01 — What happened

Two of the largest advertising platforms moved the same way inside a fortnight.

On 15 October, Google began rolling out Universal Cart to users in the United States, across Google Search — including AI Mode — and the Gemini app. It holds items from several stores at once, keeps them when the shopper returns, and alerts them to price drops and restocks.

Checkout runs two ways: the shopper pays inside the cart with Google Pay, or the cart is handed to the merchant’s own site to finish there. Both run on the Universal Commerce Protocol, which Google co-developed with retailers this year. Nike, Sephora, Target, Walmart, Wayfair and many Shopify merchants are named at launch, and Google is plain about the arrangement: “No matter which way you buy, the brand stays the merchant of record.”

Meta moved on the page before checkout. Around 14 October advertisers began receiving a notice headed: “Starting October 28, 2026, we’ll add product browsing to your existing ads.” From that date, tapping an affected Facebook or Instagram ad may open a full-screen page inside the app — product images, prices, descriptions and suggested items assembled by Meta — rather than going straight to the advertiser’s website.

The feature is not new; it has existed as an opt-in Advantage+ creative enhancement. What changes is the default. Meta’s justification is an internal experiment run from 6 to 21 August across roughly 117,700 advertisers in “3+ verticals”, reporting 3% more conversions on average at better than 99% confidence. The notice does not define which conversions were counted, describe the comparison, or state any effect on cost or return on ad spend. There are two ways out: the notice’s “Opt out” button, which works only before the date, and a per-ad toggle in Advantage+ Creative and Destination settings.

Nor does it say whether ads created after 28 October will default to on, or where conversions are recorded. As of 15 October, PPC Land could find no Meta Help Centre article or newsroom post describing the change. Advertisers got fourteen days’ notice, thirty days before Black Friday. David Herrmann, president of Herrmann Digital, called the timing “downright diabolical”.

02 — Why it matters

Strip away the product names and both moves are the same move: a page the platform owns now sits between the moment of intent and your website.

It is not a block. The shopper can still continue to your site, and Google’s cart can still hand the basket over for you to complete. The visitor is not taken away — they are delayed, filtered, and shown adjacent products you did not choose. Meta’s notice does not say whose products those are, which is a question worth asking rather than an accusation worth making.

What this costs you is not the sale. It is the record. A visitor landing on your website gives you a measurable session, a path you designed, a chance to capture an email address, and a row in your own data. One who resolves everything inside Facebook, Instagram or Google gives you an order, if you are lucky, and nothing underneath it.

And nobody is being asked. This is a default with a date on it, and defaults are not decided — they are drifted into. The same shape appeared twice this autumn: Shopify switching agentic checkout on for eligible merchants, and Google and Microsoft retiring keyword-level control with no opt-out at all. The platform changes the default, publishes an average, and leaves the specific case to you.

03 — The economics

You pay the same for the click. What the click delivers changes.

Take an illustrative retailer buying 10,000 ad taps a month. Today every tap lands on the website. Assume 2.5% of sessions convert and 6% produce a usable email address — through checkout, an account, or a capture on the page. That is 250 orders and about 600 contactable people a month.

Now apply the change. Assume Meta’s average holds and conversions rise 3%: 258 orders. Assume too that 35% of taps resolve inside the app, so sessions fall to 6,500 and captures fall with them, to roughly 390.

Eight more orders this month. Two hundred and ten fewer people you can contact next month. Those rates are assumptions, not findings — the ratio is the point.

The two sides are not measured on the same clock. The eight orders appear inside the platform’s attribution window, in the dashboard you open every morning. The two hundred and ten appear nowhere. They surface, if at all, as a flat line in repeat revenue nine months later. Businesses weigh this badly not through carelessness, but because one number is reported and the other is not.

Whether the trade is sensible depends on where your margin comes from. If your economics is a single transaction at a known margin, a 3% in-window lift for a thinner data trail is a reasonable deal. If your margin arrives on the second and third order, or starts with an enquiry somebody has to follow up, you are exchanging a durable asset for a one-off gain. That calculation needs your own repeat rate, not a quoted benchmark — the retention statistics everyone quotes are from 1990.

