TRANSCRIPT 🔗
WEBVTT
00:00:01.280 --> 00:00:03.120
Hey, it's Will Scott's VoiceClone.
00:00:03.200 --> 00:00:05.280
The content is curated and reviewed by Will.
00:00:05.360 --> 00:00:06.559
The voice is AI.
00:00:06.719 --> 00:00:10.800
Welcome to the Search Influence weekly briefing for Tuesday, September 29th, 2026.
00:00:10.960 --> 00:00:17.039
This week, search and advertising platforms gave teams more data, more matching inputs, more controls, and more markets to reach.
00:00:17.199 --> 00:00:19.920
The practical job is smaller than the announcements make it sound.
00:00:20.079 --> 00:00:25.519
Use each feature to improve one decision and document its limits before a new chart or audience becomes a client promise.
00:00:25.760 --> 00:00:27.280
Six items made the cut this week.
00:00:27.440 --> 00:00:30.079
Google started a global spam update that may take up to two weeks.
00:00:30.239 --> 00:00:34.479
Search Console added a separate view for web searches that begin with an image, photo, or screenshot.
00:00:34.640 --> 00:00:38.960
Google Customer Match added IP addresses and interaction timestamps as matching inputs.
00:00:39.039 --> 00:00:46.000
Microsoft added search to landing page reporting, device exclusions for performance max, and more LinkedIn targeting support in its advertising API.
00:00:46.159 --> 00:00:55.119
ChatGPT ads expanded into seven more Asian markets and is now available in more than 60 countries, and Meta announced an enterprise AI platform led by CJ Desai.
00:00:55.280 --> 00:01:01.840
The expert panel review adds six interpretations from John Muller, Dave Smart, Anua Degbola, Pauline Jacober, and Larry Dignan.
00:01:01.920 --> 00:01:03.439
Their comments help us decide what to do.
00:01:03.600 --> 00:01:09.519
Google's records, Microsoft Documentation, OpenAI, and Meta announcements, and the dated reporting still establish the facts.
00:01:09.680 --> 00:01:11.280
Start with the September spam update.
00:01:11.439 --> 00:01:15.200
Google's search status dashboard records the update beginning on September 24th.
00:01:15.280 --> 00:01:19.439
Google said it applies globally, across all languages, and may take up to two weeks to complete.
00:01:19.599 --> 00:01:25.200
The company did not say that the rollout would necessarily use the full two weeks, and it did not name one spam tactic as the target.
00:01:25.280 --> 00:01:26.959
That leaves teams with a familiar problem.
00:01:27.200 --> 00:01:29.840
Rankings and traffic may move while the system is still changing.
00:01:29.920 --> 00:01:34.239
A quick edit can make the diagnosis harder because we lose the clean before and after comparison.
00:01:34.400 --> 00:01:37.040
This week also brought a new Search Console reporting category.
00:01:37.200 --> 00:01:44.640
If a client's web impressions move, the cause could be the spam update, the new category entering the report, an unrelated site change, or some combination of those things.
00:01:44.799 --> 00:01:46.719
Annotate September 24th
Load full transcript
in the reporting.
00:01:46.799 --> 00:01:51.680
If performance changes, identify the pages, queries, countries, devices, and conversions that moved.
00:01:51.840 --> 00:01:55.200
Check whether the change is concentrated in one template, content type, or market.
00:01:55.439 --> 00:01:59.519
Review Google's spam policies when the evidence points to a real site level loss.
00:01:59.680 --> 00:02:05.040
Avoid a broad cleanup while the rollout is active unless there is an obvious policy problem that should have been fixed anyway.
00:02:05.200 --> 00:02:11.439
Also avoid declaring a recovery after one good day, wait for the rollout to finish, then compare a stable period with the right baseline.
00:02:11.599 --> 00:02:13.680
The second SEO story is multimodal reporting.
00:02:13.840 --> 00:02:18.960
Google added a web multimodal filter to the search results performance report and the generative AI performance report.
