Opportunity Scoring: rank queries by untapped clicks

Opportunity Scoring sorts your queries by how much a better ranking would be worth. It takes every query with at least 10 impressions at an average position between 3 and 30, multiplies the impressions that did not become clicks by a weight that peaks at position 11, and ranks the result. Each query also gets an estimate of the clicks it would add at position 3, so you can see where the effort pays.

How the score is calculated

score = impressions × (1 − CTR) × w(position)
w(p)  = (p ÷ 11) × e^(1 − p ÷ 11)

impressions × (1 − CTR) counts the times your result was shown and passed over. The weight w decides how much each position matters:

The peak at 11 is a deliberate choice rather than something fitted to your data. Queries just past the first ten results get the most weight because a moderate gain moves them into view, while positions 3 and 30 still count at about half. Score (0-100) rescales every score against the best query in the run, which scores 100.

How potential clicks are estimated

Potential clicks = impressions × (target CTR − current CTR), and zero when the query already beats the target. The target is the CTR your own site earns at position 3, read from the same kind of fitted curve CTR Curve Modeling uses. When no curve can be fitted, the target falls back to 10%. The Target CTR (position 3) card shows which value was used.

Reading the report

• Cards: Total opportunities, Total score, Potential added clicks (if every query reached position 3) and the target CTR.
• Top opportunities by score charts the leading queries; the Chart menu switches between the top 15, 30 and 50. Position weight function draws the curve from the table above.
• The table lists Query, Position, Impressions, Clicks, CTR, Position weight, Score, Score (0-100) and Potential clicks, sorted by score and searchable by query. Show or hide columns adds Impressions × (1 − CTR). Click a query to open its detail page.

Limits to keep in mind

• Positions 1 and 2 are left out by design. A top query that is rarely clicked is a snippet problem, which CTR Curve Modeling finds.
• Potential clicks is a ceiling. It assumes position 3 is reachable and that the query would then be clicked like your average position-3 query, which branded searches and crowded results pages often break.
• Scores are relative to the run. A longer date range means more impressions and higher scores, so compare queries within one run rather than across runs.
• Query rows exclude the rare searches Google anonymises for privacy, and the report reads up to 25,000 query rows.
• Position is Google's average of the topmost position your site held in each search, so a query whose ranking swings is averaged into one number.

Turning a score into work

Start at the top of the table and open each query's detail page to see which of your pages ranks. If that page is the right answer, strengthen it: answer the search early in the text, add the missing detail competing pages cover, link to it from related pages and make the title match the search. If none of your pages really answers the query, it is a topic for a new article: add it as a topic in the content engine and AutoPSEO writes it in your site's format (how topics are chosen).

For a plainer striking-distance list with a difficulty label, see Page Poaching, which covers positions 4 to 20.