Google Ads Changes in October 2026: 12 Account Fixes
October 2026 was one of the densest months of the year for Google Ads changes — and most of them switched on by default, without anyone asking you. Two metrics were redefined, a targeting control quietly turned from a hard limit into a suggestion, and Google started lifting discount codes off your own landing pages and putting them into your ads.
Here are 12 concrete fixes worth making in the account before December, plus three deadlines you cannot afford to miss. Each item follows the same shape: what changed → why it matters → what to do. One caveat up front: Google rolls these out in stages, so some features may not have reached your interface yet. Go by what you see in your own account, not by the announcement date.
Part 1. Metrics that changed meaning
1. View-through conversions in Demand Gen are counted differently
What changed. For Display inventory inside Demand Gen, the bar for counting an impression moved from a viewable impression (half the ad on screen for at least a second) to a rendered impression — so impressions register far earlier and far more often.
Why it matters. Your VTC count went up on its own, with no change in actual performance. If view-through sits in your Conversions column and a bid strategy optimises against it, the algorithm is now reading an inflated signal. If you report ROAS including VTC, that improved for free too.
What to do. Check which conversion actions are marked primary and pull view-through out of the primary column if it slipped in. Compare October onward only to October: month-over-month on this metric is meaningless across the change. The mechanics are unpacked in our piece on Google Ads metric changes, and the underlying billing model in Demand Gen CPM billing.
2. The video engagement threshold moved from 5 to 10 seconds
What changed. An engagement now requires ten seconds of viewing instead of five, across in-stream, in-feed, mastheads and Shorts. Non-skippable and bumper formats are unaffected.
Why it matters. Engagement rate and engagement counts dropped in reporting — industry coverage puts the drop somewhere in the 30–50% range, which is a rough indication rather than a figure to plan against. Nothing about campaign performance got worse. The danger is a team seeing the drop, concluding the creative stopped working, and fixing something that isn’t broken.
What to do. Rebaseline. Mark the change date in your reporting and stop comparing across it. If you run automated rules or alerts on engagement declines, pause them while you recalibrate, or they will fire on a phantom signal. Where viewers genuinely drop off is a different report — the audience retention curve — and that one is unaffected by the redefinition.
Part 2. Settings that switched themselves on
3. Text overlays on Demand Gen video are on by default
What changed. Google generates text from your existing headlines and descriptions and overlays it on video across YouTube, Discover and Gmail. The stated rationale is that most feed viewing happens without sound.
Why it matters. On videos with their own typography, tight editing or legal disclaimers, the overlay covers the frame. For regulated verticals that is a compliance risk: a mandatory disclosure can end up underneath an auto-generated banner.
What to do. Open your Demand Gen campaigns, go to asset optimisation, and look at the previews. If you don’t like what you see, the whole feature can be disabled there. Worked examples are in our article on AI text overlays on Demand Gen videos; disclosure requirements are covered in text disclaimers in Google Ads.
4. Demand Gen lookalike segments became a suggestion
What changed. The similarity tiers — Narrow 2.5%, Balanced 5%, Broad 10% — no longer act as a hard reach ceiling. They are now a signal the system weighs and can exceed. October brought the corresponding controls and labelling into the interface.
Why it matters. If you built your economics on a narrow segment and assumed delivery stayed inside it, your real reach has been wider than advertised for some time. That accounts for a share of otherwise unexplained CPA drift in Demand Gen over recent months.
What to do. Pull Demand Gen CPA by week for the whole year and find where it moved. From there you either accept the broader reach and reset targets, or tighten control through audience exclusions and better seed lists. The mechanics are in lookalike segments in Demand Gen.
5. Google reads promo codes off your website
What changed. Automated promotions scan landing pages, find live offers and attach them to eligible Search and Performance Max campaigns. It works on an opt-out basis — no approval required from you.
Why it matters. If your site still carries last spring’s “30% off through March 31” banner, or a staging page with a dead promo code, Google may faithfully surface it in an ad. What follows is an annoyed user, a complaint, and in the worst case a question about whether your ad matches its landing page.
What to do. Inventory the offers on your site: what is live, what someone forgot to remove. Delete or block anything stale. If your promotions change unpredictably, opt out and manage promotions manually instead. Full breakdown: automated promotions in Google Ads.
