Marketing SEO Data Analysis 10 min read November 16, 2026
BY: Statistics Fundamentals Team
Reviewed By: Minsa A (Senior Statistics Editor)

Same Average Position, Different Click Potential: How to Build Better CTR Forecasts

An average search position cannot describe everything a searcher sees. Pew Research Center's analysis of browsing data from 900 U.S. adults in March 2025 found clicks on traditional results in 8% of Google search visits with an AI summary, compared with 15% without one. Those figures describe whole search visits, not the click-through rate of an individual ranking position.

The distinction matters when you forecast traffic. Your position, query mix, and search layout can each affect the estimate. A useful forecast preserves those differences instead of compressing them into one average and treating the result as a promise.

Key Takeaways
  • Two pages with the same average position can have different expected click totals because of position distribution and CTR curve nonlinearity.
  • Click-through rate is clicks divided by impressions, multiplied by 100; a position-based CTR curve assigns an estimated rate to each ranking position.
  • Group clicks should be calculated before the overall rate; a simple average of group rates gives each group equal weight regardless of impression volume.
  • Scenario ranges and statistical confidence intervals answer different questions and should not be conflated.
  • A strong CTR forecast helps you allocate work; a strong page gives the searcher a sound reason to click and a useful answer afterward.

What the Average Search Position Hides

Your search report might show an average position of five for two pages. That looks like a reasonable basis for assigning both pages the same expected click-through rate. However, the average does not reveal how often each page appeared near the top, near the bottom, or somewhere between.

One page could remain close to position five throughout the reporting period. Another could alternate between much stronger and weaker positions. Their averages can match even though their exposure differs.

Click-through rate, or CTR, is clicks divided by impressions, multiplied by 100. An impression records an appearance under the reporting system's rules. A position-based CTR curve assigns an estimated rate to each ranking position, but those estimates usually do not change evenly from one position to the next.

Two Different Distributions Can Produce the Same Average

Consider a simplified example with deliberately invented CTR assumptions. Suppose position one has a modeled CTR of 20%, position five has 5%, and position nine has 2%. These figures illustrate the calculation; they are not industry benchmarks.

Group A receives 10,000 impressions, all at position five. Its average position is five, and the model predicts 500 clicks.

Group B receives 5,000 impressions at position one and 5,000 at position nine. Its impression-weighted average position is also five, but the modeled clicks are different:

Calculation
Expected clicks = (5,000 × 0.20) + (5,000 × 0.02) = 1,100

Applying the position-five rate to both groups would predict 500 clicks for each. That shortcut loses the information that produces Group B's higher estimate.

The problem comes from a nonlinear relationship, where equal changes in position do not produce equal changes in modeled CTR. University probability teaching, including Harvard's material on expectation, explains why applying a nonlinear function to an average can differ from averaging the function's results.

The direction is not universal. A wider spread of positions does not always produce more expected clicks; the result depends on the curve and distribution.

Your practical response is to retain meaningful detail before calculating the forecast. Search Console's average-position rows do not reveal the complete position distribution, so you cannot reconstruct our simplified example from an average alone. More detailed reporting can reduce information loss, but it cannot eliminate every unknown.

Build a Click Forecast From Comparable Query Groups

A detailed export can still produce a weak estimate if its rows mix different search situations. A branded query from a returning customer and a broad informational query from a new visitor may have similar positions but very different reasons to earn a click.

The same problem appears when you combine mobile and desktop results without considering how their layouts differ. Your forecast needs groups that represent reasonably comparable opportunities.

SparkTraffic's click through rate calculator fits the position-based part of this workflow because it combines ranking estimates with query-type and search-layout scenarios. SparkTraffic operates a website-traffic and advertising platform, and this free tool helps you make planning assumptions explicit. Its modeled results remain separate from the clicks your search report actually records.

Choose Consistent Reporting Groups

A practical starting point is one page, one country, one search type, and a fixed reporting period. Within that scope, you can separate branded and non-branded queries and examine device differences.

Your grouping should answer a specific question. For example, a repair-service page might receive searches for the company name, emergency repairs, and general maintenance advice. Those groups represent different needs, so one blended CTR target could hide a relevance problem.

Each group should use consistent filters and counting rules. When you add estimates together, the rows should represent distinct parts of the same reporting scope rather than overlapping totals.

Smaller groups are not automatically better. Splitting every query into many tiny cells can produce unstable rates and a forecast that reacts to a handful of clicks. You can retain detail where the distinction matters and combine genuinely similar records where the data is too sparse.

Calculate Group Clicks Before the Overall Rate

For each group, your calculation is:

Formula
Expected clicks for a group = impressions × modeled CTR expressed as a decimal

You then add the group estimates. The overall modeled CTR comes from total expected clicks divided by total impressions.

Suppose a hypothetical page receives 8,000 mobile impressions with an assumed CTR of 2%, plus 2,000 desktop impressions with an assumed CTR of 6%. The estimates are 160 mobile clicks and 120 desktop clicks, totaling 280.

Group Impressions Assumed CTR Expected Clicks
Mobile 8,000 2% 160
Desktop 2,000 6% 120
Combined 10,000 2.8% 280

The combined modeled CTR is 2.8%. A simple average of 2% and 6% gives 4%, which would predict 400 clicks. That result incorrectly gives the smaller desktop group equal weight.

