What Changed in Digital Marketing: Google's Measurement Push

Key Takeaways

• Google added IP addresses and timestamps as Customer Match signals on 25 September, giving advertisers more ways to match CRM records to Google accounts without relying solely on hashed emails.
• Search Console rolled out a multimodal filter from 24 September, showing page-level data for image-based searches including Google Lens, Circle to Search and photo uploads.
• Performance Max for store goals now has a Local Customer Optimisation toggle, prioritising nearby in-market customers across Maps, Waze and local Search.
• Google Ads Asset Studio is rolling out AI video dubbing across 33 languages, letting advertisers localise video creative without a reshoot.
• The September 2026 spam update started 24 September and may run for two weeks; a separate post covers what to check.
• The thread connecting every update this fortnight is the same: Google keeps expanding what counts as a data signal, and the campaigns already set up to feed it well widen their lead over the ones that are not.

---

The fortnight to 30 September produced four Google updates worth knowing about, plus the fourth spam update of the year. None are surprising in isolation. Taken together they describe a single direction: Google keeps widening what counts as a data input, and the accounts set up to send it clean, accurate signals widen their lead over the accounts that are not.

Search Console Can Now See Your Image Search Traffic

From 24 September, Search Engine Journal reports that Google started rolling out a multimodal filter in Search Console’s Performance report globally. It covers searches where someone used an image rather than text: Google Lens, Circle to Search on Android, image uploads to Google Search, and Chrome’s right-click “Search this image” feature.

Until now, that traffic arrived at your site as a single undifferentiated block. You could not tell which pages were getting found through visual search, let alone compare them with text search performance. The new filter sits under the Web search type and shows page-level data broken down by country and device.

There is a deliberate limitation: because multimodal searches are image-driven, there is no text query data. You will see the pages being reached, not the visual query that found them. Google is also rolling out multimodal reporting to the generative AI performance report, which reached all sites from 31 August.

What this means in practice

The businesses with the most to gain are ones with strong visual content: trade businesses with portfolio pages, interior designers, landscapers, property developers, architects, and any ecommerce store with product photography. If someone photographs your finished work with their phone and searches Google to find who did it, that visit is now visible in a way it was not before.

Two things are worth checking once the filter appears in your property. First, which pages are pulling image search visits? If the pages receiving that traffic are not your conversion pages, the question is whether to improve them or link them more directly to pages that convert. Second, how does image search traffic behave compared to text? Lower time on site would suggest people are getting a visual confirmation and leaving. High engagement would suggest genuine intent behind the search.

Neither of those questions was answerable before 24 September.

Customer Match Now Takes IP Addresses and Timestamps

On 25 September, Search Engine Land reported that Google added IP addresses and timestamps as matching signals to Customer Match. When uploading a Customer Match list, advertisers can now include the IP address associated with a user interaction alongside the timestamp of that interaction.

Unlike hashed emails and phone numbers, IP addresses and timestamps are uploaded un-hashed. If the IP address corresponds to one Google can identify, Customer Match can connect that user to their Google account even without a direct email or phone match.

One regional restriction to note: the update is unavailable for end users in the European Economic Area, UK or Switzerland. For New Zealand, it is available.

The arithmetic of a higher match rate

Customer Match is only useful if the match rate produces an audience large enough to do something with. Small lists with poor match rates generate audiences so thin that bidding adjustments and exclusions have limited practical effect.

Here is a concrete example. Suppose you have a CRM list of 2,000 past website visitors with email addresses. A typical match rate for consumer email lists sits around 50 to 60 percent, giving you a workable audience of 1,000 to 1,200 matched users. If adding IP addresses and timestamps lifts that rate to 65 or 70 percent, you get 200 to 400 more matched users in your remarketing or exclusion pool. That is a meaningful difference when your list is small to begin with.

Those extra matched users matter most when you are using Customer Match for exclusions, not just targeting. If your acquisition campaign is excluding existing customers to avoid wasting spend on people who have already converted, a higher match rate means fewer existing customers slipping through and seeing ads aimed at strangers.

Whether the lift materialises depends on your data. If your website does not log IP addresses against form submissions or session events, you cannot pass the data even if Google can now use it. The first step is checking whether your CRM or analytics platform captures IP at the point of conversion, and whether it is exportable in a format you can append to a Customer Match upload.

Performance Max Gets a Local-First Toggle and a Cleaner Sales Loop

Two features for businesses with physical locations arrived close together this fortnight, covered by Search Engine Land:

Local Customer Optimisation is a campaign-level toggle in Performance Max campaigns running store goals. When switched on, Google prioritises budget toward people who are actively nearby and in-market, using real-time navigation and local intent signals across Google Maps, Waze and local Search. Advertisers creating a new Performance Max campaign for store goals will see the toggle during campaign setup, after selecting an offline conversion goal: store visits, directions or store sales.

Store Sales in Data Manager is expected to roll out from the end of September. It gives advertisers a simpler way to feed offline transaction data back into Google Ads via CRM or Google Sheets connections. Google recommends uploading daily or weekly. Transactions within the previous 30 days are required for dynamic conversion value reporting, with data from the previous 14 days the recommended window for the model to produce accurate results.

Why both matter together

These two features only produce meaningful results in combination. Local Customer Optimisation sends budget toward people near your location. Store Sales in Data Manager tells Google what happened after those people walked in. Once both are running, Smart Bidding has a closed loop: it can learn the difference between a nearby visitor who spent $40 and one who spent $400, and adjust bids toward the more valuable customer type accordingly.

Without the measurement step, you are optimising for foot traffic with no idea whether it converts to revenue. Without the targeting step, you have offline conversion data but no mechanism to reach more people who look like your best customers.