Part 3. New levers worth pulling
6. Channel prioritisation and budget allocation
What changed. Priority controls rolled out globally: Search, Shopping, Performance Max and Demand Gen can each be flagged High or Low priority. Google’s automation funds High priority campaigns fully within your target CPA or ROAS before allocating anything to Low priority, and Low priority campaigns can be capped at a percentage of daily budget. It works at account, campaign and shared-budget level.
Why it matters. This is the first honest lever in a long time against “Performance Max ate the budget that was meant for brand search.” In practice it is cannibalisation control at the level of money rather than keywords.
What to do. Assign priorities from unit economics, not from habit. The full breakdown with example allocations is in channel prioritisation in Google Ads; the underlying overlap problem is in PMax, Search and Shopping competing for one demand pool.
7. Conversion Lift went self-serve
What changed. Incrementality studies for Search and Performance Max can now be launched by advertisers themselves from the experiments section, with no Google rep involved. Reported eligibility sits around a thousand conversions and roughly $5,000 in campaign budget — treat that as an indication and confirm the thresholds in your own account.
Why it matters. It is the only way to answer “how many conversions did this advertising actually add” without arguing about attribution models. That matters most for brand search and PMax, where click attribution systematically flatters.
What to do. Put one lift study into the Q4 plan, ideally on the campaign whose value you most suspect. Details in Conversion Lift for Search and Performance Max, with the general testing method in running experiments in Google Ads.
8. Performance Max asset experiments for everyone
What changed. A/B testing of creative sets in PMax is available to all advertisers rather than a limited group.
Why it matters. Creative in PMax used to be a matter of faith: you swapped assets and things seemed better. Now there is a controlled comparison of two sets on the same traffic.
What to do. Start early. Meaningful reads take weeks, not days, and launching a test in the middle of peak season is already too late. How it works: Performance Max asset experiments.
9. The AI Max search terms and ad combinations report
What changed. A report began rolling out that ties the query, the headlines and descriptions that served against it, and the landing page into a single row.
Why it matters. The main complaint about AI Max was opacity: the system expands queries and assembles copy on its own. Now the path is visible end to end, which makes it steerable through negatives, brand lists and text rules.
What to do. Block an hour to read the report for your most expensive campaign — the first pass usually produces a batch of negatives on its own. Detail: AI Max reporting.
10. AI campaign localisation beta
What changed. A tool that translates a Search campaign end to end — headlines, descriptions, keywords, images and the landing page — into additional languages. Access shows up via the tools menu or recommendations.
Why it matters. The cost of testing an adjacent market falls from “hire a native speaker and budget a week” to “run a test on a slice of spend.” Machine translation still needs a human pass, though: it reliably mangles legal wording and product names.
What to do. If there is an obvious neighbouring market, test on a capped share of budget — but proofread the assets before launch, every time. Breakdown: AI campaign localisation.
Part 4. Shopping and AI Mode
11. Promotional badges in AI Mode
What changed. Shopping ads from brands with active Merchant Center promotions started picking up an offer badge inside AI Mode results.
Why it matters. Shopping ad visibility in AI Mode is still small, but a badge is a rare visual differentiator in an environment that shows a handful of products per query instead of a full carousel.
What to do. Submit promotions in Merchant Center early, allowing for the review window (see deadlines below). How to structure them: Merchant Center promotions.
12. Universal Cart and merchant registration
What changed. A single cross-retailer cart launched in US Search and the Gemini app: shoppers add products from different merchants and check out without visiting the sites. Registration runs through the relevant Merchant Center hub. A dashboard showing product visibility in AI Mode arrived alongside it.
Why it matters. A purchase completed without a site visit is a purchase your web analytics will not see the way it used to. Channel reporting has to be rebuilt around that.
What to do. Decide deliberately whether you participate, rather than discovering it in a report afterwards. What it breaks in measurement: AI Mode native checkout.
Three deadlines
| Date | What happens | Who it affects |
|---|---|---|
| 16 November 2026 | Data Manager stops requiring SHA-256 pre-hashed customer data and accepts encrypted data decrypted inside a trusted execution environment | Anyone running a custom hashing pipeline before uploading audiences or offline conversions |
| Late November 2026 | Submission window for peak-season promotions: review can take up to 72 hours, so they need to go in early | Anyone running Shopping or PMax with a product feed |
| February 2027 | Legacy Display controls retire and the dynamic search campaign migration completes | Anyone still running DSA or standalone Display campaigns |
The first deadline is the most technical and the most painful. If an engineer set up hashing a year ago and has since left, and the upload breaks in November, the fix happens under pressure. Check the pipeline now. How first-party data transfer works: the Data Manager API and confidential matching.
