Why Is Programmatic Advertising on Safari Still So Difficult in 2026? ðŸ§
Here in France, Safari is an advertising paradox: it represents 27% of mobile sessions in the Statcounter figure cited in our analysis, yet just 10% of mobile web advertising investment according to the Alliance Digitale figure we referenced. Why is it still so difficult to buy Safari inventory effectively in 2026?
1. Safari's audience share versus its share of advertising spend
Let's start with the numbers that prompted our question. Safari accounts for 27% of mobile sessions (Statcounter, as cited in our original post), but only 10% of mobile web advertising investment in the Alliance Digitale figures we used.
| Indicator | Safari share cited | Source in our analysis |
|---|---|---|
| Mobile sessions | 27% | Statcounter |
| Mobile web advertising investment | 10% | Alliance Digitale |
That is a 17-percentage-point difference. The figures measure different things (sessions and investment, not identical populations or denominators), so they are a useful signal of imbalance rather than a precise measure of unmonetised Safari impressions.
In other words, Safari is not a niche browser. So why does its advertising footprint look so small?
2. Third-party cookie restrictions are only the beginning
The familiar answer is Apple's restrictions on third-party cookies. Those cookies historically helped advertisers recognise browsers across websites and supported important functions: audience targeting, frequency capping and campaign measurement.
Without a shared identifier, those functions become harder to coordinate across publishers, SSPs, data platforms and DSPs. But difficult does not mean impossible: contextual and semantic targeting, publisher first-party signals and privacy-appropriate identifiers can all play a role.
For background, see our earlier article on how to conquer Safari inventories.
3. Why do alternative IDs not fix Safari buying on their own?
Here's the surprising part. In the Curator data cited in our 2026 analysis, more than half of bid requests already contained an alternative identifier, such as First-ID, ID5, LiveRamp ID or Utiq ID.
So the identifiers exist. Why isn't the problem solved? Because an ID appearing in a bid request is not the same as an ID being usable throughout the buying workflow.
With around 80% of programmatic display investment going through the Open Auction in the Alliance Digitale figure cited in our post, the integration challenge extends across four layers:
| Layer | What needs to work |
|---|---|
| Inventory | Publishers and SSPs must support and pass usable signals. |
| Targeting | Data marketplaces and DMPs must be able to activate compatible audience segments. |
| Frequency | DSPs need appropriate mechanisms to manage exposure across impressions. |
| Measurement | Advertisers need compatible ways to evaluate outcomes. |
In our view, the sell side has made meaningful progress, while other parts of the chain have yet to catch up (actual support varies by identifier, platform, permissions and use case).
This is also why making first-party data flow more smoothly through the programmatic supply chain matters.
4. One striking exception: Google audience segments on Safari
We did observe one encouraging result. In our tests using Google audience segments, approximately 30% of delivered impressions were on Safari.
That is an observation from our own test, not a market-wide benchmark or proof of which underlying Google signals enabled delivery. It does, however, show why it is worth testing actual delivery rather than assuming Safari is inaccessible.
We have explored Google's audience data from several angles, including the unusual size of its audience segments, the signals that may underpin those segments, and where some of those signals may originate.
5. Could publishers become their own data providers?
One possible way forward is for publishers to become more active data providers: packaging their first-party audiences and making those segments directly accessible to media traders in their DSPs.
That could make publisher-controlled signals easier to activate through the Open Auction, where compatible integrations and permissions allow it. We discussed the idea in our earlier LinkedIn post on publisher first-party data.
It would not eliminate every problem. Matching, interoperability, consent, contractual controls, frequency and measurement would still need attention. But it could help move the market away from depending exclusively on third-party cookies.
For a related perspective, read our piece on first-party and 1.5-party data activation.
6. How GreyFox Ads helps media traders reconnect with Safari
Until the whole ecosystem catches up, our approach at GreyFox Ads is pragmatic: combine cookie-based, alternative-ID and semantic targeting in curated programmatic deals, selecting the method appropriate to the inventory and the applicable privacy requirements.
The objective is straightforward: reconquer Safari and avoid concentrating advertising investment on Chrome simply because Chrome is easier to buy. ðŸ§
Our Master Deal also emphasises quality supply paths, relevant inventory, attention and transparency. Reaching Safari users is valuable only if the resulting advertising is effective and responsibly delivered.
Original LinkedIn post: Read the original GreyFox Ads analysis.
Related post: Publisher first-party data and DSP activation.
Looking for a quality programmatic advertising partner?
Want to reach Safari and other Open Web audiences more effectively in France? Let's discuss your Display and Video campaigns.
Contact GreyFox AdsDiscover the Master Deal