00:02:19.039 --> 00:02:24.560
It covers lens, circle to search on Android, images uploaded to Google Search, and the search this image option in Chrome.
00:02:24.719 --> 00:02:27.439
For an image-heavy business, this fills a real reporting gap.
00:02:27.599 --> 00:02:31.360
A retailer can see which product pages appeared after a person photographed an object.
00:02:31.439 --> 00:02:35.280
A destination can see which pages surfaced after someone pointed a camera at a place.
00:02:35.439 --> 00:02:40.639
Food, home, travel, and other visual categories may find useful page level demand that was invisible before.
00:02:40.800 --> 00:02:41.919
There is a large limitation.
00:02:42.080 --> 00:02:44.719
Search Console does not provide the query dimension for this traffic.
00:02:44.800 --> 00:02:49.199
The report can show the page, country, device, date, impressions, and other available dimensions.
00:02:49.280 --> 00:02:52.000
It cannot show what the camera saw or what the person intended.
00:02:52.159 --> 00:02:57.840
John Mueller said after checking with the Search Console team that the multimodal data had not previously been included in the counts.
00:02:57.919 --> 00:02:59.919
That makes the rollout a possible reporting break.
00:03:00.080 --> 00:03:04.719
Dave Smart also reported declines in text-based impressions and clicks on a few properties around the split.
00:03:04.879 --> 00:03:07.919
His sample was small, and Google's explanation does not settle the cause.
00:03:08.080 --> 00:03:11.840
Export the new view and keep multimodal and text-based trends separate for now.
00:03:12.000 --> 00:03:17.439
Compare the pages that surface with their primary images, product information, alt text, captions, and structured data.
00:03:17.599 --> 00:03:19.840
Do not rewrite pages based on an unknown query.
00:03:19.919 --> 00:03:24.639
Use the report to identify which pages deserve inspection, then test improvements against traffic and conversions.
00:03:24.719 --> 00:03:26.159
Add a dated note to dashboards.
00:03:26.240 --> 00:03:31.840
If the top-level web total rises because the report now includes new data, do not present that as an organic growth win.
00:03:31.919 --> 00:03:36.400
If text-based data falls during the same period, do not blame the new split until the comparison is stable.
00:03:36.560 --> 00:03:41.120
For a first review, sort the multimodal export by impressions, then look at the top 20 pages.
00:03:41.280 --> 00:03:46.639
Ask whether the main image clearly shows the product, place, meal, room, or object the page is meant to describe.
00:03:46.800 --> 00:03:51.439
Check whether important details appear only inside the image, and whether the page carries matching text.
00:03:51.599 --> 00:03:56.319
For commerce pages, confirm price, availability, identifiers, variants, and shipping information.
00:03:56.479 --> 00:04:01.120
For destination and local pages, confirm the place name, address, landmarks, and useful context.
00:04:01.199 --> 00:04:05.439
Save that baseline before changing anything so the next export can show whether the work helped.
00:04:05.520 --> 00:04:06.560
Now to customer match.
00:04:06.639 --> 00:04:12.479
Google Ads documentation now allows customer match files to include a user's IP address and interaction timestamp.
00:04:12.639 --> 00:04:15.680
Those fields must be passed as plain unhashed values.
00:04:15.840 --> 00:04:21.439
Google says IP matching is not supported for end users in the European Economic Area, the United Kingdom, or Switzerland.
00:04:21.519 --> 00:04:26.000
A timestamp also cannot be sent by itself without an IP address or other identifying data.
00:04:26.079 --> 00:04:27.120
The appeal is straightforward.
00:04:27.279 --> 00:04:30.560
Another matching signal may make more of an advertiser's first party list usable.
00:04:30.639 --> 00:04:36.800
The material reviewed this week does not show a match rate increase or a campaign performance result, so the benefit is still a test, not a fact.
00:04:37.040 --> 00:04:39.279
Anuadigbola called out the operational catch.
00:04:39.439 --> 00:04:43.519
Consent, disclosure, regional exclusions, and data handling belong in the implementation plan.
