AI Overviews Traffic Drop: Why Blog Traffic Fell

An AI Overviews traffic drop can happen when Google answers a searcher’s basic question on the results page, leaving fewer reasons to visit your article. That does not mean every informational post is finished or that your site has been penalised. It means you need to identify which searches and pages lost clicks, then offer evidence, experience or tools that a summary cannot replace. Our guide to redesigning pages for AI Overviews clicks looks at the page design side of that problem.

This article explains what changed, how to diagnose a decline without guessing, and how to reshape generic posts into useful, original resources. The goal is not to chase every AI feature. It is to make your content valuable to a real buyer and clear enough for search engines to understand.

What is an AI Overviews traffic drop?

An AI Overviews traffic drop is a reduction in organic visits to a website that coincides with Google showing AI-generated summaries for relevant searches. An AI Overview is a summary displayed in Google Search that combines information and links from web pages to respond to a query. It can appear above traditional unpaid search results, although its presence and layout vary by search and location.

An informational query is a search made mainly to learn something, rather than to buy or contact a business immediately. Examples include “what is brand positioning?” and “how does a booking system work?” These searches often have a quick answer, which makes them easier to summarise than a decision that depends on a person’s requirements.

A traffic decline on its own does not prove that AI Overviews caused the change. Search demand can shift. Rankings can move. A site change, tracking issue, seasonality or a search algorithm update can also affect visits. You need to compare the right pages, queries and dates before deciding what to fix.

Why can clicks fall while impressions and rankings look steady?

An impression is counted when a result from your site is shown in Google Search. A click is counted when someone selects that result and visits your page. Click-through rate, or CTR, is the share of impressions that turn into clicks. If your result keeps appearing but fewer people select it, impressions can stay level while clicks and CTR fall.

One reason is that an AI Overview can satisfy a searcher before they open a website. In its July 2025 analysis of browsing data from 900 US adults and 68,879 Google searches, Pew Research Center reported that people clicked a traditional result in 8% of searches where an AI summary appeared, compared with 15% when one did not. The study describes observed behaviour in its sample, not a guaranteed rate for every industry or country. Read the Pew Research Center’s study of clicks when AI summaries appear.

That is a change in the search results page, not necessarily a change in your ranking. A page can retain its position while competing with an answer box, videos, maps, shopping results and other features. Search results pages, often shortened to SERPs, are the pages Google displays after someone searches. Their changing layout affects how much attention a traditional link receives.

Several studies point to a click impact, but their methods and samples differ. A field experiment described by Search Engine Journal in April 2026 reported a 38% reduction in outbound organic clicks on searches where AI Overviews appeared. Ahrefs reported in February 2026 that its analysis of informational keywords found a 58% CTR decline for position-one results associated with AI Overviews. These are study findings, not a forecast for your specific site. See the reported field experiment on AI Overviews and organic clicks and the Ahrefs analysis of AI Overviews and click-through rates.

The useful takeaway is not that every blog has lost a particular percentage. It is that a ranking report alone no longer tells the whole story. Check whether the search itself changed, whether an AI Overview appears, and whether the page still attracts the right visitors.

Why generic definition posts are more exposed

A short answer to a common definition is relatively easy to summarise. If a page says what a familiar term means and lists a handful of standard benefits, an AI system may be able to provide much of that information directly in the search experience. The searcher may feel they have what they need without opening the source pages.

This is not a claim that Google copies every article or that every simple topic will lose its clicks. It is a practical distinction between information that is widely repeated and information that depends on a specific business, process, dataset or real-world decision. A generic post about “what is customer lifetime value?” can explain the concept. It cannot, by itself, tell a retailer which customer segments are profitable, how the calculation was built, or what action the team took after reviewing the numbers.

A common older content approach was to publish many broad “what is” and “benefits of” articles to attract early-stage searchers. Those posts can still help people and support a larger website. The problem comes when the article offers no reason to read past the definition. Reaching a target word count does not add original value if the extra paragraphs repeat familiar points.

A useful test is to remove the company name from the article. If the same page could be published unchanged by dozens of competitors, it probably needs a stronger point of view, evidence or practical tool. That does not mean adding jargon or an unsupported opinion. It means showing what your team knows that a general summary cannot establish.

Does research prove that every blog has lost traffic?

No. Industry-level findings show a pattern, but they do not diagnose an individual website. Studies may examine different countries, devices, query types, time periods and definitions of a click. Their results should be treated as context for your own analysis, not copied into a business forecast.

For example, SparkToro’s June 2026 analysis reported that 68.01% of US Google searches in its clickstream panel ended without a click to the open web. That is a claim about the study’s sample and methodology, not a universal rate for every search engine, country or subject. Its research on US Google searches that end without a click is useful background when discussing zero-click searches. A zero-click search is one where the person does not select a result that takes them to another website.

Numbers can also distract from the decision you need to make. If your visits are down but qualified leads and sales are steady, the commercial impact may be smaller than the traffic chart suggests. If visits are steady but enquiries have fallen, the bigger issue may be the page’s offer, proof or conversion path. Measure traffic and outcomes separately.

