สำเนาบทสนทนา 🔗
WEBVTT
00:00:01.199 --> 00:00:03.120
Hey, it's Will Scott's VoiceClone.
00:00:03.200 --> 00:00:04.799
The content is curated and reviewed by Will.
00:00:04.879 --> 00:00:05.919
The voice is AI.
00:00:06.000 --> 00:00:09.519
This is the Search Influence weekly briefing for November 10th, 2026.
00:00:09.679 --> 00:00:13.439
This week, AI platforms moved a little farther from giving an answer toward taking an action.
00:00:13.599 --> 00:00:16.399
Google added home services booking links inside AI mode.
00:00:16.559 --> 00:00:19.519
Microsoft Clarity is showing what supported agents do on websites.
00:00:19.679 --> 00:00:26.000
On the advertising side, ChatGPT ads is moving advertisers to average cost targets, and Meta is changing how some conversions are matched.
00:00:26.079 --> 00:00:27.120
The lesson for us is simple.
00:00:27.359 --> 00:00:31.039
Give these systems accurate services, useful content, and clean conversion signals.
00:00:31.120 --> 00:00:35.039
Then compare the platform's report with appointments, orders, qualified leads, and revenue.
00:00:35.200 --> 00:00:36.719
A new service can create an opportunity.
00:00:36.880 --> 00:00:40.000
A new measurement method can also make the dashboard move before the business does.
00:00:40.159 --> 00:00:42.799
The written briefing includes an expert panel review.
00:00:42.960 --> 00:00:45.439
We use the expert panel to put the week's news in context.
00:00:45.600 --> 00:00:48.320
Announcements, reporting, and original data establish the facts.
00:00:48.479 --> 00:00:52.159
The expert views add interpretation and point out what those sources do not prove.
00:00:52.399 --> 00:00:53.520
Seven items made the cut.
00:00:53.679 --> 00:00:56.960
Google added direct home services booking links in USAI mode results.
00:00:57.039 --> 00:01:04.159
Microsoft Clarity announced agent activity, task simulations, brand agents, and a conversational analytics feature, each with its own availability rules.
00:01:04.319 --> 00:01:10.719
Search Engine Roundtable reported an unfinished OpenAI analytics screen that may show site owners their ChatGPT mentions and clicks.
00:01:10.879 --> 00:01:14.319
A small YouTube experiment linked richer descriptions with more AI retrievals.
00:01:14.480 --> 00:01:18.159
ChatGPT ads move new campaigns to Target CPA and Target CPC.
00:01:18.239 --> 00:01:24.480
Meta introduced on-device conversion matching for eligible pixels, and Meta paused the terms changes that had been scheduled for November 20th.
00:01:24.640 --> 00:01:26.560
Start with Google's home services booking links.
00:01:26.640 --> 00:01:30.079
Barry Schwartz at Search Engine Roundtable reported Google's November 4th announcement.
00:01:30.239 --> 00:01:36.079
US users can now move from research and AI mode to direct booking links through Angie, Thumbtack, HouseCall Pro, and Jobber.
00:01:36.239 --> 00:01:37.840
Service Titan is described as coming soon.
00:01:38.000 --> 00:01:40.879
Google Product Lead Robbie Stein's example is a plumbing emergency.
00:01:41.040 --> 00:01:45.359
AI Mode compares local pricing, checks reviews, makes a recommendation, and helps him book a plumber.
00:01:45.439 --> 00:01:47.920
It is Google's own example, not lead performance data.
00:01:48.079 --> 00:01:51.120
For a local service business, the service page is only part of the setup.
00:01:51.200 --> 00:01:57.120
The system also needs a clear service offering, reliable location information, current availability, and a booking path that works.
00:01:57.280 --> 00:02:01.920
If a customer can find the business but cannot schedule the right service, the new service has not done its job.
00:02:02.000 --> 00:02:05.200
So check the provider listings and booking tools clients already use.
00:02:05.359 --> 00:02:07.840
Search their priority services in the markets they serve.
00:02:08.000 --> 00:02:12.479
Confirm that the result describes the right service and that the booking link lands on a useful next step.
00:02:12.639 --> 00:02:16.800
Then add a lead source field so the team can tell an AI mode booking apart from another local lead.
