Keyword Valuation and Prioritisation
Keyword valuation is arithmetic with stated assumptions, not a score out of ten. The popular formula — value × volume ÷ difficulty — does not work: volume is an approximate yearly average and difficulty is a per-vendor score, so the product has no unit. This lesson computes the expected value of a query from quantities that each have a source.
It rests on Search Console data and the Keyword Planner metrics reference.
The chain from demand to money
There are four steps between "volume" and revenue, and the number shrinks at each:
| Step | Quantity | Where it comes from |
|---|---|---|
| 1 | Approximate monthly demand | The 12-month average plus the monthly breakdown |
| 2 | Relevance share | How much of that demand is your task rather than an adjacent one |
| 3 | CTR at a position you can hold | Your own CTR from Search Console |
| 4 | Conversion and margin | Analytics: key events in GA4 |
The formula is plain: value = demand × relevance share × CTR × conversion × margin. Its point is not precision but that every factor can be challenged on its own and checked against a source.
The honest caveat: Google does not publish a CTR curve
Any "average CTR of position one" from an article is someone else's sample, not yours. But what you need you already have: Search Console returns CTR on a 0 to 1.0 scale and the average position for every query. So the curve can be built from your own data — group existing queries into position buckets and take the median CTR in each.
That is the only defensible coefficient for step 3. Building such a curve is what CTR analysis does: it shows the CTR that is normal for your site at each position, not an internet-wide average.
A worked example
A query with approximate demand of 1,000 a month. Our page is relevant to about half of that demand — 0.5. The position realistically available, judging by the composition of the results, is 4–10, where the site's median CTR is 3%. Conversion for pages of this type is 2%, and the average margin per order is 40 units.
1,000 × 0.5 × 0.03 = 15 clicks a month. 15 × 0.02 = 0.3 orders. 0.3 × 40 = 12 units a month. Whether that justifies building the page is your call, but the call is now argued in money rather than in points.
The last two factors can be checked against reality in conversions by query and the link between positions and revenue, and conversion left on the table by live pages in conversion opportunities.
Why this is always a range, not a number
- Part of the demand is answered on the results page. A featured snippet shows the descriptive snippet first, and some people no longer click through. Zero-click risk estimates that loss.
- A query can multiply. The documentation states that AI features may use a "query fan-out" technique — issuing multiple related searches across subtopics and data sources. The one-query-one-click link gets looser in both directions.
- Demand is uneven. A yearly average hides the season — see the lesson on metrics.
- The format may not match. If the results expect another page type, CTR falls below any bucket: see the lesson on search intent.
Prioritisation: value against effort
Priority comes not from difficulty in points but from effort in work: how many pages are needed, whether we have the data and the expertise, whether a new template is required. Hence the working sort order:
- Compute the expected monthly value for each topic.
- Estimate effort in units people understand: person-days, new pages, data required.
- Sort by value-to-effort, not by volume.
- At the top of the list, check the composition of the results: if it holds only large brands with data you do not have, the topic drops.
- Record the assumptions. In three months compare the forecast with the outcome — that is how the model learns.
Mistakes
- Treating volume as traffic.
- Taking a CTR curve from someone else's article instead of your own data.
- Multiplying scores together and comparing the result across topics.
- Ignoring the relevance share: a broad query is rarely entirely yours.
- Building the estimate on an exact position instead of a bucket.
- Never revisiting the forecast after publishing: without a comparison against actual impressions and clicks the model never improves.
What comes next
The full path from query to order is assembled in the SEO funnel. And the next lesson is about seasonality: what to do when demand exists only part of the year.