When people rework a site, “apparently most traffic is mobile now” often gets used as the reason. But that figure is an average across the whole web, not a figure about your site. And your figure has already been recorded somewhere — the entire point of this article is to go and open it.
Why the average is not your answer
Visitor mix varies enormously with the kind of business and the route people arrive by. A B2B site found by someone searching at their desk during working hours, and a neighbourhood shop reached from an Instagram link, are the same web on completely different screens. A site people reach via a QR code on a business card is effectively all mobile.
So deciding from an average can be wrong in either direction: deprioritising desktop on a site where mobile is genuinely a minority, or continuing to check everything on a wide screen when almost everyone arrives on a phone. The second is more common and much quieter, because the person building the site is always sitting in front of a large monitor.
What to look at
The third row is the important one. The device mix of visits and the device mix of conversions are often different. Browsing on a phone and enquiring later from a laptop is a common pattern, and so is the reverse. Concentrate on one side based on visit share alone and you may neglect the side that actually produces revenue.
Splitting by acquisition channel earns its keep too. People arriving from search and people arriving from a social link differ in device and in how long they stay, and that difference is a reasonable basis for deciding which channel to grow.
Three places to find it
If you have an analytics tool installed, it almost certainly includes a device, browser and screen resolution report already. That is the first place to look.
The second is your search engine webmaster tools, where performance can be split by device. It is useful for looking at search visitors specifically, and it accumulates data as soon as the site is registered even if analytics is not installed yet.
The third is server access logs. Slightly more effort, but they include visitors whose browsers block analytics scripts. Anyone using ad blocking or tracking prevention is invisible to analytics, so where the two disagree, the logs are closer to reality.
What to change once you have looked
The highest-value action is simple: pick your most common device and go through the enquiry flow at that width yourself. Not just glancing at whether the layout holds, but filling in the form and submitting it. The problems that surface — a button that will not take a tap, a field hidden by the keyboard, text you have to pinch to read — are the real causes of drop-off.
Two cautions. Do not draw conclusions from percentages when the sample is small; a share calculated on a handful of daily visits will look different next week. And filter out automated traffic and your own visits, or the whole picture wobbles — most analytics tools have an exclusion setting for exactly this.
Checking across screen widths and measuring performance is covered in the Performance archive, and growing the visitors who arrive from search is the subject of the content that brings customers from search series. To have improvements made and verified on real device conditions, our optimization program includes before and after measurement.