AI traffic loss calculator: what is AI search costing you?

Multiply your monthly organic sessions by the share that come from queries showing an AI Overview, then by a published click-loss rate (34.5%, 58% or 61%). That gives clicks at risk. Multiply by conversion rate and order value for revenue. The result is a planning range for one search feature, not a forecast of your whole site.

Step by step

1. Traffic. Enter monthly organic sessions from Google Analytics or clicks from Search Console. Use organic only; paid, email and direct traffic aren't exposed to AI Overviews.
2. Exposure. Enter the percentage of that traffic landing on queries that trigger an AI Overview. This is the input that moves the answer most, so measure it if you can (see the FAQ below).
3. Scenario. Pick a CTR-loss coefficient. Moderate (58%) is the most recent large Ahrefs figure; the other two bracket it.
4. Conversion rate and order value. Use your organic conversion rate, not the site average, if your analytics separate them. For lead generation, use lead-to-customer rate and average deal size.
5. Read the four outputs. Clicks lost per month and per year, customers lost per month, and annual revenue at risk. Results update as you type.

Worked example (the default inputs)

A site with 50,000 monthly organic sessions, 45% of them on AI Overview queries, the moderate 58% coefficient, a 2% conversion rate and a $120 order value:

• Clicks at risk per month: 50,000 × 0.45 × 0.58 = 13,050
• Customers at risk per month: 13,050 × 0.02 = 261
• Annual revenue at risk: 261 × $120 × 12 = $375,840

Switch the same site to the conservative 34.5% scenario and the revenue figure drops to about $223,560. That gap is the honest width of the evidence.

What the calculator does not do

• It doesn't look up your site. Every input is typed by you, and nothing leaves your browser.
• It applies one coefficient to all exposed traffic. The studies measured the position-1 result, so for pages that rank lower the real loss may be smaller or larger; no public study pins it down.
• It doesn't model Google AI Mode, ChatGPT or Perplexity. The figures are for AI Overviews only.
• Inputs are clamped: negative numbers count as zero, and exposure and conversion rate cap at 100%.

The model is deliberately simple and transparent:

Annual revenue at risk =
  Monthly traffic
  × AIO-exposure %
  × CTR-loss coefficient
  × Conversion rate
  × Average order value
  × 12

The CTR-loss coefficient is the share of clicks the top result loses when an AI Overview appears. We offer three, all vendor-reported and graded partial by this site — the published figures, and the population each one measures:

Two caveats travel with these numbers. They measure the top result on informational queries, so applying them to every position overstates the loss for pages ranking lower. And Ahrefs' two figures moved from 34.5% to 58% between studies, which is why no single coefficient should be read as settled. The model deliberately excludes three things: brand or direct search that AI can't intercept, paid search backfill, and, importantly, the traffic you recover by being cited inside the AI answer.

How often AI Overviews appear depends heavily on the query mix, and the prevalence estimates in circulation come from vendor keyword panels, not a census. Separately, SparkToro and Datos estimated that about 58.5% of US Google searches in 2024 ended without a click — a figure about all searches, routinely confused with the AI Overview click effect above. For informational content, the old deal, write a helpful page, earn the click, is quietly breaking. The click is being absorbed into an answer box that may or may not name you.

This calculator exists to put a number on that shift so you can decide how much to invest in GEO. If the figure is small, your traffic is durable and you can relax. If it's large, the useful response is to try to become a source the answer cites, and to check which tactics for that have evidence behind them in the GEO tactic evidence lookup. You can see the wider set of published figures in Google AI Overview statistics.