Where 57 Million Traffic Sessions Actually Came From

The 57 million session study covers volume, reliability and the duration distribution. This is the part you need when you are deciding where to run a campaign: which devices, which countries, and what quality to expect in each.

All figures come from our own delivery ledger, delivered sessions only. Windows: 7 days for the country table, 3 days (5.93 million sessions) for the device split and the device quality metrics, both ending 12 November 2026, 21:21 UTC.

Device split, with quality attached#

Two things in that table matter more than the split itself.

First: desktop bounces almost four times as often as mobile - 22.39% against 5.85% - and only 54.5% of desktop sessions last five minutes or more, against 90.15% on mobile. If you are buying traffic for a desktop audience, mobile-heavy delivery will not look like what you bought. You will optimise the wrong thing.

Second: desktop sessions go deeper. 1.32 pages per session on desktop against 1.08 on mobile. The two behaviours are not the same traffic: desktop leaves more often and averages shorter visits, but opens more pages; mobile bounces less and stays longer.

Country distribution (7 days, top 25)#

In total, our 30-day window covered 205 countries. The table above is the 7-day window, because that is the one we can show in full.

The part that is worth arguing about#

The list above is not the usual "Western traffic" shape. In the same seven-day window, South Africa (686,962), Poland (614,031) and Turkey (591,707) each delivered more sessions than Australia (570,907), Germany (563,698) or Russia (548,405) - and Ukraine (449,187) was not far behind Italy (504,620). If your market is outside the obvious five, the cheapest volume is usually not where the instinct says it is - and the reason is supply, not demand: proxy capacity, not audience size.

That is also the honest caveat on this table: we are reporting where our delivery went, not where your customers are. A cheap, well-supplied market is cheap precisely because it is not contested. If your conversion rate in Turkey is a tenth of your rate in Germany, buying Turkish traffic is not a discount - it is a rounding error with a delivery fee attached.

Which is why the second question is always "how do I know the traffic is real enough to convert?" - and the honest answer is that a provider can only show you their side of it. Ask for the duration distribution and for their delivery rate, then run a small pilot with conversion tracking on your side before you scale.

What we would publish next#

This is the first of a monthly series: each month we publish the current window's device, country and delivery figures, with the failures included. The queries are public in the full study, so anyone can check the shape of the analysis.

If you would rather have the numbers for a market we are not measuring - or if you want the same treatment applied to a competitor's claimed figures - write to us and we will send the queries.