WEBVTT
00:00:28.559 --> 00:00:37.439
Econometry is a easy way to do the marketing directory and David Granger, media strategist and publication editor.
00:00:46.240 --> 00:00:56.799
Marketing marketing is developing the case for marketing and brand investment in the short, medium, and long term to internal and external stakeholders and of course to the C-suite.
00:00:57.039 --> 00:01:02.159
And as always with Fworks, we aim to do the how as well as the what and why.
00:01:02.320 --> 00:01:06.959
So in our next session, which was, I have to say, mentally moderated by Jet Cook.
00:01:07.120 --> 00:01:09.200
Unfortunately, you have me again.
00:01:09.280 --> 00:01:15.200
Um, we're gonna hear how econometrics is the key to developing the case for marketing in the C-suite.
00:01:15.439 --> 00:01:16.480
So, hello.
00:01:16.719 --> 00:01:18.000
Do you want to introduce yourselves?
00:01:18.400 --> 00:01:20.239
Yes, uh, I'm David Granger.
00:01:20.319 --> 00:01:23.120
I'm a media strategist of 25 years.
00:01:23.200 --> 00:01:28.879
I've worked at every media agency in London, like a weird personal game of Pokemon.
00:01:29.040 --> 00:01:40.799
Um, but in the context of this conversation, I've edited the brand new uh IPA publication uh called Econometrics in the C-suite, which you can pick up when you sort of leave uh the building this evening.
00:01:41.040 --> 00:01:49.040
Um yeah, I was asked to edit this publication, completed it a couple of a couple of weeks ago, forgotten everything about it.
00:01:49.200 --> 00:01:53.599
So um yeah, if if we're a little bit rusty, then please forgive us.
00:01:54.319 --> 00:01:57.920
Uh and I'm Ross Farker, I'm the marketing director at LittleMoon's.
00:01:58.079 --> 00:02:06.000
If you um haven't tried us, though I hear that we've been mentioned a couple of times already today, so I'm clearly doing a very good salience job, uh primarily through conferences.
00:02:06.319 --> 00:02:07.519
Um please do.
00:02:07.680 --> 00:02:16.319
Um and my background is partly client side, Cadbury, uh TiaGio, Wakamama, um, and partly agency side at 101 uh and great.
00:02:16.400 --> 00:02:20.639
And econometrics has been a running theme in different ways uh throughout my career.
00:02:21.039 --> 00:02:21.439
Brilliant.
00:02:21.520 --> 00:02:22.080
Well, welcome.
00:02:22.159 --> 00:02:26.560
And hopefully the questions will help push you and remember what this is all about.
00:02:26.719 --> 00:02:33.120
So the session description for this panel talks about how econometrics can bring a louder voice to the C-suite for marketing.
00:02:33.280 --> 00:02:34.400
Is that true?
00:02:35.039 --> 00:02:36.639
Well uh Ross.
00:02:37.680 --> 00:02:38.159
Brilliant.
00:02:39.759 --> 00:03:08.000
Well, do you know I am so I I think louder is the right word for this because uh for me, econometrics is primarily brilliant if it's about reframing the marketer as the person who genuinely understands the drivers of demand, of which marketing is one of them, and of which it places more rigor into the marketing bit, rather than being the marketer who walks in with the evidence for what they've been saying all along to the C-suite.
00:03:08.080 --> 00:03:14.240
So it does in the end give you more credibility and authority, so louder feels appropriate.
00:03:14.319 --> 00:03:28.960
But I do tend to think too many marketers do trend to tend to treat it as the evidence we've all been searching for that uh marketing is is real and not just a thing that you have to pay begrudgingly when you're you're the CEO or the CFO.
00:03:29.280 --> 00:03:44.639
Uh instead, it's more successful, and you find your voice becoming louder when you're suddenly able to go in and talk to the C-suite about all the different things that are driving demand in your business and ultimately create something that is of utility to your colleagues in the C-suite.
00:03:44.800 --> 00:03:49.680
So, you know, my best successes have been when the sales director will go, God, that's really interesting.
00:03:49.759 --> 00:03:51.199
I'm gonna go and do something with that.
00:03:51.360 --> 00:04:03.840
When uh when we found out at Wagamama that, like, I mean, of course it seems uh natural, but like that weather was a really big determinant of our demand, or that um refurbing restaurants was a big determinant of our our demand.
00:04:04.000 --> 00:04:08.560
So if you're the property director or you're the person doing the sales forecast, that's really, really useful.
00:04:08.639 --> 00:04:16.079
It makes you more inclined to listen to the other bit that I'll go in and say and be like, oh, and by the way, tele pays back really well, you should do more of that, give me some money.
00:04:16.399 --> 00:04:18.959
So less louder and more credible playbacks.
00:04:19.199 --> 00:04:20.240
Yeah, okay, cool.
00:04:20.480 --> 00:04:24.319
And what have been the issues in the past for econometrics, do you think?
00:04:25.120 --> 00:04:40.079
I think I think one of the certainly one of the issues that um uh that agency partners and I guess sort of new CMOs and new brand managers run into is is just that sort of change of regime.
00:04:40.240 --> 00:04:45.600
I think econometrics has typically sort of come within expiry date previously.
00:04:45.680 --> 00:04:48.319
And I think it it it's it's very often wrong.
00:04:48.399 --> 00:05:07.279
I mean, you only have to sort of um uh look at the long and short of it, and when the guys are talking about the sort of 10-year roles uh rules that sort of thing that still hold true today, typically if you've sort of commissioned an an econometric study sort of even you know sort of three or four years ago, those same laws do apply to um your business today.
00:05:07.360 --> 00:05:19.199
But I think too many sort of marketers are keen to just sort of say, well, hang on, that's that's out of date now, and we need to sort of you know keep refreshing the case study, which is I think is is uh just wrong a lot of the time.
00:05:19.439 --> 00:05:25.040
Um I think yeah, that that I think for me that's that's one of the um that's one of the big issues.
00:05:25.120 --> 00:05:45.279
And the the other thing is um is just not having um if you're if you are a sort of CMO, if you are a brand manager, just not having the sort of technical understanding behind the sort of uh the econometric model which you happen to be sort of sharing with um with your peers in the boardroom.
00:05:45.439 --> 00:05:52.480
Um I was lucky enough to be a sort of technical judge uh for the IPA effectiveness awards in 2022.
00:05:52.720 --> 00:06:02.240
And for the first time ever, I had to sort of delve into the robustness of the econometric model that sat behind every single sort of case study.
00:06:02.399 --> 00:06:10.480
Um and to educate yourself about how the model is built and the claims being made by the model in the first instance.
00:06:10.639 --> 00:06:20.160
I'm not sure that every marketer or CMO is sort of equipped with that knowledge when they um go into bat on behalf of sort of marketing budget in the boardroom.
00:06:20.560 --> 00:06:21.279
They definitely aren't.
00:06:21.439 --> 00:06:25.680
I mean, just to be clear, like I'm not a mathematician, I could not do that.
00:06:25.839 --> 00:06:37.439
Um and yeah, I do have to be the one that goes in and sells econometrics to my business, so I need to be able to explain what it is beyond just saying it's a black box that's going to give us all the answers.
00:06:37.600 --> 00:06:46.240
Um, and then I need to get to the other side on a presentation where there will be at least one person on your board who is a mathematician at heart, right?
00:06:46.319 --> 00:06:48.319
And will want to ask you lots of really tricky questions.
00:06:48.560 --> 00:06:49.759
You're like, I don't I don't really know.
