Discussion
  • AI Search
  • Content SEO

Have AI Overviews made 3,000-word guides less effective for queries the AI now answers fully, and what replaced them on your site?

Published 1 reply

Discussion starter: a content 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

    The short answer is yes, but word count is not the reason. Long guides lose value on queries where the guide mostly summarised what everyone already agrees on. On those queries an AI Overview gives the same summary before anyone reaches your page. The guide’s traffic came from hundreds of long-tail questions, and those are the queries AI features answer most often. Long-form content with something a summary cannot supply still earns clicks: original data, a working tool, first-hand testing, or advice that depends on the reader’s situation.

    So the useful question is not “long or short”. It is: which parts of each guide were only restating a consensus, and which parts gave someone a reason to click? What tends to replace the monolithic guide is a smaller core answer plus assets that do things a summary cannot. Many guides still rank and convert well, so segment your own query data before you rewrite anything.

    What the evidence shows, and where it conflicts

    The strongest public behavioural data comes from Pew Research Center. It analysed the March 2025 browsing of 900 US adults across roughly 69,000 Google searches. Users who saw an AI summary clicked a traditional result in 8% of visits, against 15% for those who did not. Clicks on links inside the summary happened in 1% of visits. The trigger pattern matters even more for guide owners. Only 8% of one- or two-word searches produced a summary, rising to 53% for searches of ten words or more and 60% for queries that started with a question word. Those long, question-shaped queries are exactly what a 3,000-word guide was built to collect.

    Google’s position is different. In August 2025 Liz Reid wrote that “total organic click volume from Google Search to websites has been relatively stable year-over-year”. She said clicks are being redistributed, and that people increasingly click through to forums, videos, in-depth reviews, original posts and “thoughtful first-person analysis”. Google has not published the data behind that claim.

    These two findings do not necessarily contradict each other. Total clicks can stay flat while a specific page type loses share to other page types. Aleyda Solis’s analysis of the May 2026 core update pointed the same way, with reference aggregators among the sources that lost visibility. A guide that rewrites the top ten results is, functionally, a reference aggregator.

    Important distinction:

    Being cited in an AI Overview is not the same as getting the click. Ahrefs found that the share of AI Overview citations coming from top-10 pages fell sharply between July 2025 and March 2026. Its detection method changed in that period, so the two datasets are not directly comparable. Rankings, citations and clicks now need to be tracked as separate outcomes.

    Why comprehensive guides are more exposed than other content

    Think of the old guide as the library’s reference desk. People came in with small questions (“what is X”, “how long does Y take”) and the desk answered them, so a comprehensive guide collected a large share of those visits. An AI Overview puts someone at the front door who answers the small questions before people come inside. People still come in, but for the things the doorkeeper cannot hand over: the source documents, a tool, a specialist’s judgement, evidence they can verify.

    In technical terms, a long guide ranks for a wide spread of query intents. Some are answer-seeking (a definition, a fact, a short procedure). Others are task-seeking (compare options, calculate something, download a template, check evidence). AI features absorb much of the answer-seeking demand. Task-seeking queries still need a page. When a guide’s clicks came mostly from answer-seeking long-tail queries, its CTR can fall a lot while its rankings barely move.

    How to diagnose which guides actually lost value

    1. Segment by query, not by page. Export the guide’s queries from Search Console and group them by intent: definition, procedure, comparison, calculation, troubleshooting, commercial. The page total hides which clusters moved.
    2. Compare CTR at similar positions. Where position held steady but CTR dropped, absorption by a SERP feature is the more likely cause. Where position dropped too, look at ranking competition and core update timing before you blame AI Overviews.
    3. Handle the impression distortion. Google stopped supporting the &num=100 parameter in September 2025, and many sites saw desktop impressions fall, while clicks were affected far less, according to Brodie Clark’s analysis. Year-on-year impression and average-position comparisons that span that date are unreliable.
    4. Check the generative AI performance report. Search Console now shows impressions (not clicks, CTR or queries) for AI Overviews and AI Mode, filterable by page. It reached all eligible sites on 31 August 2026. A guide with high AI impressions and falling clicks is a strong candidate for restructuring. AI Mode activity is also counted in the standard Web totals.
    5. Check what the guide did for the business. Look at assisted conversions, newsletter or trial signups, and the backlinks the page holds. A guide that lost half its clicks but keeps its converting visitors and its links may need little change.

