.790 --> 14:45.832
[SPEAKER_01]: Like you can have a side that only has, I don't know, maybe 10,000 sessions every month, and still have that website
14:53.093 --> 15:04.277
[SPEAKER_01]: AI is fixing the problems as they arise, instead of anticipating of what could have been wrong, or what could be wrong, or what could be improved.
15:04.917 --> 15:07.958
[SPEAKER_01]: So it's for me just a different mindset.
15:08.920 --> 15:11.842
[SPEAKER_02]: And I suppose then our A.B.
15:11.942 --> 15:25.153
[SPEAKER_02]: testing energy is left for the really big brand stuff where we're getting involved with and the things which may have not made it onto the table historically because we were focused on more of the stuff that we're now handing over to the agents.
15:25.173 --> 15:29.917
[SPEAKER_02]: With that B, is that what you're seeing happen to businesses who are starting down this pathway?
15:30.562 --> 15:34.363
[SPEAKER_01]: For sure, I feel that exactly what you said.
15:34.483 --> 15:39.105
[SPEAKER_01]: Like the small tasks and the more mundane tasks are now being handed over to agents.
15:39.825 --> 15:47.087
[SPEAKER_01]: And then agencies or like the site optimization teams can actually concentrate on those bigger chunks.
15:47.727 --> 15:56.730
[SPEAKER_01]: And like go much deeper also in the creative part, not just on like the technical like the knowledge that they also made like with the button should be read
16:00.231 --> 16:00.955
[SPEAKER_01]: Let AI do.
16:01.946 --> 16:06.127
[SPEAKER_02]: Yeah, and it's, it's all for us that we all keep using the same button example.
16:06.167 --> 16:12.088
[SPEAKER_02]: And we all know it's like so not the most important thing to test, but it's such an easy way of getting across the concept.
16:12.568 --> 16:15.249
[SPEAKER_02]: I think we've, we've both already been guilty of it once each so far.
16:15.289 --> 16:18.209
[SPEAKER_02]: So we'll see how the score card looks like at the end.
16:18.829 --> 16:22.010
[SPEAKER_02]: Going back to the, the automated AI runways.
16:22.590 --> 16:26.151
[SPEAKER_02]: It feels like this type of conversion
16:32.152 --> 16:43.196
[SPEAKER_01]: Yes, for sure, like I was just lately doing an analysis on because the reality is not that many tools right now are doing a gente conversion rate optimization.
16:43.536 --> 16:44.156
[SPEAKER_01]: It's just a few.
16:44.617 --> 16:46.877
[SPEAKER_01]: And I was looking at most of them.
16:47.298 --> 16:47.418
[SPEAKER_01]: And
16:48.418 --> 16:57.982
[SPEAKER_01]: actually what I found out is that the majority are doing optimization in terms of personalization, search optimization, catalog optimization.
16:58.002 --> 17:13.028
[SPEAKER_01]: So I think that's the kind of the core thing that's currently out there and then like here and in there you're going to find some solutions optimizing for user experience or for optimizing a beta testing like that.
17:13.108 --> 17:23.853
[SPEAKER_01]: was very interesting for me, like using AI to generate all your hypothesis that you can then go with your zero agency or team and go ahead and test.
17:24.493 --> 17:32.237
[SPEAKER_02]: I was like an unbiased set of AB tests, not the boss wants this one tested, so we better do it at times in our AI.
17:32.817 --> 17:33.137
[SPEAKER_02]: Yeah.
17:34.138 --> 17:50.106
[SPEAKER_02]: So on that basis then, I guess we're potentially in the next few years moving to a place where personalisation and conversion rate optimisation in that sense become very much into link because they're both feeding off or should be feeding off very similar data sets.
17:50.586 --> 18:07.881
[SPEAKER_02]: The conversion rate agent is going to do a better job if they've got sales data and they know that that person's coming from Facebook ads and that they've ordered five times before and last time they came in via email or whatever it might be, which is all that the personalization thrives off as well.
