iOS 11 Is Out and Safari Now Limits Cross-Site Tracking

Released yesterday

Apple released iOS 11 yesterday, and with it Safari 11. Among the changes is Intelligent Tracking Prevention, announced in June: Safari now limits the cookies companies use to follow people from site to site.

iPhones and iPads prompt their owners to update, and many will within weeks. Macs get the same Safari through their own updates this fall.

For most businesses the question is what this does to the numbers used to judge advertising. The short answer: your own analytics should look the same, and some ad reports will not.

Whose cookie it is

A cookie is tied to the domain that created it, and no other domain can read it.

Cookies set by your own pages belong to your domain. They are first-party cookies, and Universal Analytics keeps one there to recognise a returning browser.

Your pages may also load code from ad platforms and social networks, which set cookies on their own domains. Those are third-party cookies, and because the same code sits on thousands of sites, they can recognise one browser everywhere.

Safari 11 is aimed at that second kind.

What Safari 11 does

• It classifies domains that appear as third parties across many unrelated sites.
• For a day after a person visits such a domain directly, its cookies work normally.
• After that, its cookies are kept separately for each site, so visits cannot be connected.
• After thirty days without a direct visit, its cookies and data are deleted.
• First-party cookies on the site being visited are left alone.

The second is the one that shapes advertising. Few people ever visit an ad platform's tracking domain directly, so its cookies soon lose their reach across sites.

What should look the same

Universal Analytics sets its cookie on your own domain, so Safari 11 leaves it alone. Visits, sources, pages, goals and returning visitors in Analytics should look as they did last week.

Campaign tags on your links, and the click identifier AdWords adds to its ads when auto-tagging is on, travel in the page address itself. Analytics reads them on arrival, without needing anyone else's cookie.

Nothing on your site breaks, and nobody sees a warning. The change happens out of sight, in what other companies can count.

Where the numbers move

Ad platforms count conversions with their own cookies: someone clicks an ad, the platform remembers the browser, and a later booking on your site is credited to the ad.

In Safari 11, a booking made more than a day after the click may never be credited, because the platform's cookie can no longer connect the two.

Conversions from people who saw an ad without clicking will largely disappear for Safari users. Remarketing lists, the audiences that let ads follow past visitors, will hold fewer of them and lose them sooner.

Why the delay matters

The one-day window is generous for a pizza ordered twenty minutes after a click. It is tight for a holiday booked over a fortnight, a renovation planned for a month, or a course chosen after comparing three.

The longer customers take to decide, the further an ad platform's report drifts from what really happened.

Businesses selling considered purchases to iPhone users are likely to see the biggest change, and it will look like a drop in results when it is a drop in counting.

A worked example

A seaplane tour company flies scenic routes from a harbour, selling mostly to visitors who plan days or weeks ahead. It advertises on AdWords, runs remarketing, and takes bookings on its own website.

A typical customer clicks an ad on an iPhone while planning a trip, reads about the flights, leaves, and books from a hotel room several days later, often after a remarketing ad.

Under Safari 11, that booking may never be connected to the first click in the ad platform's reports, and the customer is likely to slip out of the remarketing list's reach after a day.

The mechanism is the gap between first click and booking. The longer it is, the less of it the platform can see.

So this week the company wrote down what each platform claimed for August, beside the bookings in its own system. It will compare the same figures at the end of October before changing any campaign.

Reading the reports this fall

Put a note in Analytics on September 19, using an annotation, so the change is visible to anyone reading the timeline months from now.

Then see what share of visitors arrive in Safari: the browser report lists it under Audience, then Technology. A small share means small shifts in ad reports; a large one means you should expect them.

Expect conversions reported by ad platforms to fall for Safari users. Before reading that as advertising doing worse, set it beside the bookings, calls or orders you count yourself.

Not every platform loses the same

The one-day window reopens whenever a person visits the platform's own site in Safari. A platform people browse there every day keeps more of its reach than an ad network whose domain nobody visits.

Visits through a platform's app are unlikely to count, since the app is not Safari. Someone who uses a social network only through its app may give that network no window in the browser.

So one platform's figures may fall further than another's, for reasons unrelated to which ads worked. Compare each with your own counts, not with each other.

Expect the platforms to adjust

Ad platforms have every reason to respond, and changes to their tags are likely, such as keeping more information in cookies on the advertiser's own domain.

When they do, sites with old tags may miss out. Ask whoever runs your advertising to list the tags on your site and to update them as the platforms change.

The counter-case

Plenty of businesses will barely notice: those without online advertising, and those whose customers book within hours of clicking.

Those that began counting their own results after Apple's announcement in June already have the comparison they need.

Still, any business deciding on advertising from the platforms' dashboards alone should begin checking those dashboards against its own counts.

The baseline

1. Note September 19 as an annotation in Analytics.
2. See how many visitors arrive in Safari.
3. Pull last month's bookings from your own records.
4. Write down what each platform claimed for the same month.
5. Repeat the comparison at the end of October.
6. Keep your ad tags up to date.
7. Hold campaign changes until the comparison is in.

Step four is the baseline the rest depends on. Without last month's claimed figures written down, nobody will be able to tell how much of a fall was Safari and how much was real.

What Apple announced in June, including the video changes coming to Macs, is covered in Safari will block autoplay video and limit tracking.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.