00:02:16.960 --> 00:02:18.319
Do not call this a lead channel yet.
00:02:18.400 --> 00:02:20.879
We have no booking volume, lead quality, or revenue data.
00:02:21.039 --> 00:02:23.199
Next up, Microsoft Clarity's Agent Update.
00:02:23.360 --> 00:02:29.120
Microsoft is trying to connect three journeys people browsing, people asking an AI for a recommendation, and agents acting for them.
00:02:29.280 --> 00:02:32.639
Agent Activity shows which supported agents visit a site and what they do there.
00:02:32.800 --> 00:02:37.199
Microsoft says supported operators currently include Google, OpenAI, and Anthropic.
00:02:37.439 --> 00:02:41.840
The dashboard can show searches, form submissions, logins, downloads, appointment bookings, and checkout.
00:02:42.000 --> 00:02:43.840
Clarity also adds simulation assessments.
00:02:43.919 --> 00:02:48.080
A team can test whether an agent completes an important task, how long it takes, and where it gets stuck.
00:02:48.240 --> 00:02:55.120
Brand agents, a brand's own AI assistant, are expanding into Microsoft Discovery Serfaces such as Copilot and Bing and into local service booking.
00:02:55.199 --> 00:02:57.599
Eligibility depends on the platform and the type of business.
00:02:57.759 --> 00:03:00.319
Ask Clarity is the conversational analytics piece.
00:03:00.479 --> 00:03:04.719
You ask a question about site behavior, and Clarity links its answer back to session recordings.
00:03:04.879 --> 00:03:09.599
That feature is a closed beta for retailers, so do not treat the whole announcement as generally available.
00:03:09.759 --> 00:03:18.159
Microsoft reports that 14 of 16 early access participants said they discovered something new, that is useful product feedback from a tiny company-run survey, not an independent study.
00:03:18.319 --> 00:03:20.639
For clients, pick three tasks that drive revenue.
00:03:20.800 -->
Load full transcript
00:03:23.280
Booking, a lead form, and checkout are good examples.
00:03:23.439 --> 00:03:29.599
Test whether an agent can finish each one, review the recordings, and confirm which features the account actually has before promising anything.
00:03:29.759 --> 00:03:33.039
Staying with measurement, the third story is a possible ChatGPT analytics view.
00:03:33.199 --> 00:03:35.759
Treat this one as a report about unfinished work, not a launch.
00:03:35.919 --> 00:03:41.360
Search engine roundtable reports on an unfinished OpenAI analytics screen spotted by Josez Kazu Kenas.
00:03:41.439 --> 00:03:47.199
It appears to let site owners verify that they own a site, then see how often ChatGPT mentions them and how often it sends clicks.
00:03:47.360 --> 00:03:51.280
The report calls it ChatGPT Webmaster Tools and compares it to Google Search Console.
00:03:51.439 --> 00:03:52.960
OpenAI has not announced a product.
00:03:53.039 --> 00:03:54.960
There is no official name and no release date.
00:03:55.120 --> 00:03:59.439
The report also shows no prompt data, user data, competitor comparison, or ranking view.
00:03:59.520 --> 00:04:00.639
It still matters for planning.
00:04:00.719 --> 00:04:03.520
Today it is hard to see how ChatGPT sends traffic to a site.
00:04:03.599 --> 00:04:08.080
If something like this ships, agencies would get that data straight from OpenAI instead of piecing it together.
00:04:08.240 --> 00:04:09.199
So start the baseline now.
00:04:09.360 --> 00:04:13.199
Track ChatGPT referrals and mentions with the analytics and log data we already collect.
00:04:13.360 --> 00:04:15.919
If a tool does launch, we will have something to compare it against.
00:04:16.079 --> 00:04:19.120
And do not promise clients a dashboard OpenAI has not confirmed.
00:04:19.199 --> 00:04:21.279
The fourth story is the YouTube description experiment.
00:04:21.519 --> 00:04:24.000
Peace AI looked at about 60 YouTube videos.
00:04:24.160 --> 00:04:28.319
On October 19th, researchers updated descriptions and added timestamps to 15 of them.
