Glossary

CTR curve

A CTR curve maps each search position to the share of searchers expected to click that result. Clicks fall fast after the top three. Published curves are averages across many sites, so a client's own curve, fitted from its Search Console data, predicts its clicks far better, especially on local and non-branded searches.

What it doesn't do: A curve predicts clicks from position. It can't tell you why a page underperforms its curve; a map pack or AI Overview on the results page often does.

Updated Oct 22, 2026

Why one curve doesn't fit every site

Branded searches click far more than non-branded ones at the same position, because the searcher already wants that business. Map packs, AI Overviews and ads push organic results down the screen on many money searches. A curve built from a mix of all of these overstates what a local service page will get.

In our own calibration run on three sites we operate (13 weeks to Oct 11, 2026, measured Oct 20, 2026), a standard published-style curve overstated non-branded click-through by 14 to 76 times, depending on the position band. Branded searches came close to the standard curve. The methodology page shows the method.

How it shows up in RankUpside

Clicks-at-stake estimates use the client's own curve, fitted from 13 weeks of its non-branded searches, in position bands: 1, 2, 3, 4–5, 6–10, 11–20 and 21+. A band needs at least 1,000 impressions to use the client's own rate; otherwise a conservative pooled rate stands in, and neither ever exceeds a standard curve. Branded searches keep the standard curve.

Sample data: Copperline Gutters, a fictional business on a reserved .example domain. "gutter cleaning fairview" sits at #6.0 with 495 impressions. The striking distance finding estimates the extra clicks from reaching #3 with the client's own curve, which keeps the promise small and believable.

How to fit your own

Export 13 weeks of queries from Search Console and remove branded rows.
Round each row's position and put it in a band: 1, 2, 3, 4–5, 6–10, 11–20, 21+.
Per band, add up clicks and impressions, then divide clicks by impressions.
Treat bands with few impressions as unreliable.

Common mistakes

Applying a published curve's position-1 rate to a non-branded local search.
Mixing branded and non-branded rows in one curve.
Forgetting that one page can hold several positions for different searches.

CTR curve: questions.

What is a good CTR for position 1?

It depends on the search. In our own calibration run on three sites we operate, branded searches at #1 clicked at about 23%, while non-branded searches at #1 clicked well under 1%. Compare against your own site's curve, not an industry average.

Why are my client's CTRs so far below published benchmarks?

Published curves mix branded traffic and results pages with no map pack or AI Overview. Local service sites face both on most money searches, so their non-branded CTR runs much lower at the same position.

Does RankUpside use a fixed CTR curve?

No. It fits each client's curve from 13 weeks of its own Search Console data where there's enough volume, and uses a conservative prior where there isn't.

Connect a client and see their first report.

The first report builds from 13 weeks of Search Console history as soon as you connect. Start with 3 client sites.

Start free trial See a sample report

What your client sees each weekSample data

Oxbow & Vane

Weekly search report · Copperline Gutters

Sep 14 – Oct 11, 2026 vs the 4 weeks before

Copperline Gutters got 27 clicks from named buyer searches in the last 4 weeks, up 23% from the 4 weeks before. Google hides the search terms behind 84% of all 289 search clicks. Rankings rose across the site: pages that steadily show up are up about 2.4 positions.

Buyer clicks23%

27from 22

All search clicks14%

289from 253