How do you diagnose an AI Overviews traffic drop?

Start with a clear comparison, then inspect the search results and the page itself. Do not delete a group of articles simply because their traffic fell. Use this sequence to separate a search feature effect from other causes.

1. Choose a useful comparison period. In Google Search Console, compare the same number of weeks or months year on year where seasonality matters. Also compare the period immediately before and after the decline. Search Console is Google’s free tool for reviewing how a site performs in Google Search. Its performance report documentation explains the available search performance data. The report does not, by itself, prove that an AI Overview caused a change.

2. Compare clicks, impressions, CTR and average position. Filter by the affected page and its queries. If impressions and average position are broadly stable while CTR falls, a change in the search results page or searcher behaviour is plausible. If impressions also fall, investigate demand and visibility before blaming the summary box.

3. Check the actual searches. Search several important queries yourself, ideally in the relevant country and on the device your audience uses. Search results can vary, so one screenshot is not a complete dataset. Note whether an AI Overview appears, what it answers, which sources it cites and how much space it takes above the traditional results.

4. Look for other changes. Review when the drop began and compare that date with site releases, URL changes, seasonality, indexing issues and changes to the page title or content. Check whether you changed analytics or consent settings, too. A broken measurement setup can look like a traffic loss when visits are still happening.

5. Connect visits to business outcomes. Compare enquiries, booked calls, sales and the quality of leads from organic search. Where possible, look at the landing pages and queries associated with those outcomes. A lost visit is not always a lost customer, and a click is not automatically a qualified lead.

6. Keep a record before changing the page. Save the date range, query list and screenshots or notes from your SERP checks. This gives you a baseline for judging a rewrite. Change one meaningful thing at a time when practical, rather than rewriting every post at once.

A simple diagnosis table can keep teams from treating every decline the same way:

What you see	What it might mean	What to check next
Impressions steady, CTR down	Searchers may be choosing an AI answer or another result	Inspect the live SERP and compare affected queries
Impressions and clicks both down	Lower demand, weaker visibility or an indexing issue may be involved	Review trends, rankings, index status and seasonality
Visits steady, enquiries down	The page or offer may not meet the visitor’s needs	Review intent match, proof, CTA and form experience
Visits down, qualified enquiries steady	Fewer low-intent visits may be affecting totals more than revenue	Track lead quality and sales, not sessions alone

What kind of content can earn a visit when a summary is available?

The strongest answer is not “make everything longer”. Build pages around information that is difficult to infer without the source: original evidence, first-hand experience, a clear method, or an action the reader can take. These are ways to create a reason to visit, not a guarantee of rankings or AI citations.

A case study is a detailed account of a real project or business situation, including the starting problem, what was done, what changed and what was learned. A credible case study can include the client’s context, timeframe, constraints, decision process and measurable outcome, provided the business has permission to publish those details and can substantiate the claims.

For example, a web design studio could write about how a service business reviewed its mobile enquiry flow. A useful account would describe the original friction, the analytics or customer feedback that revealed it, the design changes, the way results were measured and any limitations. If there is no verified outcome data, say so. Do not invent a percentage increase to make the story more compelling.

Other content formats can offer distinct value as well:

• A data teardown: Explain what a real dataset shows, how it was collected and what it cannot tell you. Remove personal or confidential information.
• An implementation guide: Show the actual sequence, decisions and trade-offs involved in completing a task, rather than offering a generic checklist alone.
• A decision framework: Help readers choose between options based on their business situation, using criteria they can apply themselves.
• A diagnostic tool: Offer a calculator, checklist or scorecard that helps someone assess their own situation. Explain the assumptions behind it.
• A first-hand review: Describe who tested a product or process, what they tried, what worked and what did not. Disclose relevant commercial relationships.

First-party proof means evidence collected or experienced directly by your business, rather than a claim repeated from another publisher. It can include original screenshots, interviews, test results, annotated design decisions or internal process documents that are safe to share. First-party does not mean automatically reliable. State how the evidence was gathered, what period it covers and where it may not apply.

A case study also answers a different question from a definition article. The definition helps someone understand a term. A case study helps a potential buyer judge whether the approach might work for a situation like theirs. That makes it particularly useful for readers comparing providers, platforms or methods. It does not make every case study automatically high-converting; the proof must be relevant and the next step must be clear.

How should you redesign a blog post around proof?

A blog layout affects whether readers can find the evidence, understand the argument and take the next step. Design is not decoration added after the writing. Hierarchy, spacing, tables, images and calls to action help people judge whether a page is credible and useful.

Start with the answer and the proof. In the first screen, explain what the article will help the reader decide. If it is a case study, include a concise project summary with the context, work completed and verified outcome. If the outcome is confidential or not measured, do not imply otherwise. Avoid a long, generic introduction before the reader reaches the reason the article matters.

Use a clear information hierarchy. A heading should tell a reader what the next section answers. Short paragraphs and descriptive subheadings make a page easier to scan. A comparison table is useful when the reader is weighing options. A numbered sequence is better when the order of actions matters. A checklist works when the reader needs to inspect their own site or process.