    Deciding what to do with each part of a guide

    Query pattern the guide ranked for How fully an AI summary tends to satisfy it What I would do with that section
    Definitions (“what is X”) Usually fully Keep a short, accurate answer inside the hub page. Stop investing in it as a traffic source.
    Generic procedures (“how to do X”) Often, for simple tasks Keep it if you can add something a summary cannot: a template, a checklist file, screenshots of a real interface, or video.
    Situation-dependent choices (“best X for Y”) Partly: generic advice without context Build decision pages or tools that ask for the reader’s constraints and give a specific recommendation.
    Data and benchmarks Poorly without a source to cite Publish original data with a clear method and a date. Other sites cite it and so do AI answers, which also earns links.
    Troubleshooting specific errors Partly Write first-hand diagnostic content: the actual error text, what causes it, and how to confirm the fix.
    Commercial comparisons Partly, and users often want to verify Move closer to the product: honest comparisons, pricing clarity and evidence from real use.

    What tends to replace the monolithic guide

    A tighter hub

    A shorter core page that answers the main question well and links clearly to deeper assets. It keeps the URL, its links and its rankings.

    Tools and templates

    Calculators, configurators, downloadable checklists. A summary can describe these, but the user still has to click to use them.

    Original evidence

    Proprietary data, test results, benchmarks and case evidence with a stated method. This is the information gain that summaries have to cite.

    Decision pages

    Pages built around a specific situation or trade-off, where the right answer depends on the reader’s constraints.

    The replacement is not “shorter content”. It is content where the length is justified by information someone would otherwise have to go and gather. Google’s February 2026 Discover core update, for example, said it would show more in-depth, original content from sites with topic-by-topic expertise. That is about Discover, not web search, but it points the same way.

    Watch for this:

    Do not break a strong guide into dozens of thin pages, one per question, in the hope of being cited. Google’s guide to generative AI features says you do not need artificial “chunking” or AI-specific rewrites, and that optimising for these features is still SEO. Mass-producing near-duplicate question pages also moves you towards scaled content abuse. And splitting a page that holds most of a topic’s backlinks can lose the rankings you still have.

    Hypothetical example:

    A payroll software company has a 3,500-word guide to calculating employee leave entitlements. Search Console shows clicks down sharply on “what is” and “how is leave calculated” queries at stable positions. Clicks are steady on queries that include an award, an industry or “calculator”. The sensible move is to cut the definitional sections to a concise summary and build an interactive entitlement calculator on the same URL or a closely linked one. Then add a worked-examples page per common award, reviewed by someone qualified. The URL keeps its links. The traffic that remains comes from people with a task to complete, and they are closer to a trial signup.

    How to test it instead of assuming

    • Hypothesis: reworking guides to add a tool, data or decision support recovers non-brand clicks and conversions better than leaving them as they are or simply shortening them.
    • Groups: take a set of comparable guides in the same section. Rework half and leave half unchanged as a control. Match the groups on traffic, links and intent mix.
    • Primary metric: non-brand clicks and conversions (or assisted conversions) per guide. Secondary: AI feature impressions from the generative AI report, backlinks earned, engaged sessions.
    • Duration: at least 8 to 12 weeks, noting any core update in the window, since Google has documented that smaller core updates happen beyond the announced ones.
    • What would change the decision: if the reworked guides do no better than the control on conversions, the loss is probably demand-side (fewer people need to click at all). The budget then belongs in commercial pages and brand, not in rebuilt guides.

    A practical position

    If these were my guides, I would sort every guide’s query clusters into answer-seeking and task-seeking. I would stop spending on the answer-seeking parts beyond keeping them accurate and concise. Then I would put the saved effort into one asset per guide that a summary cannot replace: a tool, a dataset, a first-hand test or a decision page. The guides that deserve that investment are the ones that still hold links and assisted conversions. For the rest, a shorter, accurate hub page is enough.

    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.