How Much Control Do You Keep When Pinterest Performance+ Is On?

More than Pinterest's default-setting pitch implies, but less than most advertisers assume — and which of the four controls you reach for first matters more than whether you reach for them at all. Since March 2026, Pinterest has been telling advertisers to make Performance+ the default way to buy, at "50% fewer required inputs." Since June 2026 it has been quietly handing back the things that removal took away: New Customer Acquisition in June, then Priority Products and self-serve A/B testing at Pinterest Presents on September 17, 2026. The useful question is not automation versus control. It is sequencing — which control earns its place first, and which one only adds noise to a report you were already reading correctly.

What Pinterest automated, and what it handed back

Neither direction is new, and neither is finished. Reading them as one timeline is more informative than reading any single announcement on its own.

One detail that gets lost in the coverage: Pinterest did not automate your creative. Performance+ automates delivery, bidding and targeting — which of your ads gets served. Standard image, video and manually uploaded creative goes out as you built it. The exception is the Sales objective run from a product catalog, where Performance+ Creative is an optional set of toggles that can convert product Pins into shopping and collections formats and generate backgrounds for eligible plain images. Pinterest recommends at least 10 ads per ad group so the system has enough to choose between. That is a materially different posture from Meta's Advantage+ Creative suite, which advertisers keep reporting re-enables itself — the story we wrote about in Meta Advantage+ Is Now the Default.

The four controls, and what each one actually changes

Two of those four are constrained in ways the announcements skip. New Customer Acquisition can only run inside Performance+ campaigns, which means it cannot share a campaign with max-width collections or Top of Search placements. Priority Products has no published cap on how many products you can shortlist and no reporting breakout that shows what the shortlist actually received — details that decide whether it is a lever or a decoration.

Priority Products is a merchandising input, not a performance setting

The feature exists because of a real tension: automation optimizes for what the data already says works, and a business sometimes needs to support a collection that has no history yet. A retailer launching a seasonal line is asking for delivery toward products the model has evidence against. Priority Products is how you say that out loud without dismantling the campaign.

It fails in three predictable ways:

1. You prioritize the bestsellers. The automation was going to find them. You have now paid for demand you already had and learned nothing about whether the promotion moved anything.
2. You prioritize new SKUs with no conversion history. You are asking the model to starve everything it has evidence for. Either it complies and performance drops — which you will read as "Pinterest got worse" — or it quietly ignores the shortlist, and the test is void.
3. You prioritize by margin and ignore stock depth. You promote what you cannot ship, then absorb the return cost of a promotion that worked.

The filter we use has four columns: margin spread (do prices actually differ enough for value optimization to matter), stock depth (can the priority survive a week of delivery), age of conversion history (is there anything to learn from, or only something to seed), and creative readiness (does each priority product have a Pin worth serving). If three of the four are empty, the correct move is not to use Priority Products. It is to fix the catalog.

Note the overlap with the problem we mapped in Pinterest Shopping Bids Work Better as a Mix Than a Setting: in both cases the platform hands you a control that only pays off once your underlying signal is trustworthy. Pinterest is not giving you a shortcut around catalog quality. It is giving you a way to point at catalog quality faster.

Most A/B tests inside Performance+ will measure the algorithm, not your change

This is the announcement we would rank above Priority Products, and the one most likely to be used badly.

Automation reallocates budget between impressions, audiences and creative continuously. If both arms of a test live inside one campaign, the system's own allocation is a confound you cannot subtract afterwards. You will get a winner, and the winner will be partly the algorithm's preference rather than your change.

Four conditions make the new testing tools read cleanly:

• Split at campaign level, with fixed budgets per arm. Not ad set, not ad group. The budget is the treatment; if the system can move it, the treatment is contaminated.
• Hold everything else constant. Test the priority list with creative fixed. Test creative with the priority list fixed. Never both in the same cell — this is the same mistake we see on Meta when an account changes bid, audience and creative in one afternoon and then argues about which one caused the result.
• Read on volume, not calendar days. Pinterest's own guidance of at least 10 ads per ad group means thin arms are the default. A 7-day test on a small catalog is a coin flip with a dashboard.
• One variable per test, with a stated graduation condition before it starts. If you cannot write down what result would make you change your mind, you are not running a test, you are looking for reassurance.

The order we would take them

On a client account, the sequence is not "turn everything on and see."

1. Fix the signal first. Tag plus Conversions API, with purchase values actually arriving. Every control below optimizes toward whatever signal you are feeding it, so a broken feed makes all four worse.
2. New Customer Acquisition, if there is a second purchase. It is the only lever that changes who you are buying, not just what you are showing, and it is scoped to Performance+ where it cannot be diluted.
3. Priority Products, but only against a dated commercial moment. Give it a defined end date so you can compare against the period before it. An evergreen shortlist is a setting, not a decision.
4. A/B testing last, as the verification layer. It does not improve performance on its own. It is how you find out whether steps 2 and 3 were real — and it is the only one of the four that can tell you you were wrong.

That ordering also answers the question in the title. You do not keep control by having more toggles available. You keep it by knowing which single toggle belongs to the decision you are actually making.

What is still unknown

Pinterest has published no pricing, no eligibility criteria and no independent performance data for the September cohort. The App Promotion figure doing the rounds — up to 85% improvement in cost per install — comes from one U.S. enterprise advertiser over an alpha test with a 30/1 attribution window. That is a case study, not a benchmark, and it is not a reason to change objective. Priority Products and self-serve A/B testing are both described as rolling out in the coming weeks with no per-market schedule. Until there is a reporting breakout that shows what a shortlist received, the honest position is the one we took on Visual Search Ads: treat it as a framework decision, not a headline decision, and let a structured test settle it.

Sources: Pinterest, "Pinterest unveils Visual Search Ads, bringing new performance advertising to visual discovery" (September 17, 2026); Business Wire, "Pinterest Launches Visual Search Ads to Expand Performance Advertising Across Visual Discovery" (September 17, 2026); Digiday, "Pinterest unveils new suite of ad tools, including Visual Search Ads" (September 21, 2026); Mary Lumley, "Pinterest Ad Specs 2026" (updated September 2026); Brainlabs, "Pinterest Presents 2026: Reaching Shoppers Through Visual Search"; Pinterest Business, "Make Pinterest Performance+ Campaigns Your New Default".