When Direct access grows as a share of your traffic, several causes are candidates: referrer policies, tagging issues, third-party tools, exempt mode, bots and more. The fastest way to find the right one is to break the data down along a few dimensions, in this order: sub-type, device and browser, other traffic sources, landing page URLs, recent changes on your side, and bot activity.
Step 1: Break down by sub-type
The three Direct access sub-types have very different causes. Look at the share of each in your data:
|
If most of the increase is in |
The cause is likely |
|---|---|
|
Direct access (continuity of visit) |
Session timeouts during long sessions, typical of long-form readers, dashboards left open and B2B audiences. This is usually harmless, but consider whether the 30-minute timeout suits your audience. |
|
Direct access (out of site) |
Referrer stripping, the dominant case for almost every Direct access spike. Continue with step 2. |
|
Direct access (local) |
Internal testing or development traffic reaching your production reports. You can usually fix it by excluding internal IP addresses. |
Step 2: Break down by device and browser
If the rise is in out of site Direct access, find out which platforms drive it. Break down Direct access traffic by device family and browser:
-
A disproportionate share of iOS Safari: Safari and iOS referrer restrictions.
-
A disproportionate share of Android Chrome: in-app browsers stripping the referrer, typical of social apps that open links in their own browser.
-
A disproportionate share of desktop: less common. It usually points to enterprise referrer policies, ad blockers or browser extensions in a specific audience segment.
If the increase is spread evenly across devices, the cause is probably related to content or campaigns rather than browsers. Continue with step 3.
Step 3: Compare with other traffic sources
Open a time series of Direct access next to your other organic sources. If they move together, you can see which source is being reclassified as Direct access:
-
Direct access rises and Search engines falls: the platforms in the Search engines category have started stripping referrers on some traffic. This is commonly seen with Google AMP, Chrome on Android and Discover-style products.
-
Direct access rises and Generative AI or Referrer sites falls: an AI tool that was recognized has started stripping referrers, or a recognized referrer has changed how its outbound links behave.
-
Direct access rises and email campaign visits fall in
src_medium: email links no longer carry campaign parameters. Check that the tagging in your email platform is intact.
Step 4: Inspect landing page URLs
If the breakdowns haven't isolated the cause, look at the landing page URLs of Direct access visits. Sort them by number of visits and inspect the top URLs:
-
The same untagged campaign URL appears repeatedly: an active campaign doesn't carry
utm_orat_parameters. Re-tag the campaign at its source. -
Specific URLs are missing the Piano Analytics tag: there's a tagging gap. Use Tag Inspector to confirm the SDK fires on those pages.
-
URLs include a redirect path: the redirect strips the referrer. Confirm that the redirect chain keeps the referrer header.
Step 5: Check recent changes on your side
If the breakdowns don't point to an external cause, something you did may have introduced the spike:
-
Front-end deployment: a change to the single-page app navigation, the service worker configuration or the Piano Analytics SDK integration can affect how the first event of a visit fires and which referrer it carries.
-
New third-party tool: a chat widget, consent management platform, A/B test framework or other tool added to your pages can change navigation or referrer behavior.
-
Site migration: a change of domain or URL structure without an updated Main URL in Piano Analytics can cause cross-site visits to be misclassified.
Compare the start date of the spike with your deployment history. If they match, work with your front-end team to check that page.display still fires correctly, with the expected referrer, on the new flow.
Step 6: Check for bot activity
If the spike is large and sudden and none of the above explains it, check whether the traffic looks like bots:
-
A disproportionate number of single-page visits, with a high bounce rate and a visit duration under 2 seconds.
-
A concentration in geographies or IP ranges that aren't typical for your audience.
-
Unusual user-agent strings in the visit data.
If bots are responsible, filter them out with Piano Analytics' bot detection and exclusion.