Meanwhile your own measurement gets noisier. Fewer taps reaching the site moves landing-page analytics, consent flows and post-click tests at once, and neither platform has said how a handed-over cart is attributed. The less a platform sees after the click, the more it leans on what you tell it — conversion feedback is now the main lever you hold.

04 — What businesses should do

1. Before 28 October, open every ad account and look. The notice does not name the account it applies to, and the opt-out button is not confirmed as covering a whole portfolio. Check account by account rather than trusting one click.

2. Answer one number before deciding. What share of your gross margin comes from customers’ second and later purchases, or from enquiries your team follows up? If you can answer that, this is arithmetic. If you cannot, that is the finding, and a bigger one than this change.

3. If your revenue depends on a captured record, opt out for now. Quotes, repeat orders, subscriptions, service bookings — anything with a follow-up sequence behind it. Trade the season on a path you control and revisit in January, when Meta has documented the feature.

4. If you opt in, keep a holdout. The per-ad toggle makes a split possible: leave it off on a meaningful slice of spend and compare. A 3% average across 117,700 advertisers says nothing about your account, and a result read off Black Friday week says nothing at all.

5. Move capture earlier than the landing page. If arrival on your site is no longer guaranteed, capture the email address at the earliest moment you still own.

6. On the Google side there is no switch, so do the unglamorous work. Universal Cart reads your product data, so price, availability and shipping accuracy now sit in front of the shopper directly — and a handed-over cart arriving unattributed in your analytics makes a working channel look idle.

05 — Where technology fits

For a large number of businesses the honest answer is nothing at all. If you do not buy paid social and do not sell products that appear in Google Shopping, this is industry news. Check it against your own order sources, then move on.

For the businesses it does touch, the answer is not AI. It is three things.

A conversion path on your own site worth landing on. Not a redesign — the specific pages paid traffic arrives at, and whether anything on them creates a record. If all a visitor can do there is buy or leave, the platform’s in-app panel is a better version of your page.

A customer record created at the first identifiable moment. The caveat from our own archive holds — a CRM is not a growth strategy, it is where the strategy gets written down. What matters is that the record exists, carries its source and is reachable: ordinary connected-systems work, not a platform purchase.

An integration sending outcomes back. Orders, qualified enquiries, cancellations and refunds returned to the ad platforms, so optimisation runs on your definition of a good customer rather than a cheap click.

What not to buy is a dashboard. No attribution tool reconstructs a record that was never created.

06 — RevenueStack’s take

Our view: the development is not the interstitial page. It is that 3% turned out to be enough.

Both platforms have made the same offer — a small, measurable, in-window gain in exchange for a slice of the customer relationship — and both are probably right that most advertisers will take it. Not because advertisers ran the numbers and agreed, but because one side of the trade is printed in a dashboard every morning and the other is reported nowhere. That asymmetry is not a side effect of the design. It is the design.

So the question is not whether product browsing is good. The businesses that come out of this well will not be the ones who opted out fastest. They will be the ones who already knew what a customer record was worth to them — the one number that makes this trade decidable, in October and at whatever the next default turns out to be.

Most do not know it. Which is precisely why, on 28 October, the default will decide it for them.

Sources

  1. Meta will turn on product browsing in existing ads on October 28 — PPC Land, 2026-10-15
  2. Performance Marketing Insider: Meta Wants To Take Performance Through Purchase — MediaPost, 2026-10-16
  3. Google's Universal Cart reaches US search and Gemini but not YouTube — PPC Land, 2026-10-15
  4. Google's Universal Cart reaches US shoppers as Meta flips browsing on — PPC Land, 2026-10-16
  5. Google's Universal Cart shopping tool launches in the US — TechCrunch, 2026-10-15
  6. Introducing the Universal Cart and more ways to help you shop — Google, 2026-05-19