Apple blocks another ads route for brands

An adtech vendor can identify a customer and still have done nothing to earn the sale. Apple’s WebKit 27 block puts a price on that distinction. Safari on iOS, iPadOS, macOS and visionOS will block network access to listed identity-resolution infrastructure. The target is the request route used to match an email collected on one site to an identity used elsewhere. WebKit’s announcement says first-party personalization on the site itself remains available.

That distinction matters because customer data can do useful work. A CRM email can bring back a grocery customer who is due to reorder. A purchase signal can help an ad platform find more people likely to buy. A cross-site resolver can make an impression appear to reach a known person. Those activities carry different costs and produce different evidence. The spend should follow what changed in the customer’s behaviour.

Better matching can find customers, and still take credit for them

In a randomized study of more than 70,000 Meta advertisers, researchers compared purchase-optimized campaigns using offsite conversion data with campaigns optimized for clicks. They estimated a $38.16 median cost per incremental customer and calculated that losing purchase optimization would raise it to $49.93, a 31% increase. The study, whose authors include Meta researchers, gives a marketer a commercial reason to keep useful purchase signals.

An extra $11.77 for each additional customer changes how far an acquisition budget goes. A business selling a low-margin product cannot absorb that indefinitely. Losing a purchase signal can mean fewer customers for the same spend, even while the platform continues to deliver clicks. Replacing a sales target with a traffic target makes the campaign easier to fulfil and the business harder to fund.

eBay’s large randomized search-ad experiments show why attribution alone cannot earn it. The company tested paid search ads against holdouts. Brand-keyword ads produced no measurable short-term benefit in the tested setting, because many people searching for eBay were already likely to visit. Nonbrand ads had value among new or infrequent buyers. The researchers’ paper found that the same advertising channel could acquire new demand and subsidize existing demand, depending on who was reached.

Take a grocery brand with a regular customer due to restock. The CRM sends a replenishment offer and a display campaign also reaches that person. The receipt lands after both contacts. A resolver can claim a match, while the sale may have come from the email or the customer’s existing habit. For a new shopper who had never bought from the brand, purchase data that helps an ad system find them has a different value. The acquisition budget has to return additional customers after media and data costs. A percentage of matched records cannot pay those bills.

Privacy changes move money between platforms

Apple’s earlier App Tracking Transparency change offers a measured example of the commercial effect. A May 2025 CESifo study using advertiser and firm data estimates conversion-optimized Meta ad click-through rates fell about 37% relative to click-optimized campaigns after ATT. Meta’s share of sampled ad spend fell, with most of the shift going to Google. Across specifications, the study estimates relative firm-revenue reductions from roughly 8% to 40% among more exposed e-commerce firms, with smaller firms hit harder. The paper reports a range because exposure and treatment varied across firms.

ATT changed app tracking across iOS. WebKit targets listed identity-resolution domains in the browser. Both changes affect where a brand can put its customer signals to work. In the ATT study, Google gained much of the budget leaving Meta. Advertisers responded to weaker targeting by moving money to another large platform. The brand’s dependence moved with the spend.

This is why “first-party data” is an incomplete budget label. Google Customer Match, for example, lets an advertiser upload a customer list for matching against Google account records and use the eligible audience within Google’s products. Google’s documentation describes that platform-held match. The advertiser supplies the file. The platform controls the account environment where the matched audience can be activated. A third-party resolver connecting identities across publisher sites sells a different route. Moving spend from one to the other can reduce a browser dependency while deepening dependence on a platform’s own audience system.

Pay for the sale the signal creates

The brand pays once to acquire a customer, then keeps paying intermediaries to recognise that customer on later visits. Some of those contacts cause the next purchase. Others arrive near a sale the customer had already decided to make. When every supplier claims the order, the plan can show several successful campaigns against the same basket. The business has one sale to pay for all of them.

That is where I would challenge an identity premium. A resolved email and a higher match rate are inputs the supplier sells. The return has to appear in additional purchases after the cost of the data and the media. If removing a matching layer leaves orders intact, the vendor has lost its reason to charge for that layer. If losing a purchase signal makes acquisition more expensive, cutting it to save a data fee can cost the business more than it saves.

Apple’s block leaves the CRM email intact and puts the cross-site route under pressure. The vendor that creates the next sale keeps its place in the plan. The one selling a better match to existing demand loses the line item.