00:04:43.680 --> 00:04:46.240
This is not a box to check in the campaign interface and forget.
00:04:46.399 --> 00:04:47.839
Before testing, map the data flow.
00:04:48.000 --> 00:04:55.279
Record where the IP address comes from, what event creates the timestamp, which users must be excluded, who can export the file, where it is stored, and when it is deleted.
00:04:55.439 --> 00:04:58.399
Confirm that the stated purpose and disclosure cover the transfer to Google.
00:04:58.480 --> 00:04:59.600
Use a bounded list first.
00:04:59.680 --> 00:05:03.920
Measure the change in match rate, then measure qualified leads, appointments, sales, or revenue.
00:05:04.079 --> 00:05:06.560
A larger audience is not automatically a better audience.
00:05:06.720 --> 00:05:10.240
If the team cannot explain the source and handling of every field, pause the upload.
00:05:10.319 --> 00:05:13.680
Run the test through the same governance used for other customer data connections.
00:05:13.839 --> 00:05:17.920
Give the file an owner, record the exact columns, restrict access, and set a deletion date.
00:05:18.000 --> 00:05:23.199
Ask the privacy or legal owner to review the plan before the first upload, not after a client asks where the data went.
00:05:23.360 --> 00:05:28.720
Keep the result in a test log with the list size, match rate, campaign dates, geographic scope, and downstream outcome.
00:05:28.879 --> 00:05:34.639
That record makes the next decision easier and prevents a temporary experiment from becoming a permanent feed nobody remembers approving.
00:05:34.800 --> 00:05:36.800
Microsoft also added reporting and controls.
00:05:36.959 --> 00:05:40.079
Microsoft's official release notes list a search term landing page report.
00:05:40.240 --> 00:05:46.480
Search engine lands dated report says teams can bring the search term, delivered headline, final URL, and performance into one view.
00:05:46.639 --> 00:05:51.839
The update also includes campaign-level device exclusions for Performance Max and additional LinkedIn segment targeting support.
00:05:51.920 --> 00:05:54.639
The search to page report may be the most immediately useful piece.
00:05:54.720 --> 00:06:02.399
It can reveal a search that received the wrong message, a good headline that sent the click to a weak page, or a landing page selected by automation that does not fit the person's intent.
00:06:02.560 --> 00:06:05.279
Device exclusions add a business rule to an automated campaign.
00:06:05.439 --> 00:06:09.199
That can help when a product or conversion flow performs badly on a device for a known reason.
00:06:09.279 --> 00:06:13.759
It can also hide a broken mobile experience if the team excludes the device instead of fixing the page.
00:06:13.920 --> 00:06:18.000
Add a recurring audit that reads across the search, headline, destination, and qualified result.
00:06:18.160 --> 00:06:20.639
Flag mismatches, then test the smallest useful change.
00:06:20.800 --> 00:06:25.120
Use exclusions when the business case is clear, and record the reason so someone can revisit it later.
00:06:25.199 --> 00:06:26.240
Keep the timeline accurate.
00:06:26.399 --> 00:06:31.759
Microsoft's release notes place the relevant interface editions in August, while the trade report was published on September 25th.
00:06:31.839 --> 00:06:36.160
The September article is current coverage of the capability not proof that every component launched that day.
00:06:36.319 --> 00:06:38.079
ChatGPT Ads is expanding again.
00:06:38.240 --> 00:06:45.199
OpenAI began rolling the product into Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam, and Taiwan on September 23rd.
00:06:45.279 --> 00:06:48.879
The company says ChatGPT ads is now available in more than 60 countries.
00:06:49.040 --> 00:06:54.399
Advertisers can work through OpenAI, agency and technology partners or self-service ads manager when eligible.
00:06:54.560 --> 00:06:56.800
Ads remain limited to people on the free and go plans.
00:06:56.959 --> 00:07:00.480
OpenAI says plus, pro, and enterprise subscriptions remain ad-free.
00:07:00.560 --> 00:07:05.360
The company also says advertiser content is labeled and separated from ChatGPT's independent answers.