18:08.862 --> 18:10.944
[SPEAKER_02]: I want, I'm going to go deep now.
18:11.204 --> 18:13.407
[SPEAKER_02]: I do wonder if we end up with a different category.
18:13.687 --> 18:18.732
[SPEAKER_02]: You know, if we start talking about these things we do with different terminology in the next few years, it's a big questionality.
18:18.812 --> 18:21.134
[SPEAKER_02]: We're welcome just to kind of go really, guys.
18:22.707 --> 18:30.253
[SPEAKER_01]: I think, I mean, what we're seeing even nowadays is that terminology changes a lot.
18:30.333 --> 18:40.180
[SPEAKER_01]: Like with AI, like the thought of having something like a genetic conversion rate optimization, like this didn't exist a year ago, right?
18:40.240 --> 18:44.223
[SPEAKER_01]: Like it's completing your term that's like being formed right now.
18:44.904 --> 18:48.846
[SPEAKER_01]: So I want to be surprised if the way
18:49.787 --> 19:05.337
[SPEAKER_01]: conversion rate optimization, personalization, but also recommendations like upsells and all those things actually change not only in terms of terminology, but also how they work with each other because ultimately where I see the industry going is.
19:06.377 --> 19:29.059
[SPEAKER_01]: data being like the ultimate king and like data not being like, you know, siloed into different platforms but having it sit on one place because ultimately this is how you actually get value and like this is how you can optimize processes can improve conversion like offer more personalization and like everything else that follows.
19:29.739 --> 19:47.653
[SPEAKER_02]: I think the person who invents the perfect hub for all of business's data securely with the right access, with the training and everything built into it, is once the winner of that comes out, I think it's going to really push things forward, but the moment it feels like it's all very siloed.
19:48.514 --> 19:51.256
[SPEAKER_01]: Yeah, that's true, but what I'm seeing is
19:53.068 --> 20:19.581
[SPEAKER_01]: our attempt to consolidate that with almost every brand or especially like most of the SaaS services offering their own MCPs, so kind of feeding the information to the large language models and I feel that I kind of that MCP game and like being able to call different tools within the LLM of your choice bit, like cloth or chatjupity is gonna be ultimately the future,
20:21.282 --> 20:36.850
[SPEAKER_01]: What other tools tried to achieve, maybe five, ten years ago, I think Signal was similar to, but I think the why they failed was that because people were able to chat with their data and I kind of chat with the information that was available.
20:36.870 --> 20:43.074
[SPEAKER_01]: We always had to do all those complex steps and processes and now everything is just conversational.
20:43.434 --> 20:46.736
[SPEAKER_01]: So I think that's a big change right now.
20:47.356 --> 21:08.047
[SPEAKER_02]: Yeah, and to be having all the data fed into something that fundamentally understands the data, you know, it's been forever trying to field match and make it all, you know, oh yeah, I was trying to do the most simple things because it thinks that's something different than what it is, whereas with the LLMs, they have a certain amount of intelligence already, of course.
21:08.748 --> 21:14.591
[SPEAKER_02]: Anything else the audience needs to know about AI and conversion rate management before
21:18.043 --> 21:21.524
[SPEAKER_01]: I would say that they shouldn't fear from it.
21:22.364 --> 21:25.425
[SPEAKER_01]: And my advice is always start with something small.
21:25.505 --> 21:29.166
[SPEAKER_01]: Some start with something that ideally won't break your website.
21:30.346 --> 21:43.009
[SPEAKER_01]: Like, be it removing friction points or starting with some personalization and just interacting with AI agents really helps kind of
21:43.947 --> 22:02.972
[SPEAKER_01]: decrease that fear of like oh my god that's like that's so bad that I cannot control it because that's not that's not the reality you always have guardrails you always have set rules you can always guard a agents to not push anything before you actually go and check it so just start
22:07.360 --> 22:08.181
[SPEAKER_02]: great advice.
