A third of link-in-bio traffic has no source at all

We looked at where 16,262 filtered page views across 470 link-in-bio pages actually came from. Instagram was the largest identifiable source at 43.5%. The second largest entry in the table is not a source at all: 36.3% of visits arrive with no referrer, which means nothing records where they came from.

That second number is the more useful one, and it is the one nobody publishes.

The table

Window: 16 July – 14 September 2026. Linkhiver pages only, demo accounts excluded, same bot filtering as our click benchmark.

What we are not telling you, and why

The tempting headline here is "TikTok drives under 1% of link-in-bio traffic". We are not publishing that, because we cannot support it.

TikTok's in-app browser suppresses the referrer. Traffic from a TikTok bio would therefore arrive in the direct bucket and be indistinguishable from everything else in it. The 0.9% figure is real and it means something other than what it appears to mean, which is the most dangerous kind of statistic.

We tested the direct bucket rather than guessing. It is 68.9% mobile and 29.7% desktop — against Instagram's 2.2% desktop. So it is not a hidden TikTok channel; it is a mixture of typed URLs, WhatsApp shares, QR scans, direct messages and some residual automated traffic. That test does not resolve the TikTok question either way, which is precisely why we are not answering it.

The finding: a third of the traffic is dark

Every one of these arrives with no referrer, and none of them can be told apart afterwards:

• a link pasted into a WhatsApp message
• a QR code scanned off a menu, a flyer or a business card
• a URL typed or pasted from a direct message
• a bookmark, or a link opened from a notes app

They land in one undifferentiated bucket that is 36.3% of all traffic — bigger than Facebook, Google, X, TikTok and YouTube combined, which together are 17.2%.

This is a category-wide measurement problem, not a Linkhiver one. Any link-in-bio analytics dashboard, ours included, is reporting confidently on roughly two thirds of what happens and quietly guessing at the rest. If your numbers have never added up, this is why.

It also converts better than anything except Google: 59.7 clicks per 100 views, above Instagram's 50.8. Whatever that traffic is, it arrives more deliberate than social traffic does — which is what you would expect from someone who scanned a code or followed a link a friend sent them personally.

Two smaller things worth stating

Meta is more than half of everything identifiable. Instagram and Facebook together are 53.5% of all traffic and 84% of the traffic we can name. For most creators on our platform, "social traffic" means Meta.

Google referrals convert at 90.5 clicks per 100 views, far above any social source. The sample is small — 581 views — and we are stating it with that number attached rather than rounding it into a headline.

What we cannot tell you

• Views are page loads, not people. No session identifier exists in this data.
• The direct bucket cannot be decomposed. That is the whole finding, and it applies to our own reporting too.
• Bot filtering is behavioural. We remove profile-days where views spike to at least 200 and at least 20× that page's ordinary day. It cannot see steady low-volume automated traffic, and some of that will sit in the direct bucket.
• This describes Linkhiver pages, not the category. Our user base is weighted towards Brazil and Turkey, and an Instagram-heavy audience produces an Instagram-heavy table. A tool whose users are mostly on TikTok would report something different.

Method

One query against our analytics table over the window above, bucketing the referrer field, excluding demo accounts and suspected automated traffic. Sources are grouped by domain match; "other" is every referrer that did not match a known platform.

Data collected 16 July – 14 September 2026. Published 18 September 2026. Free to cite with attribution.