Written by Jeremy Souffir Founder, JTS Tech Services
The short version: on 18 August 2026, Google sunset the Content API for Shopping, the interface that has pushed product data into Merchant Center for the better part of a decade. Its replacement, the Merchant API, has been generally available since August 2025. And per Google's own migration guide, "starting September 1, 2026, Content API for Shopping requests will experience progressive errors." That sentence is the entire news, and the important word in it is progressive. This is not a plug being pulled. It is a tap being closed slowly, over a period Google has not published, on a system whose failure mode is not an error page but a slightly out-of-date price. If you are on Shopify's Google & YouTube channel or Woo's Google Listings & Ads, the migration was handled on the platform side and you can read the rest of this out of professional interest. If your catalogue reaches Google by any other route, the next twenty minutes are the cheapest twenty minutes in your quarter.
What exactly did Google switch off?
Here is what is actually on the record, taken from Google's developer documentation and Merchant Center help pages rather than from anyone's summary of them. We have kept it to things Google says in its own words, because this is a subject where the trade coverage has been looser than the source material.
- The Content API for Shopping was sunset on 18 August 2026. Google's migration guide states it in the past tense: "Content API for Shopping was sunset on August 18, 2026." The Merchant Center help page announcing the change had committed to providing access "until August 18, 2026, after which the Content API will be shut down."
- Since 1 September 2026, requests to it experience progressive errors. Google's wording, repeated on both the migration guide and the Merchant API latest-updates page, is: "Starting September 1, 2026, Content API for Shopping requests will experience progressive errors."
- Google has not published a schedule for that ramp. We looked. There is no public table of percentages, no stated end date for the degradation, and no announced hour at which the last request fails. We are not going to invent one, and you should be sceptical of anyone who quotes you a figure.
- The Merchant API is the replacement and has been generally available since August 2025. It is not a beta you are being pushed into early; it has been shipping for a year, and Google directs all calls to its v1 or v1alpha versions.
- There is an extension route. Google's documentation says that if you need more time, you can apply for extended access to the Content API through your Google representative. This is a real escape hatch, and it is one reason not to panic — but it is an application, not a setting, and it has to be made by a person.
- This is the write path, not the reporting path. It governs how product data gets into Merchant Center. It is a different thing from the Merchant Center reporting definitions Google restated on 24 August, which we covered a fortnight ago and which changed what your numbers meant rather than whether they arrived.
Why is "progressive" the word that matters?
Because every instinct you have about integrations breaking is calibrated for the loud kind. A credential expires, a certificate lapses, an endpoint returns 404, and within the hour somebody is in a channel asking why the sync is red. Those failures are unpleasant and they are also self-reporting: the system tells you it has stopped. A progressive error rate does none of that. Some requests succeed. Some fail. Your integration, if it is like most integrations, retries the failures a couple of times, logs something nobody reads, and moves on to the next batch. The dashboard stays green because the job completed. It completed having quietly dropped a portion of its payload.
Now think about what a product feed actually carries and what a partial delivery does to it. It is not one artefact that is either present or absent. It is a stream of small corrections: this SKU went out of stock at 11am, this price dropped for the promotion, these forty items got new titles, this one is discontinued. Miss a share of those and nothing disappears. Your catalogue in Merchant Center stays fully populated, fully approved, and increasingly fictional. The products people see are your products. The prices they see are the prices you charged before the sync started dropping updates.
What we know, what we don't, and what we are inferring
Three things, separated deliberately, because they get run together everywhere else. What we know, because Google states it: the sunset happened on 18 August, progressive errors began on 1 September, the Merchant API is the replacement, and an extension can be applied for. What we do not know, and neither does anyone quoting you a number: how steep the ramp is, whether it is uniform across accounts, and when it reaches complete failure. Google has published none of that, and the honest answer to "how long have I got?" is that the only safe assumption is less time than you would like. What we are inferring, and flagging as inference rather than fact: that the practical symptom for most affected merchants will be stale data rather than visible errors. That follows from how retry logic and feed processing normally behave rather than from anything Google has said about it, and if your integration surfaces hard failures loudly, you are in a better position than this post assumes.
So does this land on you?