22:08.943 --> 22:10.886
[SPEAKER_02]: It's been great picking your brains about all of this.
22:11.346 --> 22:22.604
[SPEAKER_02]: Listers, remember we're going to stay tuned right to the end so you don't miss out on Elana's insider tips on conversion rates and my suggestion for more free resources to help you improve things even further in your businesses.
22:24.703 --> 22:28.406
[SPEAKER_02]: Okay, and then as so far, we've gone deep into how can AI help?
22:28.906 --> 22:32.149
[SPEAKER_02]: Now you get to ask your insider knowledge about the whole of conversion rate.
22:32.549 --> 22:38.174
[SPEAKER_02]: So for the following questions, you're going to be anything to do with that, which of course does include everything we've already been talking about.
22:38.314 --> 22:39.194
[SPEAKER_02]: And then are you ready?
22:39.835 --> 22:40.055
[SPEAKER_02]: Yes.
22:40.656 --> 22:40.996
[SPEAKER_02]: Awesome.
22:41.056 --> 22:42.597
[SPEAKER_02]: Let's start with new advice.
22:42.997 --> 22:48.422
[SPEAKER_02]: If we've inspired someone to take their first step, what do they need to know to give themselves the best chance of success?
22:49.342 --> 23:01.229
[SPEAKER_01]: I'm going to repeat myself by saying that the one thing that you really need to do is start by measuring a problem and from there on it's really easy to find the right tools that can help you optimize it.
23:02.950 --> 23:04.011
[SPEAKER_01]: So start with that.
23:04.391 --> 23:10.755
[SPEAKER_02]: Yeah, that one thing I don't think is ever going to change in conversion rates is yeah, we're start with the numbers.
23:11.575 --> 23:14.717
[SPEAKER_02]: Now once you've started, of course, you've got to keep optimizing.
23:14.757 --> 23:17.139
[SPEAKER_02]: So what's your favorite way to improve performance?
23:18.426 --> 23:32.921
[SPEAKER_01]: So one thing I really I think people forget a lot of times, but I feel it's so crucial is to do regular checkups on their stores and especially on their apps and what's left from the apps they deleted.
23:33.401 --> 23:37.025
[SPEAKER_01]: So my advice to you is like every six months go ahead.
23:37.405 --> 23:40.666
[SPEAKER_01]: check all the apps that you have in your Shopify store.
23:41.107 --> 23:42.867
[SPEAKER_01]: Make sure if you're using them.
23:43.527 --> 23:45.188
[SPEAKER_01]: Check if they're actually making your money.
23:45.248 --> 23:47.789
[SPEAKER_01]: If they're giving you the value of your thought, they're going to give you.
23:48.309 --> 23:50.350
[SPEAKER_01]: And which are on just delete them.
23:50.870 --> 23:56.253
[SPEAKER_01]: Like make sure that you delete it also like the small increments of code they usually leave behind.
23:57.273 --> 24:05.477
[SPEAKER_01]: what I'm saying that is because that's something people almost always forget, but this leads to problematic points like friction.
24:06.157 --> 24:09.859
[SPEAKER_01]: And the friction happens in the most unfortunate moments.
24:10.059 --> 24:18.783
[SPEAKER_01]: So when someone tries to click on to card button and this button doesn't work because you have so many apps that actually block that process.
24:19.844 --> 24:21.925
[SPEAKER_01]: That's very unfortunate.
24:22.145 --> 24:24.646
[SPEAKER_01]: So make sure to keep your
24:27.027 --> 24:29.949
[SPEAKER_01]: Thank you, Laker.
24:30.530 --> 24:31.511
[SPEAKER_02]: Brilliant advice.
24:32.071 --> 24:37.075
[SPEAKER_02]: If someone listing wants to learn more, is there a cheap or free resource you would recommend?
24:37.922 --> 24:38.122
[SPEAKER_01]: Yeah.