00:06:49.839 --> 00:06:52.480
I've sort of got to the ends of my understanding.
00:06:52.639 --> 00:06:58.399
Um so that is really hard because marketers don't tend to be people who build models in their spare time.
00:06:58.720 --> 00:07:01.199
They're not that that kind of personality.
00:07:01.360 --> 00:07:18.399
So for me, one of the most important things has been getting the econometricians that I work with to get to know all of my board stakeholders along the way, not least because it builds belief in them as well, but also because it means they can answer some of the more tricky questions about what this is all about, right?
00:07:18.480 --> 00:07:29.120
So my CEO, for example, is one of those sales directors, turns CEO who is absolutely shithop, pardon my language, on numbers and modelling.
00:07:29.199 --> 00:07:31.680
He'll open up an Excel and build his own models himself.
00:07:31.839 --> 00:07:33.519
I'm never going to compete with that.
00:07:33.680 --> 00:07:41.040
So I needed Grace Kite, who is around somewhere, who just did the project with us last, to be in the room and be able to answer some of those tricky questions.
00:07:41.120 --> 00:07:44.480
Otherwise, it we the model wouldn't have had the belief in it among stakeholders.
00:07:44.879 --> 00:07:50.800
And do you think marketers are too protective of it at the moment, or is it just fear of having to present it?
00:07:51.120 --> 00:08:01.680
Uh I as I say, I think it's partly that marketers are um fearful of having to answer the tricky questions about how these numbers came out, other than just saying it's a black box.
00:08:01.920 --> 00:08:06.240
Um and partly also I think just looking at it too narrowly.
00:08:06.399 --> 00:08:18.800
I talked to quite a lot of marketers who do really go into these projects just thinking what I really want is the bar chart that shows me the ROI on marketing that I can then go off uh and tell everyone about how well it's going.
00:08:18.959 --> 00:08:28.160
And they kind of miss the fact that there's lots more useful and interesting stuff in there that, as I say, builds belief amongst the the broader stakeholder group.
00:08:28.480 --> 00:08:29.600
Yeah, I think I think that's right.
00:08:29.759 --> 00:08:36.000
I think we, you know, we're we're here at the Institute of Practitioners in advertising sort of conference.
00:08:36.080 --> 00:08:43.759
So we tend to sort of focus on advertising's contribution to sales, to uh to revenue, um, to profit.
00:08:43.919 --> 00:08:53.759
And you know, if you've ever looked at a sort of stacked bar chart and worked out that our contribution is like 4% of sort of total sales, then um sometimes you think, why do we bother?
00:08:53.919 --> 00:09:02.879
Uh but uh but you know, I think you forget about the interdependency with all the other sort of uh business drivers with within an organization.
00:09:02.960 --> 00:09:06.559
Um, and that's what we need to be sort of more cognizant of.
00:09:06.879 --> 00:09:10.399
Yeah, and when we heard you talk about it, is that inputs and outputs, isn't it?
00:09:10.480 --> 00:09:13.440
Like it's all it's all the elements that make that up.
00:09:13.600 --> 00:09:19.600
Um, so then how important is it to bring other departments into the conversation and leaders into this process?
00:09:20.159 --> 00:09:58.320
I mean, absolutely crucial from my side, particularly your CFO or your finance director, or frankly your whole finance team, and not just because they have to provide lots of the inputs typically, but but because good looks like you walking into the boardroom almost arm in arm with your CFO, with them going, This is my model too, and I believe in it, and I've got lots of things that you're going to learn from it, and that takes time and them being involved all the way through the process, and often actually, even if they can't be the ones in every single meeting, that the people on their team are the ones who've been really involved in it because they're going to be more effective messengers to your CFO than you are as the marketer, even though they're they're your peer.
00:09:58.399 --> 00:10:00.080
So I think that's really important.
00:10:00.240 --> 00:10:11.039
I think the other thing I'd say is like all good econometrics models are built on good hypotheses up front, and the marketer doesn't have a monopoly on the hypotheses of what's driving demand.
00:10:11.200 --> 00:10:20.159
And certainly in every model that I've ever been a part of, we've had these brilliant workshops at the start where everyone's chucking in their own hypotheses about what could be driving it.
00:10:20.240 --> 00:10:21.440
And there's been stuff that I've never heard.
00:10:21.519 --> 00:10:23.120
I'm like, oh, wow, that's really interesting.
00:10:23.200 --> 00:10:23.919
Where did that come from?
00:10:24.000 --> 00:10:27.039
And then it proves out in the model to be a really big, big driver.
00:10:27.200 --> 00:10:32.720
So again, it just builds buy-in and you get interesting stimulus that you wouldn't get on your own.
00:10:33.919 --> 00:10:35.840
And should you be doing that at the very beginning?
00:10:36.159 --> 00:10:37.440
I I I think so.
00:10:37.600 --> 00:10:47.120
I mean, it's kind of a bit like you're all agency folk as well, like when you have the kickoff at the start where clients get to chuck in their instincts and you politely go, Yeah, okay, sure.
00:10:47.360 --> 00:10:49.759
We're not going to be going anywhere near that, but thanks very much.
00:10:49.919 --> 00:10:55.600
Um there is a bit of that upfront that's just bringing people into the process.
00:10:55.679 --> 00:11:04.399
And as I say, you go in perhaps cynically about it, and then you find loads of really good stuff that you weren't expecting, and it enhances the process that then follows.
00:11:05.440 --> 00:11:05.919
Cool.
00:11:06.240 --> 00:11:18.080
And in this report, which you will all have a copy of, have you heard at the beginning, um, there's a lot of reporting uh around the differences AI can bring to econometrics, and particularly relating to speed and updating the data.
00:11:18.320 --> 00:11:18.480
Yeah.
00:11:18.720 --> 00:11:19.840
How have you found this?
00:11:20.399 --> 00:11:24.559
I think I mean this this was this was sort of brand new knowledge to me.
00:11:24.639 --> 00:11:42.559
You know, this I was entering into this world for the first time, and um there's a lovely sort of chapter at the end of this publication where uh Dom and Winnie from Cantar summarize the potential contribution of sort of AI to sort of econometric modelling, but they do stress that this is in its infancy.
00:11:42.639 --> 00:11:53.440
And I think um uh what I think what's interesting when you read this sort of chapter back, what there's so much um sort of humanity in that's in that sort of chapter.
00:11:53.519 --> 00:12:06.720
So what they're really talking about is you know, still establishing as a sort of human hypothesis for what you want um econometrics to sort of prove, and um establishing uh uh the right sort of data capability.
00:12:06.799 --> 00:12:15.279
So it's not like we're just sort of pressing a button and just handing over control of building models to um you know some weird sort of supercomputer somewhere.
00:12:15.440 --> 00:12:23.840
Um I think uh what what I love about this chapter in particular is that to date, um and the oh there's some people coming on to talk about AI in a minute, isn't there?
00:12:24.000 --> 00:12:26.559
So um I'm gonna sort of damn them uh right now.
00:12:26.720 --> 00:12:29.840
So I think AI's contribution to marketing.
00:12:30.080 --> 00:12:39.440
For some reason, we've we've deemed it, deemed that its role is to sort of build lots of stuff really, really quickly, be that sort of crap creative uh mostly.
00:12:39.600 --> 00:12:44.399
Whereas I think AI's greatest contribution could be providing foresight.