Make images carry evidence. An annotated wireframe can show why a layout changed. A screenshot can explain a user flow. A chart can make a dataset easier to interpret. Add a caption that says what the reader is seeing and why it matters. Use original images only when you have the rights to share them, and remove customer details that should remain private.

A call to action, or CTA, is the visible prompt asking a reader to take a next step. Place it where it fits the reader’s stage, such as after the evidence or at the end of a decision guide. One relevant action is usually clearer than several competing buttons. If the article is about search visibility, a free AI search audit is a more natural next step than an unrelated sales prompt.

Use semantic structure: headings, lists and tables that communicate what each section is for, rather than visual styling alone. This helps readers navigate and helps software interpret the content. It is not a special trick that guarantees Google or an AI system will cite the page. Good structure supports comprehension; it cannot compensate for weak evidence.

For more on how page structure and visual hierarchy support discovery, see our guide to website design for AI search. If the article is part of a broader site problem, our website design service can help bring content, user journeys and conversion goals together.

Should you delete or rewrite old definition posts?

Usually, assess each page before deciding. A traffic decline alone is not a reason to delete an article that still supports customers, earns links, assists other pages or attracts valuable enquiries. Nor is every old post worth keeping if it is inaccurate, duplicative or unrelated to what the business offers.

Use this decision path:

• Keep and improve it if the page still meets a real need, is accurate and has a useful audience. Add original examples, clearer steps, evidence and links to relevant services.
• Combine overlapping pages if several posts answer the same question without offering distinct value. Choose the strongest URL, bring across useful material and redirect retired pages where appropriate.
• Reframe it if the basic answer remains useful but readers need help applying it. Add a decision guide, a worked example or a real project insight.
• Remove or redirect it carefully if it is obsolete, misleading or has no continuing purpose. Check links, search visibility and the destination before making a change.

A redirect sends visitors and search engines from an old URL to a relevant new one. Do not redirect every deleted post to the homepage; a destination should make sense to the person who followed the old link. Preserve the original page’s useful information where it remains accurate, and check your site’s links after the change.

What should a small business publish instead?

You do not need to abandon informational content. You need to connect it to the decisions your business can help with. A useful editorial plan balances practical answers with proof and commercial relevance.

For each planned article, write down four things before drafting:

1. The reader’s real question. What are they trying to understand or decide? Use customer conversations, sales questions, search data and support requests as inputs.
2. The value that requires your source. What can you add from direct experience, a real example, original analysis or a useful tool?
3. The decision the page supports. Is the reader choosing a design approach, service, platform or provider? Make that decision visible rather than hiding it beneath general background.
4. The next sensible action. Could the reader use a checklist, compare options, see relevant work or ask for help? Match the action to the information they have just read.

A business selling design services, for example, can explain what a homepage does and also show how it prioritises information for a particular audience. It can discuss what the team considered, what was rejected and what evidence shaped the final design. A business selling a product could pair a how-to with an original test or a transparent comparison. The specific examples must come from real work, not a template dressed up as experience.

There is still a place for clear definitions. Put the short answer near the top, then build out the parts that require judgement: how to apply the idea, where it fails, what it costs in time or effort, and how to choose between alternatives. That structure serves a reader who wants a fast answer and one who needs the deeper explanation.

How should you measure whether the new approach works?

Do not judge a content change by organic sessions alone. Track a small set of measures that connect discovery to business value. The right measures depend on the page and what the reader should do next.

Useful measures can include:

• Search Console impressions, clicks, CTR and average position for the page and its target queries.
• Organic visits to the page and the share of those visits that continue to a relevant service or product page.
• Enquiries, bookings or purchases associated with organic landing pages, where your analytics setup can measure them reliably.
• Lead quality, such as whether enquiries match your service area, budget or customer type.
• The page’s usefulness to sales and customer support, based on whether people share or refer to it during real conversations.

Keep a before-and-after record and allow enough time to understand normal variation. Do not treat a short-lived movement as proof that a rewrite worked. Search demand changes, competitors publish new pages and search features shift. Note what you changed, when you changed it and which outcome you expected to improve.

If you are monitoring AI visibility, be precise about what the available data can show. A citation or referral from an AI product is not the same as a Google organic click. Search Console performance data is useful for Google Search, but do not label a change as AI Overview traffic unless the data you have actually identifies it. Maintain a query sample and check the live results periodically rather than presenting a guess as a measurement.

The practical response to an AI Overviews traffic drop

The best response to an AI Overviews traffic drop is not to publish more generic articles or to rewrite every page for a speculative algorithm. First confirm which pages and searches lost clicks. Then decide whether the cause is a results-page change, lower demand, ranking loss, measurement problem or a page that no longer meets the visitor’s needs.

Where a post mostly repeats common knowledge, keep the concise answer but add a reason to visit: first-hand evidence, a tested process, a meaningful comparison or a tool that helps the reader act. Show that value through page design, with a strong hierarchy and proof people can inspect. Then measure whether the change improves qualified visits and outcomes, not just the total number of sessions.

If you want to understand how your brand and website appear across AI search, take the next step with DesignFIT’s free AI search audit.