The third deadline looks distant, but the action is an October one: if you still run Display traffic, move it to Low priority with a capped allocation now, so the distribution has settled by the time the legacy controls disappear.
Other Google Ads changes worth noting from October
- The Data Strength Uplift metric appeared in the interface — an estimate of how much your data completeness contributes to results. Useful ammunition when arguing for investment in tracking.
- Demand Gen performance-based cost adjustment extended its evaluation window from three weeks to 60 days.
- Google representatives confirmed that target and budget edits on a steady-state campaign scale bids proportionally rather than resetting learning. That retires one myth without licensing reckless edits — context in the Smart Bidding learning period.
The headline lesson of October: three of the twelve Google Ads changes altered your reports, not your campaigns. Before fixing anything, make sure the number actually fell rather than being redefined.
This week’s running order
- Check which conversion actions are primary and remove view-through if it ended up there.
- Mark metric redefinition dates in reporting and pause alerts that will fire on a phantom drop.
- Open Demand Gen: text overlays, lookalike segments, year-to-date CPA trend.
- Inventory the offers on your website — before the automation finds them.
- Assign channel priorities based on economics.
- Verify the customer-data upload pipeline before 16 November.
- Submit Merchant Center promotions with review time to spare.
- Schedule one lift study and one asset experiment.
If that pass turns up disapproved ads or questionable policy decisions, the appeal process is covered in Policy Manager and appeals, and what a Google employee can actually do about it is in today’s companion piece on Google Ads support and your account rep. If you pull reporting through the API, several October changes alter the values you export — see writing GAQL queries against the Google Ads API.
For large accounts where walking this list by hand takes days, it is worth looking at agency account setups with sane access structure, the Google Ads guide for systematising the process, and the PPC Rebels team if you want an outside pair of hands on the audit.
FAQ
Why don’t I see some of these features in my account?
Google rolls changes out in stages, by country, account type and spend level. An announcement date is not an availability date. Work from what is visible in your own interface and recheck every couple of weeks.
Does any of this matter if I don’t run Demand Gen?
Yes. Channel prioritisation, automated promotions, self-serve Conversion Lift and the November customer-data deadline are independent of Demand Gen and apply to most accounts.
My video engagement rate collapsed. Is that the 10-second threshold?
Most likely, if the drop lines up with the change date and impressions, clicks and conversions held steady. Check the shape first: a metric redefinition produces a one-day cliff, while genuine creative fatigue declines gradually.
How do I turn off text overlays on video?
Demand Gen campaign settings include an asset optimisation section with a disable option. Look at the previews first — on some videos the overlay genuinely helps.
What happens if I ignore automated promotions?
Google will attach whatever offers it finds on your site to eligible campaigns. If every offer is current, that can work in your favour. If expired promotions are still live on the site, you are risking a mismatch between ad and landing page.
Should Performance Max be set to Low priority?
It depends on which campaigns actually produce margin. The working rule: High priority for campaigns with demonstrated incrementality, Low for those suspected of harvesting demand you already had. Test that with a lift study rather than intuition.
What exactly breaks on 16 November?
Nothing breaks by itself — the requirement for how data is transmitted changes. The risk is that bespoke pipelines built around the old pre-hashing scheme need to be brought in line ahead of time.
Should I register for the universal cart?
That is a sales channel decision rather than an advertising one. The upside is presence where the purchase completes. The downside is that the transaction happens off your site and conventional web analytics will not see it. Decide deliberately and plan the measurement first.
Can I restore the old lookalike behaviour?
The hard reach ceiling is gone. Control now lives in seed list quality, audience exclusions, and the targets your bid strategy works against.
How do I know the changes haven’t broken my economics?
Watch money, not intermediate metrics: weekly spend, orders and revenue. A redefinition is obvious — engagement and view-through jump or fall in a step while spend and revenue stay flat.
How often do Google Ads changes like this ship?
Major updates come in waves, typically several notable ones a month, plus the regular API release cycle. A sensible habit is one hour a month spent reviewing changes and adjusting the account, rather than discovering them through a broken report.
What if a change has already hurt a campaign?
Fix the date first and confirm causation through change history, then roll back what can be rolled back and reset targets to the new reality. Panic bid edits at that moment usually just add noise.