00:12:44.559 --> 00:13:11.600
I think the um uh I got to I got to spend some time with Dom talking about AI, and he told me this lovely sort of anecdote about a client um using so in instead of instead of econometrics being seen as a sort of presentation, you need to sort of think of it more as a mixing desk, which a sort of boardroom can sort of huddle around and sort of pull the levers and sort of start to sort of develop sort of better scenario planning capabilities, uh, which I I love that.
00:13:11.679 --> 00:13:12.080
I love that.
00:13:12.159 --> 00:13:17.200
So it's still it's not it's not that you're losing the rigour of building the model in the first place.
00:13:17.440 --> 00:13:24.240
AI is simply enabler to sort of play with some of the um uh some of the sort of levers within the model.
00:13:24.320 --> 00:13:27.200
And I thought I thought that was a really sort of uh nice development.
00:13:27.519 --> 00:13:32.480
So are we still in the experimental stage or we're actually seeing evidence of it now using AI?
00:13:32.879 --> 00:13:36.559
Uh well as I said, I mean this is completely new to me.
00:13:36.879 --> 00:13:58.639
Uh, but it's great that there are more sort of uh potential applications for um uh for more uh for using AI within econometrics, and there's more and more uh people like sort of Google and Meta who's who are sort of developing um sort of modelling capabilities as well, uh very much sort of focused on AI.
00:13:59.440 --> 00:13:59.759
Right.
00:13:59.919 --> 00:14:00.240
Okay.
00:14:00.720 --> 00:14:06.399
And then what kind of level of detail do boards want to see in these econometric reports?
00:14:06.480 --> 00:14:07.919
I mean, how much is too much?
00:14:09.039 --> 00:14:14.320
I mean, in my case, my board want to see more detail than they have the time to actually engage with, right?
00:14:14.399 --> 00:14:22.080
So left to their own devices, they would happily spend three hours um going through all of the detail, lapping it all up, being fascinated by it.
00:14:22.320 --> 00:14:39.039
Um probably not the best use of their time, but but um if you can draw out the key themes with enough detail that gives them confidence in the result, and you make sure those themes aren't just about the marketing bit you want to persuade them of, that's really effective.
00:14:39.200 --> 00:14:50.720
I think the other thing is picking off your stakeholders who you just know are fascinated by this stuff, whether it's whether it's their role to be or not, is just helpful because they will be the ones who are your biggest advocates in the room.
00:14:50.960 --> 00:15:05.279
So you know, I can think of examples where certainly at Wagamama, like I went and did the little tour to the the CEO and the CFO of the restaurant group, the parent company, because I knew that they were going to be really, really interested in this before we got into a proper board setting.
00:15:05.440 --> 00:15:13.600
And it just meant that when we did the presentation at board, I wasn't getting lots of tricky questions that were expanding the whole thing out um in the room.
00:15:13.759 --> 00:15:15.919
They were advocates on on behalf of it.
00:15:16.080 --> 00:15:18.159
So yeah, I think detail's a good thing.
00:15:18.240 --> 00:15:21.600
You just might not want to go through it all in the room in the presentation.
00:15:22.000 --> 00:15:23.679
No your audience, yeah, exactly.
00:15:23.919 --> 00:15:42.000
I think well, I think what the what um econometrics and the C suite does, it provide if you if you're a CMO and you and you're not working with uh an econometric model at present, this this sort of creates a sense of sort of professional envy because it's almost like a sales pitch for you know putting a sort of model in place.
00:15:42.159 --> 00:15:56.559
Um and you know, Ross uh works with sort of relatively sort of small advertising budgets compared to um a lot of brands, and yet it I think his is is great testimony to the sort of power of having an econometric model to help him make the sort of right decisions.
00:15:56.720 --> 00:16:07.360
And I think what when I when I was the technical judge last year, it was amazing just how uh I mean it was it was really beneficial sometimes.
00:16:07.679 --> 00:16:22.639
Sometimes the the sort of simplest models, the most sort of parsimonious models, were were the sort of best because they didn't allow for lots of assumptions, they were just trying to prove prove the contribution of sort of um three or four factors on sales of revenue or profitability.
00:16:22.799 --> 00:16:28.320
And I think it it gave you confidence that any business could actually sort of build their own sort of model.
00:16:28.639 --> 00:16:30.559
And it's also really, I mean, you're right.
00:16:30.639 --> 00:16:51.200
I I have that econometric models now have been part of those projects for smaller businesses relatively or smaller media spends, but it does frustrate me so much that there's this perception that econometrics is just for the big media spenders, that um, unless I'm spending 20 million plus a year and have insight budgets that are absolutely huge, econometrics isn't for me.
00:16:51.360 --> 00:17:02.639
Um, and I just find the almost the opposite is true, where for small businesses you're so lacking in real, genuine, rigorous insight that will tell you what the right choices to make are.
00:17:02.879 --> 00:17:09.680
And in the grand scheme of things, the cost of an econometric model is a lot less than I think people have in their heads it is typically.
00:17:09.920 --> 00:17:26.640
Like for both of the last two businesses I've worked for, they've been absolutely invaluable, and we've almost got more out of them because we're so motivated to eke out every last bit of insight out of them than the ones I've known for larger organizations that I've I've either been a part of or or worked with when I was agency side, you know.
00:17:26.880 --> 00:17:27.200
Right.
00:17:27.440 --> 00:17:32.079
So then how important is it to find the right external partner to work with?
00:17:32.240 --> 00:17:34.079
And um how how do you do that?
00:17:34.160 --> 00:17:40.079
What what what's your criteria for selection so that it doesn't look like you're making up your own homework or all those things we can often hear?
00:17:40.559 --> 00:17:42.160
I think that is the hardest bit, right?
00:17:42.240 --> 00:17:44.480
Because you're doing this primarily.
00:17:44.640 --> 00:17:45.920
Well, actually, I'll take that back.
00:17:46.240 --> 00:18:07.599
You're either doing it because you're a big media spender and econometrics are a really effective way of you working out how to optimize that, in which case I actually think there is quite a good rationale for why you might work with some of the econometricians within the media agency you're working with, because they're much closer to the action and the things that you might do with that model make sense.
00:18:07.680 --> 00:18:08.799
But exactly what you say, Karen.
00:18:09.119 --> 00:18:09.920
It's a downside.
00:18:10.000 --> 00:18:20.079
I think that board stakeholders are more likely to think, oh, what a shock, media agency declares media works really well, uh, and your whole thing is is undermined.
00:18:20.240 --> 00:18:40.240
Whereas I think when you when your objective is more a I want to understand the broader drivers of demand of which marketing investment is part of it, there is a bit more merit of having separation of powers here and having someone who is independent of any of your agencies, um, or effectively your organization as well.
00:18:40.319 --> 00:18:48.400
Some businesses have in-house econometricians as well, but like someone independent who will come and provide fresh eyes and go, no, I'm going to adjudicate here.
00:18:48.559 --> 00:19:04.480
So because of the businesses I've worked in, I I tend to work with econometricians who are um independent of any of those stakeholders, and I've found that's been helpful, but I I totally appreciate if I was in a global multinational, I might take a different view because my spend is much bigger.
00:19:06.480 --> 00:19:06.720
Okay.
00:19:08.640 --> 00:19:22.000
Yeah, I mean, just to add to that, I think there's I think there's certainly over my career, um, it was uh there was a fierce rivalry over sort of over sort of building econometric models about sort of 15 years ago.
00:19:22.160 --> 00:19:25.200
It was just a better margin business than sort of media as a whole.
00:19:25.440 --> 00:19:32.559
Um but uh but luckily I think you know we're that sort of rivalry, rivalry has sort of calmed down a bit.
