Tips
What website traffic is and how to read it
A visit count on its own tells you nothing. Three numbers next to it turn traffic into something you can act on, and one common metric is worth less than people think.
Correction, 23 September 2026: GA4 bounce rate now uses its engagement definition. The earlier single-page definition did not apply to GA4.
A visit count on its own is not information. Traffic went up is a fact with no decision attached to it. Reporting it is how a channel that produces nothing survives for a year.
Three numbers next to it turn traffic into something you can act on.
The three that matter
Where it came from. Organic search, paid, direct, referral, social, email. The same thousand visits mean completely different things depending on the split. A change in the split is usually more meaningful than a change in the total.
What it did. Enquiries, sign-ups, purchases, downloads. Pick the one action that matters to your business and count it. If you cannot name the action, that is the first problem to fix.
Which pages received it. Traffic is not distributed evenly. Usually a small number of pages carry most of it, and knowing which ones tells you what to write next and what to protect.
What the common metrics actually measure
Sessions count visits, not people. One person who comes back three times is three sessions.
Users are estimates under the tool’s identity rules. A person on two devices can be counted separately unless the setup can connect them.
Pageviews count page loads, and go up when your navigation is confusing.
Bounce rate depends on the product’s definition. GA4 measures sessions that were not engaged. A single-page visit can be engaged, so it is not automatically a bounce.
Average position in Search Console averages across every query and every impression. It moves for reasons that have nothing to do with your ranking for the query you care about. Look at per-query positions.
How to read a change
Traffic doubled. Before celebrating, check three things.
Did one page produce it? A single post picked up by an aggregator produces a spike that never repeats and converts at nothing.
Did the source change? A jump in direct traffic with no campaign usually means tracking broke or a redirect stripped the parameters.
Did conversions move with it? If visits doubled and enquiries did not, you bought the wrong visits.
The same three questions apply to a drop, in reverse.
Where the numbers disagree
Analytics, Search Console, ad platforms and server logs measure different events. Their totals need not match.
Search Console defines impressions, clicks and position for its own reports. Analytics counts what the browser reported after a script ran, which can miss blocked requests or visits without analytics consent. The ad platform counts a conversion within its own attribution window and will claim credit that another channel also claims.
Choose a reporting baseline for each question and document its coverage. Server logs include automated traffic and do not measure every in-page action.
What to actually watch weekly
Conversions, by source. Organic impressions on the pages you worked on. The top ten landing pages and whether the list is changing.
That is a five-minute review and it is worth more than a twenty-tab dashboard nobody opens.
If your analytics tell you less than that today, the fix is usually measurement setup, which is part of how we run conversion work.