00:04:28.480 --> 00:04:33.279
They compared AI retrievals, meaning how often AI tools pulled those videos before and after the change.
00:04:33.519 --> 00:04:38.720
Search Engine Land reports a 359% increase in daily AI retrievals for the updated videos.
00:04:38.879 --> 00:04:40.399
Unchanged videos rose 17%.
00:04:40.720 --> 00:04:44.720
The test measured retrievals in tools including Google AI overviews and AI mode.
00:04:44.879 --> 00:04:47.759
The researchers could not prove the description changes caused the rise.
00:04:47.920 --> 00:04:50.959
The test was small, short, vendor-run, and not randomized.
00:04:51.040 --> 00:04:55.120
A retrieval also does not tell us whether a person watched the video, became a lead, or bought something.
00:04:55.279 --> 00:04:56.879
Still, it supports a reasonable test.
00:04:57.040 --> 00:05:01.040
Descriptions and timestamps tell an AI tool what a video covers and where each part starts.
00:05:01.199 --> 00:05:08.879
Refresh a set of useful older videos with accurate summaries and timestamps, leave a similar group unchanged, and measure retrievals, views, site visits, and conversions separately.
00:05:09.040 --> 00:05:10.879
OpenAI shows up again on the paid side.
00:05:11.040 --> 00:05:13.360
The fifth story is ChatGPT adds target bidding.
00:05:13.439 --> 00:05:18.240
Starting November 9th, Target CPA replaces max bid for new conversion campaigns that use a cost target.
00:05:18.399 --> 00:05:20.959
Target CPC is available for new click campaigns.
00:05:21.120 --> 00:05:28.079
Existing campaigns can keep max bid or max CPC for now, but OpenAI says those controls retire January 1st, 2027.
00:05:28.240 --> 00:05:33.759
A target CPA or target CPC is an average cost goal, not a maximum bid, and not a guaranteed price.
00:05:33.920 --> 00:05:38.160
The system can bid above or below the target in any one auction while it aims for the average.
00:05:38.319 --> 00:05:41.680
OpenAI also warns that a target set too low can reduce delivery.
00:05:41.759 --> 00:05:45.600
The same can happen when there are few chances to show or conversion signals are weak.
00:05:45.759 --> 00:05:49.120
A campaign can leave budget unspent even when the advertiser wants more volume.
00:05:49.279 --> 00:05:55.759
For conversion campaigns, confirm that the chosen event is actually arriving from the OpenAI pixel, the conversions API, or both.
00:05:56.000 --> 00:06:02.240
Setup alone does not prove usable signals are reaching the campaign, then set the target from recent actual cost over a consistent period.
00:06:02.399 --> 00:06:07.439
If the recent cost per qualified lead is$80, typing in$30 does not make the economics true.
00:06:07.600 --> 00:06:12.319
Move one campaign at a time and compare spend, delivery, lead quality, and actual average cost.
00:06:12.480 --> 00:06:14.959
There is a separate API change starting November 18th.
00:06:15.120 --> 00:06:20.480
Partners need to move to the new target CPA and target CPC fields before the old fields retire on January 1st.
00:06:20.639 --> 00:06:23.040
The sixth story is Meta's on-device conversion matching.
00:06:23.199 --> 00:06:28.959
Meta says its pixel can now connect an ad view with a later purchase on the customer's own device, then report the conversion back to the campaign.
00:06:29.040 --> 00:06:31.279
The setting is on by default for eligible pixels.
00:06:31.360 --> 00:06:34.800
It works alongside the Pixel and Conversions API and does not replace either one.
00:06:34.959 --> 00:06:38.480
Meta also says the attribution window and attribution model do not change.
00:06:38.560 --> 00:06:41.920
The setting changes which conversions can be measured, not how they are credited.
00:06:42.079 --> 00:06:45.759
Meta says the feature does not apply to people in the European Union and United Kingdom.
00:06:45.920 --> 00:06:48.000
Meta's public help page does not show a start date.
00:06:48.079 --> 00:06:52.000
MeasureBase reports that Events Manager shows November 12th for eligible pixels.
00:06:52.160 --> 00:06:55.439
That date matters because the report can improve without the business improving.
00:06:55.600 --> 00:07:00.720
More purchases may be matched to campaigns after November 12th, even if the store records the same number of orders.