00:19:32.640 --> 00:19:42.079
So my own lived experience um sort of recently working with IKEA, so Olgra, I'm not sure if you see it, Olgra at MediaCom runs this sort of econometric model for IKEA.
00:19:42.400 --> 00:19:52.799
And while when I was at Dentu at iProspect, I was sort of uh you know doing the sort of media buying and sort of taking the findings of the econometric model and applying them to uh to the advertising budget.
00:19:53.039 --> 00:19:58.960
So I think there's there's sort of less rivalry over ownership of the data than there's than there's sort of been previously.
00:19:59.440 --> 00:20:11.039
Uh I I I I pr I completely understand the sort of um the concern about if you're if you're an advertiser then trusting the agency to sort of run their own sort of model.
00:20:11.279 --> 00:20:12.559
I completely understand that.
00:20:13.440 --> 00:20:13.759
Okay.
00:20:14.160 --> 00:20:23.359
Um if there's anyone in the audience here uh who wants to start this type of a journey and uh you know uh how do they know if they will have enough of the right kind of data?
00:20:23.519 --> 00:20:24.880
What do you recommend?
00:20:25.680 --> 00:20:28.960
And how long would it take before you actually see return on it?
00:20:29.119 --> 00:20:30.480
Because we're all about return.
00:20:30.799 --> 00:20:39.200
I think I think the well I think going through this process, what you realise is that there's a dearth of good econometricians out there.
00:20:39.359 --> 00:20:41.039
It's like finding a builder, basically.
00:20:41.279 --> 00:20:48.559
So I think a lot of the delay is almost you know finding the right practitioner who can actually sort of help you um straight away.
00:20:48.799 --> 00:20:53.279
Um I'd love I think I should retrain as an econom uh econometrician basically.
00:20:53.519 --> 00:20:58.559
But um I think once the model is built, you know, the I mean Ross, you're very familiar with this.
00:20:58.640 --> 00:21:03.680
The you know the findings are sort of they come on board sort of quite quickly and the things you can apply quite quickly.
00:21:04.079 --> 00:21:07.519
But um I don't know about I don't know about the sort of the sort of data taxonomy.
00:21:07.680 --> 00:21:13.680
I think I think they can start a lot simpler than you actually sort of give them credit for and then sort of build over time.
00:21:14.079 --> 00:21:15.519
Yeah, that was my experience, right?
00:21:15.680 --> 00:21:22.720
So it answered the question if you're if you're thinking about it but you're not sure you've got enough data, the the answer is to ask, right?
00:21:22.880 --> 00:21:36.480
So I uh when I first started thinking about econometrics at Little Moons, it was just after I'd sold into the boards the idea that it was about time in our journey for us to start investing in paid media.
00:21:36.559 --> 00:21:50.640
So it was well before we actually made an ad, bought any media, and I was already thinking about how am I going to measure this because if we do it once and it's perceived that it doesn't work, then I'm scuppered, right?
00:21:50.720 --> 00:21:53.519
So so we have to measure this, and you think about it.
00:21:53.599 --> 00:21:59.759
But I also had many of the limiting beliefs that I'm sure many people um have experienced of going, oh God, we're not buying enough.
00:22:00.480 --> 00:22:08.480
Epos data and and our maybe our finance um capability isn't quite what it would be in um in much larger organizations.
00:22:08.640 --> 00:22:27.359
Um and I vividly remembered like I I went for a coffee with Grace because I knew Grace because I'd worked with her on um on the Wacamama model, and I just kind of said, Would you do me a favor and spend half an hour just talking to me about um when would be the right time for us to do this kind of modelling and what kind of data would we need uh in order to do it?
00:22:27.440 --> 00:22:41.440
And I walked away from that conversation going, oh, actually, we'll probably be at the point where we'll be buying that EPOS data that we need by the time that the media goes out on air anyway, and I can buy it um backdated anyway, so I'll have it.
00:22:41.519 --> 00:22:45.119
And then, you know, we're talking about the costs involved, now it's probably manageable as well.
00:22:45.200 --> 00:22:58.400
So you come away from it feeling really reassured um that actually the answer is probably you can do it earlier than you think you can, and you probably should do it earlier than um than than you think you can.
00:22:58.559 --> 00:23:02.559
So um, so yeah, absolutely asking is the answer.
00:23:02.640 --> 00:23:16.400
And and David's point, like finding a practitioner who you've worked with in the past or who you trust, or or whoever is the invaluable thing there, because then you've got a sounding board who can just take away some of those beliefs and and make it concrete for you.
00:23:17.359 --> 00:23:29.920
Yeah, I think I think the you know, uh any sort of an objective of econometrics is just to sort of turn a hypothesis or a qualitative belief into a quantitative statement.
00:23:30.079 --> 00:23:36.480
So instead of just saying, oh, I think if we spend 500 grand on advertising this year, I think we'll be in a better place than we were, you know, yesterday.
00:23:36.720 --> 00:23:42.000
It's to actually understand that for every one pound you spend, you're gonna get one pound sort of 20 returns.
00:23:42.079 --> 00:23:49.759
So you've got to go and find the simple data sets which are gonna sort of go and prove that hypothesis and and and believe it to be true.
00:23:49.839 --> 00:23:51.119
So I think it is better.
00:23:51.519 --> 00:23:53.920
It's simpler than a lot of people make out, isn't it?
00:23:54.240 --> 00:24:02.559
Just that I don't know, I think I think we've sort of dressed it up to be something scary because the best models can do so much.
00:24:04.240 --> 00:24:06.799
And in editing this, then what was the most surprising thing?
00:24:06.880 --> 00:24:10.799
I mean, obviously we've talked about the AIPs, but what else did you think was illuminating?
00:24:11.119 --> 00:24:24.319
I think I think look, if you're if if you sat here now and you're an agency practitioner um and you're like one of those people who like to shut shoot their mouth off in meetings about um you know, like you know how agency people do that?
00:24:24.480 --> 00:24:28.720
Say, oh maybe we should change your product or maybe we should sort of change your distribution network or things like that.
00:24:28.960 --> 00:24:30.000
Just with no idea.
00:24:41.039 --> 00:24:42.960
So I think that was one of the sort of keys findings.
00:24:43.200 --> 00:24:44.000
It's all there.
00:24:44.160 --> 00:24:48.079
The models are fantastic and they can show you the contribution of everything.
00:24:48.240 --> 00:24:51.599
So before you go crazy, just yeah, do your homework basically.
00:24:51.839 --> 00:24:52.880
It's also quite humbling.
00:24:52.960 --> 00:24:59.839
I think when I was at agency side, and I'd see that and I'd be like, to your point, advertising is contributing three, four percent of your base.
00:24:59.920 --> 00:25:11.839
And it does give you a bit of an understanding of why clients uh tend to be kept up at night by things other than the media bits, because it can go very, very wrong with just a couple of other things going um going a bit off, you know.
00:25:12.240 --> 00:25:12.799
Brilliant.
00:25:13.039 --> 00:25:13.279
Okay.
00:25:13.599 --> 00:25:15.920
Well, I think that has been incredibly helpful.
00:25:16.000 --> 00:25:24.000
And uh make sure you read your copy and you'll all be experts at the at the end of it, and you won't make yourself look like a fool of the meeting, which is always a good thing.
00:25:24.400 --> 00:25:24.880
Objective done.
00:25:25.440 --> 00:25:25.839
Amazing.
00:25:26.000 --> 00:25:26.880
Thank you both very much.
00:25:28.160 --> 00:25:28.559
Thank you.