24:38.723 --> 24:48.749
[SPEAKER_01]: So whenever you want to learn more about conversion rate optimization, there are a lot of great resources in terms of podcasts and shows this being one of them.
24:49.389 --> 24:54.513
[SPEAKER_01]: I always say that chatjipiti or occult are always also a great start.
24:55.613 --> 25:01.357
[SPEAKER_01]: One thing I really like and like a resource that I don't see being talked enough in the industry
25:07.641 --> 25:18.888
[SPEAKER_01]: Yeah, by Sara Karamelagos every month she analyzes all Shopify teams and she measures what's their effect on user experience.
25:19.048 --> 25:26.152
[SPEAKER_01]: So if they make your site slower, if they call some friction points or if they call some instability.
25:26.172 --> 25:32.076
[SPEAKER_01]: So like if things are moving up and down your store and I think that's the first place.
25:35.778 --> 25:44.348
[SPEAKER_01]: just to make sure that the theme they use is actually optimized for good site and user, good speed and user experience.
25:44.828 --> 25:50.795
[SPEAKER_02]: What an awesome site and yet one which I suspect many of the things that I'm not sure I want to look.
25:51.975 --> 26:00.098
[SPEAKER_02]: You know, if you're a few years into using the same theme and you went on that website and discovered it was, it was not a good one, you would be like, oh, no.
26:00.398 --> 26:02.839
[SPEAKER_02]: Like, approach it your own risk, everybody that one.
26:03.099 --> 26:04.600
[SPEAKER_02]: So that was theme vitals.
26:05.240 --> 26:10.682
[SPEAKER_02]: And finally, crystal ball time, what's coming up in the next six to 12 months that we should be getting ready for?
26:11.853 --> 26:16.637
[SPEAKER_01]: I think that more AI is just going to enter this sphere.
26:17.598 --> 26:24.924
[SPEAKER_01]: As we just talked about a genetic conversion rate optimization, I'm foreseen that much more companies are going to jump on that.
26:25.264 --> 26:28.467
[SPEAKER_01]: We are going to see even more optimizations happening.
26:28.527 --> 26:31.270
[SPEAKER_01]: Like, I'll automate it and autonomous way.
26:32.130 --> 26:48.653
[SPEAKER_01]: Additionally, I feel that sidekick is going to get even more powerful, so I'm very excited to see how it starts to talk more with the apps that are currently in the app store, and maybe also some apps that are outside of the app store.
26:48.693 --> 26:58.875
[SPEAKER_01]: We'll see about that, but I just the ease of building a website and using AS every point of like building an optimizing a website that's going to
27:00.884 --> 27:09.411
[SPEAKER_02]: Now, I think you're spot on the, um, we are very nearly into the show though, so could you please at the list as now where they can find you and UXify?
27:10.292 --> 27:14.095
[SPEAKER_01]: Yes, uh, you can always visit our website, UXify.com.
27:14.596 --> 27:18.219
[SPEAKER_01]: We are also very active on LinkedIn, so make sure to give us a follow.
27:18.539 --> 27:26.646
[SPEAKER_01]: And yeah, if you ever want to talk more about conversion rate optimization and Daniela Gentiquey, we are here.
27:27.545 --> 27:38.284
[SPEAKER_02]: Awesome, there we go everybody and I know thank you so much for coming on and talking about this I think you'll have made AI and conversion rates a lot more understandable for everyone listening so we really appreciate it.
27:38.304 --> 27:39.105
[SPEAKER_02]: Thank you for being here.
27:39.827 --> 27:41.069
[SPEAKER_01]: Thank you for having me, Chloe.
27:43.032 --> 27:56.761
[SPEAKER_02]: So there you go, a clear guide to what's happening with AI and conversion rail to myization at the moment, I think you give it in those kind of three runways I was talking about, it's probably the best, my head, at least it's the best way to make it make sense.
27:56.781 --> 27:57.482
[SPEAKER_02]: You've got the...
