Discussion
  • AI Search
  • Enterprise SEO

When leadership asks how much traffic AI Overviews cost the business, what method do you use, and how confident are you in the number?

Published 1 reply

Discussion starter: an enterprise SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.

Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.

1 reply

  • admin
    Admin
    Original poster

    I give leadership a range, built from two or three methods that point the same way, with the assumptions written down, and I tell them how confident I am in each part. No platform reports clicks lost to AI Overviews, so any number is an estimate. The most defensible estimate compares click-through rates on queries where AI Overviews appeared with comparable queries where they did not, before and after, using your own Search Console data and a record of which queries showed AI Overviews. Third-party studies give a sense of scale, and the “Search generative AI performance” report shows which pages are exposed, but neither produces your number on its own.

    What the public evidence suggests about scale

    Two widely cited sources give a sense of the size of the effect. Pew Research Center tracked the browsing of US adults in March 2025 and found that when a search showed an AI summary, users clicked a traditional result in 8% of visits, compared with 15% when no summary appeared. Ahrefs, using aggregated Search Console data, estimated that the presence of an AI Overview was associated with a substantially lower click-through rate for the top-ranking result, comparing December 2025 with December 2023. Both are useful context; neither describes your audience, your query mix or your positions.

    Methods I combine

    Method What it needs Strength Weakness
    Query-level comparison A rank tracker that records whether an AI Overview appears for each tracked query Uses your own data; controls for position Tracked queries are a sample; AI Overview presence varies by user
    Before and after on affected queries Dates when AI Overviews started appearing for query groups Direct Confounded by seasonality, updates and reporting changes
    Page-level exposure Search Console’s generative AI report (impressions by page) Shows which pages are exposed Impressions only, no clicks
    Benchmark application Published CTR effects applied to your exposed traffic Quick Imports someone else’s audience and method

    Controlling for the things that also moved

    • Google’s removal of the &num=100 parameter in September 2025, which changed reported impressions and average position for many sites. Compare clicks rather than impressions across that date, or compare periods on the same side of it.
    • Core updates, which move rankings independently of AI features. Annotate them and exclude update windows from comparisons where possible.
    • Seasonality: compare with the same period the previous year.
    • Demand changes: branded search and market demand can change clicks without any AI effect.
    • AI Mode in the totals: Search Console counts AI Mode within web search totals without a separate click breakdown.

    How I present it

    1. A range, not a point. For example: “We estimate AI Overviews reduced organic clicks by between X and Y over the year, concentrated in informational content.”
    2. Where the loss sits: which page types and topics, and whether they carried revenue.
    3. Confidence levels: high confidence that informational queries lost clicks; lower confidence about the exact size.
    4. What offsets it: AI impressions for your pages, citations, and any change in branded search or conversions.
    5. What we are doing about it: which content we are changing, which we are accepting as lost, and how we will measure.

    Hypothetical example:

    A home improvement retailer tracks 3,000 queries, of which about 40% showed AI Overviews by mid-2026. At comparable positions, informational queries with AI Overviews lost markedly more click-through than those without; commercial queries showed little difference. Applying the observed gap to the retailer’s informational traffic gives an estimated loss of 12% to 18% of total organic clicks, almost all on guides that rarely led directly to purchases. Revenue-linked queries were largely unaffected. The team presents the range, the method and the finding that most revenue was not exposed, and proposes restructuring guides to link more directly to products.

    Watch for this:

    Presenting every organic decline since 2024 as the cost of AI Overviews is tempting and will not survive scrutiny. Leadership will eventually hear about core updates, reporting changes and competitor gains. Separate the causes, or the whole estimate loses credibility.

    Show the other side of the ledger

    A loss estimate on its own invites the conclusion that search is in decline. Put it next to what is growing or holding. Search Console’s generative AI report shows how many impressions your pages receive in AI Overviews and AI Mode, which is exposure the business did not have before, even without clicks. Branded search, conversions from organic and revenue per organic session may be stable or rising even as informational clicks fall, because the visitors who still click are further along in their decision. And some content that lost clicks was never commercially important. Presenting losses and offsets together gives leadership a fair picture and moves the conversation from “how much did we lose” to “where should we invest now”.

    How confident to be

    Most teams can say with reasonable confidence which content types lost clicks to AI Overviews and roughly how much, as a range. Few can give a precise figure, and none can give one that isolates AI Overviews perfectly from everything else that changed. Saying that clearly is part of the answer. Leadership generally prefers an honest range and a plan to a precise number that falls apart when questioned.

    Estimate with your own query-level data, sense-check against published studies, control for reporting changes and updates, present a range with confidence levels, and focus the conversation on which revenue was actually exposed.

    Need help with this on your own site?

    Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

    See SEO services

Add a reply

Replies are open to approved members. Log in or apply to join.