00:00:28.559 --> 00:00:37.439
Econometry is a easy way to do the marketing directory and David Granger, media strategist and publication editor.
00:00:46.240 --> 00:00:56.799
Marketing marketing is developing the case for marketing and brand investment in the short, medium, and long term to internal and external stakeholders and of course to the C-suite.
00:00:57.039 --> 00:01:02.159
And as always with Fworks, we aim to do the how as well as the what and why.
00:01:02.320 --> 00:01:06.959
So in our next session, which was, I have to say, mentally moderated by Jet Cook.
00:01:07.120 --> 00:01:09.200
Unfortunately, you have me again.
00:01:09.280 --> 00:01:15.200
Um, we're gonna hear how econometrics is the key to developing the case for marketing in the C-suite.
00:01:15.439 --> 00:01:16.480
So, hello.
00:01:16.719 --> 00:01:18.000
Do you want to introduce yourselves?
00:01:18.400 --> 00:01:20.239
Yes, uh, I'm David Granger.
00:01:20.319 --> 00:01:23.120
I'm a media strategist of 25 years.
00:01:23.200 --> 00:01:28.879
I've worked at every media agency in London, like a weird personal game of Pokemon.
00:01:29.040 --> 00:01:40.799
Um, but in the context of this conversation, I've edited the brand new uh IPA publication uh called Econometrics in the C-suite, which you can pick up when you sort of leave uh the building this evening.
00:01:41.040 --> 00:01:49.040
Um yeah, I was asked to edit this publication, completed it a couple of a couple of weeks ago, forgotten everything about it.
00:01:49.200 --> 00:01:53.599
So um yeah, if if we're a little bit rusty, then please forgive us.
00:01:54.319 --> 00:01:57.920
Uh and I'm Ross Farker, I'm the marketing director at LittleMoon's.
00:01:58.079 --> 00:02:06.000
If you um haven't tried us, though I hear that we've been mentioned a couple of times already today, so I'm clearly doing a very good salience job, uh primarily through conferences.
00:02:06.319 --> 00:02:07.519
Um please do.
00:02:07.680 --> 00:02:16.319
Um and my background is partly client side, Cadbury, uh TiaGio, Wakamama, um, and partly agency side at 101 uh and great.
00:02:16.400 --> 00:02:20.639
And econometrics has been a running theme in different ways uh throughout my career.
00:02:21.039 --> 00:02:21.439
Brilliant.
00:02:21.520 --> 00:02:22.080
Well, welcome.
00:02:22.159 --> 00:02:26.560
And hopefully the questions will help push you and remember what this is all about.
00:02:26.719 --> 00:02:33.120
So the session description for this panel talks about how econometrics can bring a louder voice to the C-suite for marketing.
00:02:33.280 --> 00:02:34.400
Is that true?
00:02:35.039 --> 00:02:36.639
Well uh Ross.
00:02:37.680 --> 00:02:38.159
Brilliant.
00:02:39.759 --> 00:03:08.000
Well, do you know I am so I I think louder is the right word for this because uh for me, econometrics is primarily brilliant if it's about reframing the marketer as the person who genuinely understands the drivers of demand, of which marketing is one of them, and of which it places more rigor into the marketing bit, rather than being the marketer who walks in with the evidence for what they've been saying all along to the C-suite.
00:03:08.080 --> 00:03:14.240
So it does in the end give you more credibility and authority, so louder feels appropriate.
00:03:14.319 --> 00:03:28.960
But I do tend to think too many marketers do trend to tend to treat it as the evidence we've all been searching for that uh marketing is is real and not just a thing that you have to pay begrudgingly when you're you're the CEO or the CFO.
00:03:29.280 --> 00:03:44.639
Uh instead, it's more successful, and you find your voice becoming louder when you're suddenly able to go in and talk to the C-suite about all the different things that are driving demand in your business and ultimately create something that is of utility to your colleagues in the C-suite.
00:03:44.800 --> 00:03:49.680
So, you know, my best successes have been when the sales director will go, God, that's really interesting.
00:03:49.759 --> 00:03:51.199
I'm gonna go and do something with that.
00:03:51.360 --> 00:04:03.840
When uh when we found out at Wagamama that, like, I mean, of course it seems uh natural, but like that weather was a really big determinant of our demand, or that um refurbing restaurants was a big determinant of our our demand.
00:04:04.000 --> 00:04:08.560
So if you're the property director or you're the person doing the sales forecast, that's really, really useful.
00:04:08.639 --> 00:04:16.079
It makes you more inclined to listen to the other bit that I'll go in and say and be like, oh, and by the way, tele pays back really well, you should do more of that, give me some money.
00:04:16.399 --> 00:04:18.959
So less louder and more credible playbacks.
00:04:19.199 --> 00:04:20.240
Yeah, okay, cool.
00:04:20.480 --> 00:04:24.319
And what have been the issues in the past for econometrics, do you think?
00:04:25.120 --> 00:04:40.079
I think I think one of the certainly one of the issues that um uh that agency partners and I guess sort of new CMOs and new brand managers run into is is just that sort of change of regime.
00:04:40.240 --> 00:04:45.600
I think econometrics has typically sort of come within expiry date previously.
00:04:45.680 --> 00:04:48.319
And I think it it it's it's very often wrong.
00:04:48.399 --> 00:05:07.279
I mean, you only have to sort of um uh look at the long and short of it, and when the guys are talking about the sort of 10-year roles uh rules that sort of thing that still hold true today, typically if you've sort of commissioned an an econometric study sort of even you know sort of three or four years ago, those same laws do apply to um your business today.
00:05:07.360 --> 00:05:19.199
But I think too many sort of marketers are keen to just sort of say, well, hang on, that's that's out of date now, and we need to sort of you know keep refreshing the case study, which is I think is is uh just wrong a lot of the time.
00:05:19.439 --> 00:05:25.040
Um I think yeah, that that I think for me that's that's one of the um that's one of the big issues.
00:05:25.120 --> 00:05:45.279
And the the other thing is um is just not having um if you're if you are a sort of CMO, if you are a brand manager, just not having the sort of technical understanding behind the sort of uh the econometric model which you happen to be sort of sharing with um with your peers in the boardroom.
00:05:45.439 --> 00:05:52.480
Um I was lucky enough to be a sort of technical judge uh for the IPA effectiveness awards in 2022.
00:05:52.720 --> 00:06:02.240
And for the first time ever, I had to sort of delve into the robustness of the econometric model that sat behind every single sort of case study.
00:06:02.399 --> 00:06:10.480
Um and to educate yourself about how the model is built and the claims being made by the model in the first instance.
00:06:10.639 --> 00:06:20.160
I'm not sure that every marketer or CMO is sort of equipped with that knowledge when they um go into bat on behalf of sort of marketing budget in the boardroom.
00:06:20.560 --> 00:06:21.279
They definitely aren't.
00:06:21.439 --> 00:06:25.680
I mean, just to be clear, like I'm not a mathematician, I could not do that.
00:06:25.839 --> 00:06:37.439
Um and yeah, I do have to be the one that goes in and sells econometrics to my business, so I need to be able to explain what it is beyond just saying it's a black box that's going to give us all the answers.
00:06:37.600 --> 00:06:46.240
Um, and then I need to get to the other side on a presentation where there will be at least one person on your board who is a mathematician at heart, right?
00:06:46.319 --> 00:06:48.319
And will want to ask you lots of really tricky questions.
00:06:48.560 --> 00:06:49.759
You're like, I don't I don't really know.