27:58.302 --> 28:07.088
[SPEAKER_02]: The obvious stuff we should be doing day to day anyway, like image, optimization, fixing broken things, et cetera, et cetera, et cetera, that you get the AI handling so it just happens.
28:07.909 --> 28:26.202
[SPEAKER_02]: Then you've got the things that you can train AI agents to work on on the fly as people are actually experiencing the website, like preloading pages and all sorts of other clever clever things that just they're not the risky things that could destroy your site or damage your business, but they are the things which
28:26.902 --> 28:33.086
[SPEAKER_02]: are worth kind of ongoing testing, optimizing, improving, but with a small layer of humanet oversight especially to get them up and running.
28:33.847 --> 28:45.434
[SPEAKER_02]: And then that third runway, which is where you and your team, with some help from AI, where you need it, tackle the big stuff, the branding, the wording, all those big, big things.
28:45.695 --> 28:51.819
[SPEAKER_02]: And of course, that whole automated piece then starts to massively overlap in my head with the world of personalisation.
28:52.479 --> 28:57.905
[SPEAKER_02]: You can get links to everything we discussed, the transcripts are notes and more at keepoptimizing.com.
28:58.385 --> 29:03.471
[SPEAKER_02]: You can also use our direct episode short links that's keepopt.com for such the episode number.
29:04.091 --> 29:07.175
[SPEAKER_02]: And when you get to the website, you can add yourself to our email list.
29:07.655 --> 29:11.156
[SPEAKER_02]: Thank you for tuning in to this episode of the Keep Optimizing Podcast.
29:11.796 --> 29:16.158
[SPEAKER_02]: All five of our episodes revisiting the conversion rates throughout our now live.
29:16.558 --> 29:22.520
[SPEAKER_02]: Think of it as you'll get me up to date on conversion rates fast, masterclass, so please do have a listen to them all.
29:23.300 --> 29:32.583
[SPEAKER_02]: Either scroll up your podcast feed and you'll find them or use the short link, keep up.com forward slash CRO and our next cast masterclass topic.
29:32.883 --> 29:33.363
[SPEAKER_02]: Ready for this?
29:33.803 --> 29:36.564
[SPEAKER_00]: We are doing Google Ads.
29:37.234 --> 29:41.337
[SPEAKER_02]: Yes, we're going back to the world of PMAX and Google ads.
29:41.897 --> 29:45.739
[SPEAKER_02]: And we are going to be revisiting all of that in a couple of weeks.
29:46.059 --> 29:51.283
[SPEAKER_02]: You can get that catch that thread so far even at keepopt.com for such Google ads.
29:51.343 --> 29:54.665
[SPEAKER_02]: Just do that and you'll go straight to the right page on the website or scorn on your feed.
29:55.385 --> 30:02.891
[SPEAKER_02]: make sure you tune in in two Wednesday's time to catch the start of our dive into all that's new that you need to know about Google Ads.
30:03.391 --> 30:05.833
[SPEAKER_02]: So yes there is no episode next week.
30:06.173 --> 30:17.202
[SPEAKER_02]: That's going to give you time to catch up on any of the CRO episodes that you missed and re-list into any that you need to or share them with your team and then put into action what you've learned, which is the most important thing to do.
30:17.722 --> 30:23.124
[SPEAKER_02]: Make sure you subscribe to us or following us on your favourite podcast player so you don't miss out when we come back in two weeks.
30:23.564 --> 30:35.209
[SPEAKER_02]: And if you know someone who's particularly interested in improving their conversion rates or their Google ads, please tell them we're covering it all as this show exists to help you and them improve your marketing.
30:35.909 --> 30:40.831
[SPEAKER_02]: Have a great week and make sure you listen to the next episode so I can help you to keep optimizing your marketing.
30:44.035 --> 30:46.040
[SPEAKER_00]: Access everything, keep optimizing.
30:46.722 --> 30:50.290
[SPEAKER_00]: At keepoptimizing.com, that's within us, not as in.
