Quick answer: how did privacy changes affect AI ads?
Since iOS 14.5, apps need permission to track people across other apps, so platforms see fewer conversions directly. Chrome kept third-party cookie choice, but browsers still block much tracking. Platforms responded with AI modelling and server-side data. Your best move is clean, consented first-party data sent through server events.
What actually changed
Ad platforms used to follow people across apps and websites with device identifiers and third-party cookies. Two shifts reduced that. For how the platforms decide who sees an ad, see how AI ads work on Meta and Google.
Apple's App Tracking Transparency
Apple's developer rules say that in iOS 14.5 and later, apps need the user's permission through App Tracking Transparency to track them or access the device's advertising identifier. Without permission, the identifier is all zeros. Many people decline, so platforms lost a large share of direct iPhone tracking.
The cookie debate
Google spent years planning to remove third-party cookies from Chrome. In an October 2025 update, Google confirmed Chrome will keep its current approach of offering users third-party cookie choice, and announced it is retiring several Privacy Sandbox technologies, including the Attribution Reporting API, Protected Audience and Topics.
So is tracking back?
No. Safari and Firefox already limit cross-site tracking, ad blockers are common, many users opt out of app tracking, and privacy laws keep tightening. The direction is still toward less third-party tracking and more first-party data.
How ad platforms adapted
More modelling
When platforms cannot observe a conversion directly, they estimate it from patterns in the data they can see. Reported results now mix observed and modelled conversions.
More first-party data
Platforms ask advertisers to send data from their own systems: server-side purchase events, hashed customer emails and phone numbers, and offline conversions.
More automation
Broad targeting and automated campaigns rely less on tracking individuals across sites and more on aggregated signals and conversion feedback. That is one reason Advantage+ and Performance Max became the default.
Aggregated reporting
Some results now arrive with delays or only as totals rather than person-level detail, especially for app campaigns and iPhone users. Short attribution windows and day-by-day numbers are less reliable than they used to be, so weekly and monthly views give a truer picture.
What this means for advertisers
The AI is only as good as the data you send. Two stores with the same products and budgets can see very different results simply because one sends clean server-side data and the other relies on a browser pixel alone.
Before and after: what the platforms can see
| Signal | Before | Now |
|---|---|---|
| iPhone app tracking | Advertising identifier widely available | Only with ATT permission |
| Browser pixel on your site | Most purchases captured | Some lost to browsers and blockers |
| Server-side events | Rarely used | Recommended alongside the pixel |
| Hashed customer data | Optional extra | Core to matching conversions |
| Reported conversions | Mostly observed | Observed plus modelled |
Why reported numbers differ from your store
Platform reports now include modelled conversions and their own attribution windows. Your store counts actual orders. Expect a gap, track its size month to month, and investigate when it suddenly grows. A sudden jump usually means a broken tag, a checkout change or a new app version, not a change in customer behaviour. Our guide to measurement and attribution after cookies and iOS covers attribution models in depth.
What to do now: a first-party data checklist
- Send purchases from your server. Meta recommends the Conversions API alongside the pixel with deduplication, so each purchase counts once and fewer are lost.
- Turn on enhanced conversions in Google Ads. Enhanced conversions send hashed first-party data such as email with each conversion to improve matching.
- Pass customer details you are allowed to share, hashed, with each event: email and phone improve match rates.
- Upload offline outcomes, such as delivered cash-on-delivery orders, so the AI learns from orders that actually paid.
- Ask for consent clearly and keep your privacy policy accurate about what you share with ad platforms.
Our step-by-step guide to pixel, CAPI and store data walks through the setup on common store platforms.
A worked example: the cost of a data gap
Worked example: a beauty brand on a browser pixel only
The gap
The store records 1,000 orders from paid social in a month. Meta's pixel reports 700 purchases, so 30% of real purchases are missing from what the AI learns from.
After adding server events
With the Conversions API and customer details sent, Meta reports 900 purchases, closing two thirds of the gap.
Signal coverage = platform-reported purchases ÷ store orders from that channel
Before: 700 ÷ 1,000 = 70%. After: 900 ÷ 1,000 = 90%.
The AI now sees 200 more buyers a month to learn from. Even with no change in budget or creative, delivery usually improves as the model gets a fuller picture of who buys.
Note that higher coverage can also make reported ROAS rise without any extra sales. Judge real gains with your store data: orders, new customers and profit after ad spend over the following month.
Consent and the law in India
India's Digital Personal Data Protection Act, 2023 sets rules on collecting and using personal data, including consent. Ad platforms' own terms also require you to have the right to share customer data with them.
Practical steps
Tell customers in plain words what data you collect and that you use it for advertising and measurement. Share only what you need, hashed where the platform supports it. Keep a record of how consent is collected.
Keep data tidy
Delete customer lists you no longer use from ad accounts, limit who in the team can upload data, and remove former agencies and freelancers from Business Manager and Google Ads access when they stop working with you. Fewer people with access means fewer accidents.
Get proper advice
This is general information, not legal advice. For your specific setup, check with a lawyer who knows Indian data protection rules.
Customers who understand and agree to how their data is used are less likely to complain or block tracking, which keeps your signal cleaner over time.
What most privacy articles tend to get wrong
"Cookies are gone, so tracking is dead"
Chrome kept third-party cookie choice, and first-party tracking on your own site still works. The real losses came from app tracking rules, browsers and blockers.
"Cookies are staying, so nothing changes"
App tracking permission, browser limits and privacy laws still cut signals. First-party data still matters.
"Platform numbers are now useless"
They are estimates, not fiction. Use them to compare campaigns and your store data to judge profit. For the full operating plan, see our complete guide to AI ads.
Privacy rules and browser plans change often. The facts here were checked in November 2026 against Apple and Google's own pages.
Read next in this series
Key takeaways
- Since iOS 14.5, apps need permission through App Tracking Transparency to track users or access the advertising identifier.
- In 2025 Google said Chrome will keep offering users third-party cookie choice, and retired several Privacy Sandbox APIs.
- Platforms now fill gaps with modelled conversions, which work better with more first-party data.
- Server-side events (Meta Conversions API) and enhanced conversions on Google recover many lost signals.
- Consent and clear privacy notices are part of good data, not just compliance.
Ai ads: frequently asked questions
What did iOS 14.5 change for advertisers?
Are third-party cookies going away in Chrome?
How can I recover lost conversion data?
Why do Meta and my store show different numbers?
Does iPhone tracking loss affect website ads too?
Do I need customer consent to share data with ad platforms?
Sources
- Apple Developer: User privacy and data use
- Privacy Sandbox: Update on plans for Privacy Sandbox technologies (Oct 2025)
- Meta Business Help: Conversions API best practices
- Google Ads Help: About enhanced conversions
Let ScaleWin read your ad data every day
ScaleWin checks your Meta and Google Ads daily, judges every ad on profit and tells you exactly what to change.
Start free, no card