The gap we see most often is the measurement side not being wired up. Most businesses running store goals in Performance Max have set a store visit goal, but few are feeding actual transaction data back in. This update makes the measurement side easier to connect. If you have a POS system or CRM that records transaction values, the Google Sheets path is now a viable integration without custom development. The practical prompt is simple: what did you earn from in-store customers last month, and is that number visible to Google Ads?

AI Video Dubbing in Asset Studio: 33 Languages, Free for Select Users

Search Engine Land reports that Google Ads is rolling out Video Ads Dubbing in Asset Studio. It takes an existing video ad and dubs it into 33 languages and locales using AI, generating translated audio to match the original delivery. The feature is currently free for select users.

The appeal is clear: a video you have already produced can reach audiences who do not speak English, without a production budget or a return trip to the studio. Google’s own promotional data cites an 86 percent increase in click-through rate and a 75 percent reduction in cost per click, though those figures come from Google’s own marketing materials and are not independently verified. They describe a direction, not a guarantee.

What to expect in practice

AI dubbing results vary based on source material. Videos with clear single-speaker audio, direct delivery and minimal background music dub cleanest. Videos with multiple speakers, heavy sound design or complex audio layering will produce inconsistent results, and the mismatch between mouth movements and dubbed audio becomes noticeable in longer formats.

For most NZ businesses, the lowest-risk path is to take a performing video, dub it into one additional language relevant to a specific target market, run it with a modest budget against an audience matched to that language, and compare cost per lead against the English baseline. A defined pass/fail metric is essential. Without one, a small test runs for months and produces no conclusion. With one, you know within six to eight weeks whether to scale or move on.

The September 2026 Spam Update

On 24 September, Google started rolling out the September 2026 spam update, the fourth of the year. It applies globally and may take up to two weeks. We published a full breakdown of what to check and what not to do on the day it started.

The short version: if your traffic is steady, do nothing. If it drops, start with Manual actions and the Pages report before drawing any conclusions. Making large changes to your site during a rollout window makes it impossible to isolate what caused what.

What Connects All Four

Every substantive update this fortnight either adds a new data signal or closes a feedback loop that was previously open.

Customer Match accepts more identifier types. The GSC multimodal filter surfaces a class of traffic that was invisible until now. Local Customer Optimisation makes real-time geographic intent a bidding input, and Store Sales in Data Manager closes the revenue loop from nearby click to in-store transaction. Video dubbing is the least obvious fit, but it is still fundamentally about signal: does this creative generate conversions in a second language, and by how much?

This pattern is consistent with how Google has been building for two years. Its AI campaign systems perform best when they have wide, accurate conversion and audience signal. Widening what counts as a signal is not random product work. It is Google extending the reach of systems that already perform well when fed accurately, into territories where they previously had no data.

The practical consequence is the same one it has been: campaigns with clean, complete conversion tracking and rich audience data capture the benefit of these updates automatically. Campaigns with incomplete measurement, no offline data, and thin Customer Match lists do not. This fortnight added four more places where that gap widens.

What To Do This Week

None of these updates require the kind of urgent action the AI Max migration dates demanded in August. Four checks are worth scheduling in the next fortnight.

Check Search Console for the multimodal filter. Log in, open Performance, filter by Web search type, and look for the multimodal option. If it has not appeared yet, the rollout is global and should arrive shortly.

Review whether your CRM or website logs IP addresses against conversions. If it does, check whether that data is exportable in a format you can append to a Customer Match upload. If it does not, this is a configuration question worth raising with whoever manages your analytics or CRM setup.

Check whether your Performance Max campaigns for physical locations have the Local Customer Optimisation toggle switched on, and whether Store Sales in Data Manager is available in your account. If you have a POS or CRM with transaction values, the Google Sheets integration is the fastest path to closing the offline conversion loop.

Hold off on reacting to spam-related traffic fluctuations until around 8 October, which is two weeks from the rollout start. Rankings can move in either direction during a rollout window, and acting before it is finished makes diagnosis unreliable.

---

We manage Google Ads and SEO for NZ businesses, which means the above is the same checklist we run through in our own accounts each fortnight. If you would rather have someone else run it on yours, get in touch.

Frequently Asked Questions

Is the Customer Match IP address update available for New Zealand advertisers?

Yes. The restriction applies to end users in the European Economic Area, United Kingdom and Switzerland. New Zealand is not in any of those regions, so the update is available here.

What is the multimodal filter in Google Search Console?

It is a new filter in the Performance report’s Web search type that shows which pages receive traffic from image-based searches, including Google Lens, Circle to Search, photo uploads and Chrome’s right-click image search. It shows page-level data by country and device, but does not include the visual query itself because most multimodal searches do not involve a text component.

Do I need to upload my own transaction data for Store Sales in Data Manager to work?

Google can apply default conversion values without your data, but dynamic value reporting requires your own transaction uploads covering at least the previous 30 days. Without your data, Google assigns a default value to every store conversion, which will be inaccurate for most businesses and will lead Smart Bidding to optimise toward the wrong customer type.

Can AI video dubbing replace professional translation for video ads?

For simple, direct-to-camera formats with clear audio, it can produce results worth testing. For complex productions with multiple speakers, heavy music, or nuanced scripting, professional translation will produce significantly better output. Set a clear pass/fail metric before running any dubbed test so you can draw a conclusion rather than running an inconclusive small budget for months.

Should NZ businesses be worried about the September spam update?

Only if your site has practices that match Google’s spam policies, such as large volumes of thin AI-generated content, purchased links, or unrelated third-party content hosted on your domain. We published a full breakdown of what the September spam update targets and what to do if traffic drops.