00:06:49.839 --> 00:06:52.480
I've sort of got to the ends of my understanding.
00:06:52.639 --> 00:06:58.399
Um so that is really hard because marketers don't tend to be people who build models in their spare time.
00:06:58.720 --> 00:07:01.199
They're not that that kind of personality.
00:07:01.360 --> 00:07:18.399
So for me, one of the most important things has been getting the econometricians that I work with to get to know all of my board stakeholders along the way, not least because it builds belief in them as well, but also because it means they can answer some of the more tricky questions about what this is all about, right?
00:07:18.480 --> 00:07:29.120
So my CEO, for example, is one of those sales directors, turns CEO who is absolutely shithop, pardon my language, on numbers and modelling.
00:07:29.199 --> 00:07:31.680
He'll open up an Excel and build his own models himself.
00:07:31.839 --> 00:07:33.519
I'm never going to compete with that.
00:07:33.680 --> 00:07:41.040
So I needed Grace Kite, who is around somewhere, who just did the project with us last, to be in the room and be able to answer some of those tricky questions.
00:07:41.120 --> 00:07:44.480
Otherwise, it we the model wouldn't have had the belief in it among stakeholders.
00:07:44.879 --> 00:07:50.800
And do you think marketers are too protective of it at the moment, or is it just fear of having to present it?
00:07:51.120 --> 00:08:01.680
Uh I as I say, I think it's partly that marketers are um fearful of having to answer the tricky questions about how these numbers came out, other than just saying it's a black box.
00:08:01.920 --> 00:08:06.240
Um and partly also I think just looking at it too narrowly.
00:08:06.399 --> 00:08:18.800
I talked to quite a lot of marketers who do really go into these projects just thinking what I really want is the bar chart that shows me the ROI on marketing that I can then go off uh and tell everyone about how well it's going.
00:08:18.959 --> 00:08:28.160
And they kind of miss the fact that there's lots more useful and interesting stuff in there that, as I say, builds belief amongst the the broader stakeholder group.
00:08:28.480 --> 00:08:29.600
Yeah, I think I think that's right.
00:08:29.759 --> 00:08:36.000
I think we, you know, we're we're here at the Institute of Practitioners in advertising sort of conference.
00:08:36.080 --> 00:08:43.759
So we tend to sort of focus on advertising's contribution to sales, to uh to revenue, um, to profit.
00:08:43.919 --> 00:08:53.759
And you know, if you've ever looked at a sort of stacked bar chart and worked out that our contribution is like 4% of sort of total sales, then um sometimes you think, why do we bother?
00:08:53.919 --> 00:09:02.879
Uh but uh but you know, I think you forget about the interdependency with all the other sort of uh business drivers with within an organization.
00:09:02.960 --> 00:09:06.559
Um, and that's what we need to be sort of more cognizant of.
00:09:06.879 --> 00:09:10.399
Yeah, and when we heard you talk about it, is that inputs and outputs, isn't it?
00:09:10.480 --> 00:09:13.440
Like it's all it's all the elements that make that up.
00:09:13.600 --> 00:09:19.600
Um, so then how important is it to bring other departments into the conversation and leaders into this process?
00:09:20.159 --> 00:09:58.320
I mean, absolutely crucial from my side, particularly your CFO or your finance director, or frankly your whole finance team, and not just because they have to provide lots of the inputs typically, but but because good looks like you walking into the boardroom almost arm in arm with your CFO, with them going, This is my model too, and I believe in it, and I've got lots of things that you're going to learn from it, and that takes time and them being involved all the way through the process, and often actually, even if they can't be the ones in every single meeting, that the people on their team are the ones who've been really involved in it because they're going to be more effective messengers to your CFO than you are as the marketer, even though they're they're your peer.
00:09:58.399 --> 00:10:00.080
So I think that's really important.
00:10:00.240 --> 00:10:11.039
I think the other thing I'd say is like all good econometrics models are built on good hypotheses up front, and the marketer doesn't have a monopoly on the hypotheses of what's driving demand.
00:10:11.200 --> 00:10:20.159
And certainly in every model that I've ever been a part of, we've had these brilliant workshops at the start where everyone's chucking in their own hypotheses about what could be driving it.
00:10:20.240 --> 00:10:21.440
And there's been stuff that I've never heard.
00:10:21.519 --> 00:10:23.120
I'm like, oh, wow, that's really interesting.
00:10:23.200 --> 00:10:23.919
Where did that come from?
00:10:24.000 --> 00:10:27.039
And then it proves out in the model to be a really big, big driver.
00:10:27.200 --> 00:10:32.720
So again, it just builds buy-in and you get interesting stimulus that you wouldn't get on your own.
00:10:33.919 --> 00:10:35.840
And should you be doing that at the very beginning?
00:10:36.159 --> 00:10:37.440
I I I think so.
00:10:37.600 --> 00:10:47.120
I mean, it's kind of a bit like you're all agency folk as well, like when you have the kickoff at the start where clients get to chuck in their instincts and you politely go, Yeah, okay, sure.
00:10:47.360 --> 00:10:49.759
We're not going to be going anywhere near that, but thanks very much.
00:10:49.919 --> 00:10:55.600
Um there is a bit of that upfront that's just bringing people into the process.
00:10:55.679 --> 00:11:04.399
And as I say, you go in perhaps cynically about it, and then you find loads of really good stuff that you weren't expecting, and it enhances the process that then follows.
00:11:05.440 --> 00:11:05.919
Cool.
00:11:06.240 --> 00:11:18.080
And in this report, which you will all have a copy of, have you heard at the beginning, um, there's a lot of reporting uh around the differences AI can bring to econometrics, and particularly relating to speed and updating the data.
00:11:18.320 --> 00:11:18.480
Yeah.
00:11:18.720 --> 00:11:19.840
How have you found this?
00:11:20.399 --> 00:11:24.559
I think I mean this this was this was sort of brand new knowledge to me.
00:11:24.639 --> 00:11:42.559
You know, this I was entering into this world for the first time, and um there's a lovely sort of chapter at the end of this publication where uh Dom and Winnie from Cantar summarize the potential contribution of sort of AI to sort of econometric modelling, but they do stress that this is in its infancy.
00:11:42.639 --> 00:11:53.440
And I think um uh what I think what's interesting when you read this sort of chapter back, what there's so much um sort of humanity in that's in that sort of chapter.
00:11:53.519 --> 00:12:06.720
So what they're really talking about is you know, still establishing as a sort of human hypothesis for what you want um econometrics to sort of prove, and um establishing uh uh the right sort of data capability.
00:12:06.799 --> 00:12:15.279
So it's not like we're just sort of pressing a button and just handing over control of building models to um you know some weird sort of supercomputer somewhere.
00:12:15.440 --> 00:12:23.840
Um I think uh what what I love about this chapter in particular is that to date, um and the oh there's some people coming on to talk about AI in a minute, isn't there?
00:12:24.000 --> 00:12:26.559
So um I'm gonna sort of damn them uh right now.
00:12:26.720 --> 00:12:29.840
So I think AI's contribution to marketing.
00:12:30.080 --> 00:12:39.440
For some reason, we've we've deemed it, deemed that its role is to sort of build lots of stuff really, really quickly, be that sort of crap creative uh mostly.
00:12:39.600 --> 00:12:44.399
Whereas I think AI's greatest contribution could be providing foresight.