00:07:05.519 --> 00:07:11.040
Wider availability matters, but OpenAI says more formats, optimization tools, and measurement solutions are still coming.
00:07:11.199 --> 00:07:12.720
Distribution is ahead of maturity.
00:07:12.879 --> 00:07:20.079
A market being available does not tell us whether the channel creates incremental demand, captures demand created elsewhere, or moves qualified customers at an acceptable cost.
00:07:20.240 --> 00:07:22.720
Pauline Jacober's paid media guidance is useful here.
00:07:22.879 --> 00:07:27.439
More capture budget does not create more demand when a company cuts the channels that build awareness and trust.
00:07:27.600 --> 00:07:31.439
Search ads, local services ads, and conversational ads can collect intent.
00:07:31.519 --> 00:07:35.439
They cannot replace every exposure that caused the person to recognize and trust the brand.
00:07:35.600 --> 00:07:39.439
For an eligible advertiser, start with one market, one offer, and a clear success event.
00:07:39.600 --> 00:07:44.319
Keep the claims current, track qualified outcomes in the CRM, and compare the result with other markets or channels.
00:07:44.480 --> 00:07:47.839
Do not fund the test by automatically stripping money from the work that creates demand.
00:07:48.000 --> 00:07:49.920
The last story is Meta Enterprise Platform.
00:07:50.000 --> 00:07:56.240
Meta announced a new enterprise AI business and named CJ Desai as Chief Enterprise Platform Officer, reporting to Mark Zuckerberg.
00:07:56.319 --> 00:08:01.920
The initial list includes the Muse agent, Meta Business Agent, Muse API, Muse Code, and other tools for businesses and developers.
00:08:02.079 --> 00:08:04.079
The announcement names a direction and an executive.
00:08:04.240 --> 00:08:10.800
It does not give buyers pricing, general availability, implementation requirements, support terms, detailed security documentation, or customer results.
00:08:10.959 --> 00:08:13.839
Larry Dignan's launch day analysis is appropriately cautious.
00:08:13.920 --> 00:08:19.120
He said the details are too sparse for enterprises to evaluate and that the product and partner ecosystem will take time to build.
00:08:19.199 --> 00:08:23.040
Meta may publish those details soon, but they are not in the launch material reviewed here.
00:08:23.199 --> 00:08:24.160
Keep it on the watch list.
00:08:24.319 --> 00:08:33.600
Before recommending a pilot, ask for product documentation, data handling terms, security controls, pricing, integration requirements, service commitments, and a customer example that resembles the proposed use.
00:08:33.759 --> 00:08:35.759
An announcement is enough to start a question list.
00:08:35.840 --> 00:08:37.279
It is not enough to start a migration.
00:08:37.440 --> 00:08:39.279
Keep these six lines handy for client calls.
00:08:39.440 --> 00:08:41.200
Google is still rolling out the spam update.
00:08:41.279 --> 00:08:44.720
We will isolate the affected pages and outcomes before recommending broad changes.
00:08:44.879 --> 00:08:49.919
Search Console can now show which pages appear after image-led searches, but it cannot show the image or query behind them.
00:08:50.080 --> 00:08:55.919
IP and timestamp matching may increase reach, but we will review consent, regional exclusions, and file handling before testing it.
00:08:56.000 --> 00:08:59.840
Microsoft's new report can show whether the search, ad headline, and landing page lineup.
00:09:00.000 --> 00:09:01.919
We will use it to find problems, then test the fix.
00:09:02.080 --> 00:09:04.159
ChatGPT ads has wider distribution.
00:09:04.240 --> 00:09:08.080
We still need qualified lead and sales data before treating it like a mature acquisition channel.
00:09:08.240 --> 00:09:10.559
Meta announced a direction, not a complete buying decision.
00:09:10.639 --> 00:09:13.440
We will wait for product, security, pricing, and customer details.
00:09:13.519 --> 00:09:14.240
That is the week.
00:09:14.399 --> 00:09:15.200
Keep influencing.