Probably not, if you are on a first-party channel, and we would rather lose your attention here than hold it dishonestly. Feed specialists writing about the migration have been consistent on the split: merchants using Shopify's Google & YouTube channel or WooCommerce's Google Listings & Ads have been carried across by their platform, and the same broadly goes for the larger established feed-management tools, which have had a year of notice and commercial reasons not to break their customers. That is genuinely most of the market, and it is the reason this is not a five-alarm post. The exposure is concentrated somewhere much less visible.
- Custom integrations. Anything a developer or agency wrote directly against the Content API — the script that pushes stock levels from your warehouse system every night, the middleware between your ERP and Merchant Center. These do not migrate themselves and nobody is watching them on your behalf.
- ERP and PIM connectors, especially older or self-hosted ones. The vendor may have shipped a Merchant API version; whether your instance has been updated to it is a separate question, and on self-hosted software it is usually your question.
- Smaller or dormant feed apps. A tool still installed and still charging you is not the same as a tool still being maintained. The ones that stopped being actively developed are exactly the ones that will not have migrated.
- Anyone running a hybrid setup. If the platform channel handles your core catalogue but something custom pushes supplemental data — promotional pricing, local inventory, custom attributes — the primary feed can be perfectly healthy while the supplement quietly stops arriving.
- Second and third storefronts. The main store gets attention. The regional storefront, the B2B store, the one from the acquisition that still runs on its own stack — those are where a 2019 integration lives undisturbed.
One further note, offered carefully because we have not verified it ourselves: some feed specialists have reported that merchants moved onto the newer API have seen configuration behave differently afterwards — feed rules not applying as before, supplemental feeds needing to be relinked. Treat that as a prompt to check your settings after any migration rather than as an established defect. The general principle holds regardless: a migration that someone else performed on your behalf is still a change to your setup, and it is worth looking at the result.
The part that connects to everything else we have written this year
For most of the last decade, a stale product feed was a paid-search problem with a well-understood cost. Your price on Google did not match your price on the landing page, Google noticed the mismatch, and your items got disapproved. Painful, but loud — disapprovals appear in a dashboard, and somebody eventually looks at the dashboard. That is a system with an alarm attached, even if the alarm rings late.
That same catalogue data now also feeds the surfaces where an assistant assembles a recommendation. This is the thread running through nearly everything we have covered since the spring: what your feed contains determines whether you are in Google's cross-store shopping experiences, what your pages say in their opening sentences determines what an assistant repeats about you, and the quality of your structured product attributes decides whether you are eligible for placements at all. The failure mode on that side has no dashboard. An assistant does not disapprove you and it does not write to tell you your data looked old. It recommends a competitor whose price it trusts, and the only signal you get is an absence you cannot measure. We spent last Tuesday's post on the fact that Google has finally started reporting your AI impressions back to you. It is worth noticing that a supply line into those same surfaces started degrading the day before that report finished rolling out.
The two conclusions that both get this wrong
The first wrong conclusion is that this is a developer's problem and therefore not yours. It is not a developer's problem; it is a merchandising problem with a technical cause. The person who will discover it is not the person who can fix it, and they will discover it in October, in the form of an order at a price you withdrew in August or a bestseller that has been showing as in stock since it sold out. The second wrong conclusion is the mirror image and it is the more common one this week: reading a deadline post like this, feeling the jolt, and firing off a panicked message to your agency asking whether you are broken. That is not a bad instinct, it is just an imprecise one, and imprecise questions get reassuring answers. "We're on the Merchant API" is true of the main channel in almost every case and tells you nothing about the nightly stock script or the supplemental promotional feed. The useful question is not whether you have migrated. It is which specific systems write to Merchant Center, what each one authenticates as, and when each of them last successfully wrote something. Ask that and you will get a real answer, possibly an uncomfortable one. Both mistakes share a root: treating a slow failure as though it will announce itself the way a fast one does.
What would we actually do this week?
- Open Merchant Center and look at your data sources. Every source is listed there along with how it is fed. This is the single highest-value five minutes available, because it turns "I think we're fine" into a list you can actually check off.
- Write down every system that writes to Merchant Center, not just the obvious one. The platform channel, the feed tool, the nightly script, the promotional supplement, the regional storefront. The ones nobody names in the first pass are precisely the ones this affects.
- Check the last-updated timestamps on your product data rather than trusting a green status. A job that completes having dropped part of its payload reports success. A price that has not changed since the third week of August reports the truth.