00:12:44.559 --> 00:13:11.600
I think the um uh I got to I got to spend some time with Dom talking about AI, and he told me this lovely sort of anecdote about a client um using so in instead of instead of econometrics being seen as a sort of presentation, you need to sort of think of it more as a mixing desk, which a sort of boardroom can sort of huddle around and sort of pull the levers and sort of start to sort of develop sort of better scenario planning capabilities, uh, which I I love that.
00:13:11.679 --> 00:13:12.080
I love that.
00:13:12.159 --> 00:13:17.200
So it's still it's not it's not that you're losing the rigour of building the model in the first place.
00:13:17.440 --> 00:13:24.240
AI is simply enabler to sort of play with some of the um uh some of the sort of levers within the model.
00:13:24.320 --> 00:13:27.200
And I thought I thought that was a really sort of uh nice development.
00:13:27.519 --> 00:13:32.480
So are we still in the experimental stage or we're actually seeing evidence of it now using AI?
00:13:32.879 --> 00:13:36.559
Uh well as I said, I mean this is completely new to me.
00:13:36.879 --> 00:13:58.639
Uh, but it's great that there are more sort of uh potential applications for um uh for more uh for using AI within econometrics, and there's more and more uh people like sort of Google and Meta who's who are sort of developing um sort of modelling capabilities as well, uh very much sort of focused on AI.
00:13:59.440 --> 00:13:59.759
Right.
00:13:59.919 --> 00:14:00.240
Okay.
00:14:00.720 --> 00:14:06.399
And then what kind of level of detail do boards want to see in these econometric reports?
00:14:06.480 --> 00:14:07.919
I mean, how much is too much?
00:14:09.039 --> 00:14:14.320
I mean, in my case, my board want to see more detail than they have the time to actually engage with, right?
00:14:14.399 --> 00:14:22.080
So left to their own devices, they would happily spend three hours um going through all of the detail, lapping it all up, being fascinated by it.
00:14:22.320 --> 00:14:39.039
Um probably not the best use of their time, but but um if you can draw out the key themes with enough detail that gives them confidence in the result, and you make sure those themes aren't just about the marketing bit you want to persuade them of, that's really effective.
00:14:39.200 --> 00:14:50.720
I think the other thing is picking off your stakeholders who you just know are fascinated by this stuff, whether it's whether it's their role to be or not, is just helpful because they will be the ones who are your biggest advocates in the room.
00:14:50.960 --> 00:15:05.279
So you know, I can think of examples where certainly at Wagamama, like I went and did the little tour to the the CEO and the CFO of the restaurant group, the parent company, because I knew that they were going to be really, really interested in this before we got into a proper board setting.
00:15:05.440 --> 00:15:13.600
And it just meant that when we did the presentation at board, I wasn't getting lots of tricky questions that were expanding the whole thing out um in the room.
00:15:13.759 --> 00:15:15.919
They were advocates on on behalf of it.
00:15:16.080 --> 00:15:18.159
So yeah, I think detail's a good thing.
00:15:18.240 --> 00:15:21.600
You just might not want to go through it all in the room in the presentation.
00:15:22.000 --> 00:15:23.679
No your audience, yeah, exactly.
00:15:23.919 --> 00:15:42.000
I think well, I think what the what um econometrics and the C suite does, it provide if you if you're a CMO and you and you're not working with uh an econometric model at present, this this sort of creates a sense of sort of professional envy because it's almost like a sales pitch for you know putting a sort of model in place.
00:15:42.159 --> 00:15:56.559
Um and you know, Ross uh works with sort of relatively sort of small advertising budgets compared to um a lot of brands, and yet it I think his is is great testimony to the sort of power of having an econometric model to help him make the sort of right decisions.
00:15:56.720 --> 00:16:07.360
And I think what when I when I was the technical judge last year, it was amazing just how uh I mean it was it was really beneficial sometimes.
00:16:07.679 --> 00:16:22.639
Sometimes the the sort of simplest models, the most sort of parsimonious models, were were the sort of best because they didn't allow for lots of assumptions, they were just trying to prove prove the contribution of sort of um three or four factors on sales of revenue or profitability.
00:16:22.799 --> 00:16:28.320
And I think it it gave you confidence that any business could actually sort of build their own sort of model.
00:16:28.639 --> 00:16:30.559
And it's also really, I mean, you're right.
00:16:30.639 --> 00:16:51.200
I I have that econometric models now have been part of those projects for smaller businesses relatively or smaller media spends, but it does frustrate me so much that there's this perception that econometrics is just for the big media spenders, that um, unless I'm spending 20 million plus a year and have insight budgets that are absolutely huge, econometrics isn't for me.
00:16:51.360 --> 00:17:02.639
Um, and I just find the almost the opposite is true, where for small businesses you're so lacking in real, genuine, rigorous insight that will tell you what the right choices to make are.
00:17:02.879 --> 00:17:09.680
And in the grand scheme of things, the cost of an econometric model is a lot less than I think people have in their heads it is typically.
00:17:09.920 --> 00:17:26.640
Like for both of the last two businesses I've worked for, they've been absolutely invaluable, and we've almost got more out of them because we're so motivated to eke out every last bit of insight out of them than the ones I've known for larger organizations that I've I've either been a part of or or worked with when I was agency side, you know.
00:17:26.880 --> 00:17:27.200
Right.
00:17:27.440 --> 00:17:32.079
So then how important is it to find the right external partner to work with?
00:17:32.240 --> 00:17:34.079
And um how how do you do that?
00:17:34.160 --> 00:17:40.079
What what what's your criteria for selection so that it doesn't look like you're making up your own homework or all those things we can often hear?
00:17:40.559 --> 00:17:42.160
I think that is the hardest bit, right?
00:17:42.240 --> 00:17:44.480
Because you're doing this primarily.
00:17:44.640 --> 00:17:45.920
Well, actually, I'll take that back.
00:17:46.240 --> 00:18:07.599
You're either doing it because you're a big media spender and econometrics are a really effective way of you working out how to optimize that, in which case I actually think there is quite a good rationale for why you might work with some of the econometricians within the media agency you're working with, because they're much closer to the action and the things that you might do with that model make sense.
00:18:07.680 --> 00:18:08.799
But exactly what you say, Karen.
00:18:09.119 --> 00:18:09.920
It's a downside.
00:18:10.000 --> 00:18:20.079
I think that board stakeholders are more likely to think, oh, what a shock, media agency declares media works really well, uh, and your whole thing is is undermined.
00:18:20.240 --> 00:18:40.240
Whereas I think when you when your objective is more a I want to understand the broader drivers of demand of which marketing investment is part of it, there is a bit more merit of having separation of powers here and having someone who is independent of any of your agencies, um, or effectively your organization as well.
00:18:40.319 --> 00:18:48.400
Some businesses have in-house econometricians as well, but like someone independent who will come and provide fresh eyes and go, no, I'm going to adjudicate here.
00:18:48.559 --> 00:19:04.480
So because of the businesses I've worked in, I I tend to work with econometricians who are um independent of any of those stakeholders, and I've found that's been helpful, but I I totally appreciate if I was in a global multinational, I might take a different view because my spend is much bigger.
00:19:06.480 --> 00:19:06.720
Okay.
00:19:08.640 --> 00:19:22.000
Yeah, I mean, just to add to that, I think there's I think there's certainly over my career, um, it was uh there was a fierce rivalry over sort of over sort of building econometric models about sort of 15 years ago.
00:19:22.160 --> 00:19:25.200
It was just a better margin business than sort of media as a whole.
00:19:25.440 --> 00:19:32.559
Um but uh but luckily I think you know we're that sort of rivalry, rivalry has sort of calmed down a bit.