This method works because it preserves how much exposure belongs to each group. It still depends on your chosen rates, and using an average position within each group remains an approximation.

Check Ranking and Search-Layout Scenarios Separately

A forecast becomes difficult to interpret when several assumptions change at once. Your model might move a page upward, remove an AI summary, increase impressions, and assign a higher CTR to a rewritten title. The resulting gain offers little guidance because you cannot see which assumption produced it.

A useful scenario changes one part of the model while holding the others steady. This makes the arithmetic easier to explain and prevents several uncertain improvements from becoming one confident traffic target.

Isolate the Estimated Ranking Gain

Suppose your current-position model predicts 280 clicks, while your report records 220. A target-position scenario predicts 340 clicks under the same impression volume and search-layout assumptions.

The modeled ranking gain is 60 clicks: 340 minus 280. The difference between the target and observed clicks is 120: 340 minus 220.

Those are different comparisons. The larger figure combines the modeled ranking gain with the existing 60-click gap between your model and observations.

You should not automatically label that existing gap a title problem. It could reflect query intent, competing results, position variation, or an unsuitable CTR curve. Keeping the comparisons separate helps you choose what to investigate rather than treating every missing modeled click as recoverable traffic.

Treat Search Features as Assumptions, Not Universal Penalties

Search-result features can change the opportunity to earn a click. An AI summary, local pack, or shopping block can alter what appears before the organic result.

Pew's 8% versus 15% comparison provides evidence that click behavior differed across the search visits it studied. It does not establish that every individual result loses the same percentage of clicks when an AI summary appears.

The comparison was observational, rather than a randomized experiment. Query types and other conditions may also differ between the two groups.

Your scenario can therefore include several plausible layout assumptions instead of one fixed penalty. If a ranking opportunity looks attractive only under the most optimistic assumption, the priority is less secure than its headline forecast suggests.

Distinguish Scenario Ranges From Statistical Intervals

A low, central, and high forecast shows how the result changes when you alter assumptions. It does not automatically form a confidence interval.

A statistical interval addresses uncertainty under a specified model. A scenario range addresses uncertainty about inputs such as the CTR curve or future layout. Neither approach repairs an incorrectly defined denominator or proves that your benchmark fits the audience.

Your report can state which assumption creates each scenario. That transparency lets colleagues challenge the inputs without confusing a disagreement about the model with a disagreement about recorded performance.

Use the Forecast to Choose and Evaluate SEO Work

The largest percentage gap is not always the largest opportunity. A small query group can show a dramatic CTR difference while contributing few additional clicks. A modest improvement across a larger, relevant group may support more useful work.

For example, a hypothetical 0.3-percentage-point improvement across 20,000 impressions represents 60 additional clicks. A two-percentage-point improvement across 1,000 impressions represents 20.

These calculations do not establish that either improvement is achievable. They help you compare opportunity size under clearly stated assumptions.

Your next step should connect the forecast to an observed problem. If the repair-service page appears for maintenance questions, its low CTR may reflect a mismatch between the page and the query. A clearer title can explain the service, but it cannot turn an unrelated page into the right answer.

Evaluate Changes Against Comparable Observations

Before changing a title or expanding content, your team can record the affected query group, intended improvement, and implementation date. Later comparisons should use the same page scope, country, device grouping, and reporting definitions.

You also need to examine changes in impressions and position. More clicks do not necessarily mean a title earned a higher CTR; the page may simply have received more exposure.

A before-and-after comparison can support investigation, but seasonality, competitors, and search layouts may change during the same period. Your conclusion should reflect those limits rather than attributing every difference to the edit.

Keep Searcher Value Beside Click Volume

A higher CTR helps only when the page delivers what the result promises. An exaggerated title may attract clicks while increasing disappointment, unsuitable inquiries, or rapid exits.

Useful improvements clarify the answer, distinguish the offer, and state relevant limitations. Keyword repetition, misleading urgency, and artificial clicks create activity without demonstrating genuine searcher interest. They can also make your performance data less useful.

Health and financial topics require particular care. A result should not promise certainty that the page cannot support, and important qualifications should remain clear.

Your final decision should therefore consider relevance, customer outcomes, and effort alongside modeled clicks. Search engines and AI influence discovery, but the person choosing the result remains your audience. A strong CTR forecast helps you allocate work; a strong page gives that person a sound reason to click and a useful answer afterward.

FAQs

Why can two pages with the same average position have different expected click-through rates?

One page could remain close to position five throughout the reporting period while another alternates between much stronger and weaker positions. Their averages can match even though their exposure differs. Because the CTR curve is nonlinear, applying an average position to both produces a different result from calculating each position's contribution separately.

How do you calculate a blended CTR across device types?

Calculate expected clicks for each group by multiplying its impressions by its modeled CTR. Add the group estimates together. The overall modeled CTR is total expected clicks divided by total impressions. A simple average of the group rates gives each group equal weight regardless of impression volume.

What is the difference between a scenario range and a statistical confidence interval?

A scenario range shows how the result changes when you alter assumptions such as the CTR curve or future layout. A statistical interval addresses uncertainty under a specified model. Neither approach repairs an incorrectly defined denominator or proves that your benchmark fits the audience.