00:07:00.879 --> 00:07:05.519
Reported cost per purchase can fall, and return on ad spend can rise just because the measurement changed.
00:07:05.680 --> 00:07:10.959
Before November 12th, save a baseline, campaign purchases, value, cost per purchase, and return on ad spend.
00:07:11.040 --> 00:07:13.839
Put orders and revenue from the store or CRM beside those figures.
00:07:14.000 --> 00:07:17.040
Note the date in reporting and tell the people who read the dashboard what changed.
00:07:17.199 --> 00:07:23.199
Meta's documentation leaves open questions about duplicate counting, purchase value, consent, and whether the extra matches feed bidding.
00:07:23.360 --> 00:07:26.560
Until those answers arrive, treat a jump as a measurement change first.
00:07:26.720 --> 00:07:28.879
The last story is Meta's pause terms update.
00:07:29.040 --> 00:07:37.519
Meta says proposed updates to its commercial terms, product catalog terms, self-serve ad terms, and ad creative generative AI terms will not take effect on November 20th.
00:07:37.680 --> 00:07:38.959
Existing terms remain in force.
00:07:39.120 --> 00:07:42.399
Neiman Lab reports that Meta confirmed the delay has no end date.
00:07:42.560 --> 00:07:47.360
The proposed terms had drawn objections over how Meta could use content from websites connected to MetaBusiness accounts.
00:07:47.519 --> 00:07:48.399
The deadline is gone.
00:07:48.560 --> 00:07:49.519
The issue is not settled.
00:07:49.680 --> 00:07:54.240
Meta has not published new wording or a new date, and it says its crawling practices are unchanged.
00:07:54.399 --> 00:07:56.399
A pause does not mean Meta dropped the idea.
00:07:56.560 --> 00:08:00.480
For publishers and meta business users, keep the current assessment and any robots.
00:08:01.040 --> 00:08:05.040
XT decisions, meaning the file that tells crawlers what they may access As they are.
00:08:05.199 --> 00:08:06.720
Do not act on a pause draft.
00:08:06.800 --> 00:08:10.480
When new text appears, compare it with the current terms and document what actually changed.
00:08:10.639 --> 00:08:12.639
Keep these six lines handy for client calls.
00:08:12.800 --> 00:08:15.279
For local service clients, Google can now send some U.S.
00:08:15.360 --> 00:08:18.240
Home Services customers from AI mode to a booking partner.
00:08:18.399 --> 00:08:22.639
We'll check that your services, availability, and lead tracking are ready before we call it a lead channel.
00:08:22.800 --> 00:08:27.759
For analytics conversations, Microsoft Clarity is starting to show what some AI agents do on your site.
00:08:27.920 --> 00:08:31.839
We'll test the tasks that bring in leads or sales and confirm which features your account has.
00:08:32.000 --> 00:08:36.879
For ChatGPT advertisers, ChatGPT ads is moving from maximum bids to average cost targets.
00:08:37.039 --> 00:08:41.200
We'll set the target from recent results and clean conversion data, then watch delivery and lead quality.
00:08:41.279 --> 00:08:46.960
For meta advertisers, Meta may report more conversions after November 12th because it added a new way to match them to ads.
00:08:47.039 --> 00:08:51.039
We'll compare Meta's numbers with orders and CRM revenue before calling it better performance.
00:08:51.120 --> 00:08:55.440
For publishers, Meta paused the November 20th terms update, so the current terms still apply.
00:08:55.519 --> 00:08:58.000
We'll wait for new wording instead of acting on a pause draft.
00:08:58.080 --> 00:09:04.399
For video teams, a small YouTube test suggests better descriptions may help AI tools find videos, but it did not prove the edits caused the lift.
00:09:04.559 --> 00:09:06.720
We'll run a controlled test and measure business results.
00:09:06.879 --> 00:09:08.320
That's the briefing for November 10th.
00:09:08.480 --> 00:09:13.120
The written edition includes the full expert panel review, source links, scope, and caveats.
00:09:13.360 --> 00:09:19.200
This podcast was generated with an AI clone of Will Scott's voice using the Eleven Labs text to speech platform to keep influencing.