00:19:32.640 --> 00:19:42.079
So my own lived experience um sort of recently working with IKEA, so Olgra, I'm not sure if you see it, Olgra at MediaCom runs this sort of econometric model for IKEA.
00:19:42.400 --> 00:19:52.799
And while when I was at Dentu at iProspect, I was sort of uh you know doing the sort of media buying and sort of taking the findings of the econometric model and applying them to uh to the advertising budget.
00:19:53.039 --> 00:19:58.960
So I think there's there's sort of less rivalry over ownership of the data than there's than there's sort of been previously.
00:19:59.440 --> 00:20:11.039
Uh I I I I pr I completely understand the sort of um the concern about if you're if you're an advertiser then trusting the agency to sort of run their own sort of model.
00:20:11.279 --> 00:20:12.559
I completely understand that.
00:20:13.440 --> 00:20:13.759
Okay.
00:20:14.160 --> 00:20:23.359
Um if there's anyone in the audience here uh who wants to start this type of a journey and uh you know uh how do they know if they will have enough of the right kind of data?
00:20:23.519 --> 00:20:24.880
What do you recommend?
00:20:25.680 --> 00:20:28.960
And how long would it take before you actually see return on it?
00:20:29.119 --> 00:20:30.480
Because we're all about return.
00:20:30.799 --> 00:20:39.200
I think I think the well I think going through this process, what you realise is that there's a dearth of good econometricians out there.
00:20:39.359 --> 00:20:41.039
It's like finding a builder, basically.
00:20:41.279 --> 00:20:48.559
So I think a lot of the delay is almost you know finding the right practitioner who can actually sort of help you um straight away.
00:20:48.799 --> 00:20:53.279
Um I'd love I think I should retrain as an econom uh econometrician basically.
00:20:53.519 --> 00:20:58.559
But um I think once the model is built, you know, the I mean Ross, you're very familiar with this.
00:20:58.640 --> 00:21:03.680
The you know the findings are sort of they come on board sort of quite quickly and the things you can apply quite quickly.
00:21:04.079 --> 00:21:07.519
But um I don't know about I don't know about the sort of the sort of data taxonomy.
00:21:07.680 --> 00:21:13.680
I think I think they can start a lot simpler than you actually sort of give them credit for and then sort of build over time.
00:21:14.079 --> 00:21:15.519
Yeah, that was my experience, right?
00:21:15.680 --> 00:21:22.720
So it answered the question if you're if you're thinking about it but you're not sure you've got enough data, the the answer is to ask, right?
00:21:22.880 --> 00:21:36.480
So I uh when I first started thinking about econometrics at Little Moons, it was just after I'd sold into the boards the idea that it was about time in our journey for us to start investing in paid media.
00:21:36.559 --> 00:21:50.640
So it was well before we actually made an ad, bought any media, and I was already thinking about how am I going to measure this because if we do it once and it's perceived that it doesn't work, then I'm scuppered, right?
00:21:50.720 --> 00:21:53.519
So so we have to measure this, and you think about it.
00:21:53.599 --> 00:21:59.759
But I also had many of the limiting beliefs that I'm sure many people um have experienced of going, oh God, we're not buying enough.
00:22:00.480 --> 00:22:08.480
Epos data and and our maybe our finance um capability isn't quite what it would be in um in much larger organizations.
00:22:08.640 --> 00:22:27.359
Um and I vividly remembered like I I went for a coffee with Grace because I knew Grace because I'd worked with her on um on the Wacamama model, and I just kind of said, Would you do me a favor and spend half an hour just talking to me about um when would be the right time for us to do this kind of modelling and what kind of data would we need uh in order to do it?
00:22:27.440 --> 00:22:41.440
And I walked away from that conversation going, oh, actually, we'll probably be at the point where we'll be buying that EPOS data that we need by the time that the media goes out on air anyway, and I can buy it um backdated anyway, so I'll have it.
00:22:41.519 --> 00:22:45.119
And then, you know, we're talking about the costs involved, now it's probably manageable as well.
00:22:45.200 --> 00:22:58.400
So you come away from it feeling really reassured um that actually the answer is probably you can do it earlier than you think you can, and you probably should do it earlier than um than than you think you can.
00:22:58.559 --> 00:23:02.559
So um, so yeah, absolutely asking is the answer.
00:23:02.640 --> 00:23:16.400
And and David's point, like finding a practitioner who you've worked with in the past or who you trust, or or whoever is the invaluable thing there, because then you've got a sounding board who can just take away some of those beliefs and and make it concrete for you.
00:23:17.359 --> 00:23:29.920
Yeah, I think I think the you know, uh any sort of an objective of econometrics is just to sort of turn a hypothesis or a qualitative belief into a quantitative statement.
00:23:30.079 --> 00:23:36.480
So instead of just saying, oh, I think if we spend 500 grand on advertising this year, I think we'll be in a better place than we were, you know, yesterday.
00:23:36.720 --> 00:23:42.000
It's to actually understand that for every one pound you spend, you're gonna get one pound sort of 20 returns.
00:23:42.079 --> 00:23:49.759
So you've got to go and find the simple data sets which are gonna sort of go and prove that hypothesis and and and believe it to be true.
00:23:49.839 --> 00:23:51.119
So I think it is better.
00:23:51.519 --> 00:23:53.920
It's simpler than a lot of people make out, isn't it?
00:23:54.240 --> 00:24:02.559
Just that I don't know, I think I think we've sort of dressed it up to be something scary because the best models can do so much.
00:24:04.240 --> 00:24:06.799
And in editing this, then what was the most surprising thing?
00:24:06.880 --> 00:24:10.799
I mean, obviously we've talked about the AIPs, but what else did you think was illuminating?
00:24:11.119 --> 00:24:24.319
I think I think look, if you're if if you sat here now and you're an agency practitioner um and you're like one of those people who like to shut shoot their mouth off in meetings about um you know, like you know how agency people do that?
00:24:24.480 --> 00:24:28.720
Say, oh maybe we should change your product or maybe we should sort of change your distribution network or things like that.
00:24:28.960 --> 00:24:30.000
Just with no idea.
00:24:41.039 --> 00:24:42.960
So I think that was one of the sort of keys findings.
00:24:43.200 --> 00:24:44.000
It's all there.
00:24:44.160 --> 00:24:48.079
The models are fantastic and they can show you the contribution of everything.
00:24:48.240 --> 00:24:51.599
So before you go crazy, just yeah, do your homework basically.
00:24:51.839 --> 00:24:52.880
It's also quite humbling.
00:24:52.960 --> 00:24:59.839
I think when I was at agency side, and I'd see that and I'd be like, to your point, advertising is contributing three, four percent of your base.
00:24:59.920 --> 00:25:11.839
And it does give you a bit of an understanding of why clients uh tend to be kept up at night by things other than the media bits, because it can go very, very wrong with just a couple of other things going um going a bit off, you know.
00:25:12.240 --> 00:25:12.799
Brilliant.
00:25:13.039 --> 00:25:13.279
Okay.
00:25:13.599 --> 00:25:15.920
Well, I think that has been incredibly helpful.
00:25:16.000 --> 00:25:24.000
And uh make sure you read your copy and you'll all be experts at the at the end of it, and you won't make yourself look like a fool of the meeting, which is always a good thing.
00:25:24.400 --> 00:25:24.880
Objective done.
00:25:25.440 --> 00:25:25.839
Amazing.
00:25:26.000 --> 00:25:26.880
Thank you both very much.
00:25:28.160 --> 00:25:28.559
Thank you.
