Amazon Just Set a Deadline to Kill Your Old Ads
Amazon has now attached a real deadline to a transition it's been quietly running since January. According to Amazon's own announcement, confirmed through third-party ad management platform Metoda, the legacy Sponsored Brands Product Collection ad format will shut off entirely in January 2027, with a meaningful interim step arriving this month.
If you've been running Product Collection ads without much urgency about the newer Collections format, that grace period is now on the clock.
The Two Dates That Actually Matter
Starting this month, any newly created Sponsored Brands Product Collection campaign gets automatically converted into a Manual Collection instead. If you submit a custom image with a new campaign during this window, Amazon will still accept the upload, but it won't actually render in the ad. That's a subtle trap worth knowing about: you can go through the motions of building a legacy-style ad and not realize the creative work didn't carry through until you check live performance.
The second date is the real deadline. Amazon shuts off the Product Collection format completely in January 2027. After that point, Product Collection ad creation requests won't be supported at all, through the interface or through the API. Existing campaigns will keep serving impressions and can still be managed in the meantime, so nothing breaks overnight, but the format itself has a confirmed expiration date now rather than an open-ended “eventually” the way it's been described since the format first changed back in January.
What's Actually Replacing It
The new Sponsored Brands Collections format gives you meaningfully more room than the legacy version. Where Product Collection ads maxed out at three products with a custom headline and brand logo, Collections supports up to 10 related ASINs, split between two distinct modes.
Manual Collections put you in control: you select three to 10 products yourself, and any Sponsored Brands targeting type is available. Automatic Collections work differently. You set keyword targets, and Amazon's own system selects which products from your catalog it considers most relevant to those keywords and to the actual queries shoppers are typing, pulling from your existing catalog data rather than anything you build manually. You can exclude specific ASINs from an Automatic Collection if you don't want certain products surfaced, but you don't get to hand-pick what does show up.
The creative rules changed too. Lifestyle images are gone entirely under the new format. Brand logos are now optional rather than required. Automatic Collections generate their own landing pages without your input, while Manual Collections let you choose between an automatically generated title or one you write yourself.
What You Should Actually Do Before January
If you manage campaigns through bulk import tools, download the updated Campaigns bulk import template before your next import. The template now includes the new Collections-specific fields, and using an outdated template is a straightforward way to have an import silently fail or behave unexpectedly.
If you use the Amazon Ads API directly, note that Product Collection endpoints keep working until the January 2027 shutoff, but you'll eventually need to migrate to the corresponding Manual or Automatic Collection endpoints. Amazon has said migration tooling for moving existing ads over is expected to become available, though a specific release date for that tooling hasn't been confirmed yet.
The most useful thing to do right now is treat this the same way you'd treat any format your account depends on losing its runway: audit which of your active campaigns still use the legacy Product Collection format, and start testing Manual and Automatic Collections in parallel well before January, rather than waiting for the shutoff to force a scramble. Given that Automatic Collections hand Amazon's own system control over which products get shown, run a real side-by-side test against Manual Collections on a few campaigns to see whether ceding that control actually helps or hurts your specific catalog before you commit to one mode over the other at scale.
