Platform

SEO forecasting from your own baseline

Forecasting and Experiments

A forecast nobody can check is a wish. Here the SEO traffic forecast is built from your data, states its assumptions, and is later compared against what actually happened.

Tools in this part of the platform

How the forecast is built

Take the baseline: current positions, impressions and click-through rate for the chosen queries.
Apply a position scenario — what a move into the top three would mean for this cluster.
Convert clicks to sessions and key events using your own GA4 conversion rate, not an industry average.
State the assumptions openly: seasonality, cannibalisation, the CTR curve used and the confidence attached.
Record the change as an experiment, then compare actual against forecast when the period closes.

Why most SEO forecasting templates mislead

A spreadsheet that multiplies search volume by a generic CTR curve produces a confident number and a wrong one. Volume is not impressions, generic curves ignore how crowded your particular result page is, and no template knows that half your clicks were already going to a feature block.

Learning how to forecast SEO growth honestly means forecasting the range, not the point, and saying which assumption would have to break for the number to be wrong.

The value is not the prediction. It is that when the period ends you can say whether the work caused the result, and by how much you were off.

Frequently asked questions

How accurate is an SEO traffic forecast?

Accurate enough to prioritise, not to promise. Ranges built from your own baseline and conversion rate beat any generic template, but demand shifts and algorithm updates stay outside anyone's model.

How do I calculate SEO ROI from it?

Multiply the forecast key events by your own value per event, then subtract the cost of the work. Because the conversion rate comes from your GA4 rather than a benchmark, the result is checkable.

What makes this an experiment rather than a report?

A recorded change date, a test and control group of pages, external factors noted, and a verdict at the end: the edit worked, did nothing, or the data cannot tell.

How long before a forecast can be judged?

Usually a full period plus the time search needs to recrawl and settle — for most sites, four to eight weeks before actual versus forecast means anything.