- Spot-check a dozen live listings against your own admin. Pick items whose price or stock has moved since mid-August — promotional lines and fast-moving SKUs — and compare what Google shows to what your system says. This takes ten minutes and it is the only test here that cannot be fooled by a dashboard.
- If you find something unmigrated, ask about the extension route in the same message as the fix. Google's documentation says extended access to the Content API can be applied for through your Google representative. It buys breathing room for a proper migration rather than a rushed one, and it is worth knowing whether it is available to you before you need it.
- Set an alert on feed freshness, not just feed errors. The lasting fix here is not this migration, it is having something that notices when your catalogue stops changing. Any monitoring that can say "this data source has not successfully updated in 48 hours" would have caught this class of failure, and will catch the next one.
- Do not rip out a working integration this week on the strength of a blog post. If your data is current and your timestamps are moving, you are fine, and September is a bad month to perform unnecessary surgery on the pipe that feeds your Q4.
The genuinely reassuring part
Set the warnings aside for a paragraph, because the underlying situation is better than the framing of any deadline post suggests. Google gave a year of notice on a well-documented replacement, the major platforms did the work for the majority of merchants without asking them to think about it, the degradation is gradual rather than instant, and there is a documented route to ask for more time. That is close to the most humane way to retire a piece of critical infrastructure, and it is worth saying so in a week where the reflexive tone about large platforms is that they break things carelessly. The genuinely useful takeaway is also not really about this API at all. It is that you now have a reason to write down every system that touches your product data, which is a list most businesses have never made and every business benefits from having. The migration is a deadline and it will pass. The inventory outlives it, and the next time something in that chain changes — and something will, because this is the third change to how product data reaches Google that we have written about since July — you will be answering the question in an afternoon instead of a fortnight.
Where we fit
The reason a business would bring us in over this is not the migration itself. If you have one integration and a competent developer, that is a scoped piece of work and you do not need us for it. What businesses actually get stuck on is the question underneath it: nobody can produce a current list of what writes to what. The nightly script has no owner because the person who wrote it left, the supplemental feed was set up during a campaign in 2023, and the honest answer to "what would break if this stopped?" is a shrug. That is precisely the shape of an AI Ops Automation Sprint. We inventory every integration and scheduled job that touches your product and customer data, name an owner and a credential for each, find the ones that are failing silently or running on nothing but momentum, and leave you with freshness monitoring on the paths that matter — so the next deprecation notice is a calendar entry rather than an archaeology project. The migration is this month's symptom. The absence of an inventory is the condition, and it is the one that keeps producing months like this.
Sources
- Google for Developers — Start your migration: Content API for Shopping to Merchant API (the primary source for the two dates this post turns on: "Content API for Shopping was sunset on August 18, 2026" and "Starting September 1, 2026, Content API for Shopping requests will experience progressive errors", plus the migration steps themselves)
- Google for Developers — Merchant API latest updates (independent confirmation of both statements on Google's own release feed, the instruction to direct calls to v1 or v1alpha, and the note that merchants needing more time can apply for extended access to the Content API)
- Google Merchant Center Help — Introducing Merchant API (Google's original announcement to merchants: the Merchant API as the new primary tool for managing product data, and the commitment to provide Content API access "until August 18, 2026, after which the Content API will be shut down")
- Google for Developers — Migrate from Content API for Shopping to Merchant API (the compatibility overview, and the place to check whether the specific methods your integration calls have direct Merchant API equivalents)
- Producthero — Google is sunsetting the Content API for Shopping: what you need to know (the source for our platform-by-platform split — that Shopify and WooCommerce merchants on the first-party channels are generally handled by their providers while custom integrations and file-based setups need action — attributed here because it is a specialist's reading rather than a Google statement)
- JTS Tech Services — Google restated your Merchant Center numbers (the 24 August reporting-definitions change in the same product: what your Merchant Center numbers mean, as opposed to this post's question of whether your data still arrives)
- JTS Tech Services — Google switched on AI visibility reporting for every website (last week's post, and the other half of the picture: the surfaces your catalogue feeds have finally started reporting back, in the same fortnight the supply line into them began degrading)
- JTS Tech Services — Google built a cart that spans every store (why product-feed accuracy stopped being purely a paid-search concern, and what your feed decides about your eligibility across Google's shopping surfaces)
