Why Your Coaching Website Stopped Getting Traffic in 2026

Open Google Search Console and look at the last 12 months. If your impressions are flat or up but your clicks have fallen off a cliff, you are not imagining it. That gap is the exact signature of AI Overviews answering buyers without sending them to your site.

Key takeaways

• The traffic drop most coaches are seeing in 2026 has one specific signature in Google Search Console: impressions steady or rising, clicks falling. That pattern means AI Overviews is answering the query inside the search page.
• The fix is not "more SEO." Writing more blog posts with the same keyword targets makes the problem worse because AI Overviews scrapes your answer and keeps the click.
• The highest-leverage move is getting cited by AI engines directly (ChatGPT, Claude, Perplexity, Microsoft Copilot) so buyers find you inside the answer, not after it.
• BakingSubs, a niche site I built, earned 144,321 Microsoft Copilot citations in a single quarter using The Citation Cluster Method, with no ads, backlinks, or social.
• The free AI Visibility Check at /visibility-check runs 8 discovery questions across the 4 main engines and tells you exactly where you are invisible right now.

The Google Search Console pattern that tells you it's AI Overviews, not a penalty

If your clicks dropped but your impressions held steady or went up, the cause is almost always AI Overviews, not a Google penalty. Penalties knock impressions down. AI Overviews keeps you "visible" in the search results page but answers the question above your link, so nobody clicks.

Here is how to confirm it in 5 minutes. Open Google Search Console, go to Performance, and set the date range to the last 16 months. Compare the first half to the second half. Look at three numbers: impressions, clicks, and average position.

The AI Overviews fingerprint:

• Impressions: flat or up
• Average position: flat or slightly improved
• Clicks: down 30 to 60 percent on informational queries

A penalty looks completely different. Impressions fall hard, position drops, and clicks fall along with them. If your position is still page one and your impressions are healthy, Google is showing your page. The buyer just never gets to it because the AI Overview at the top of the page already answered the question.

Run the same check on a query-by-query basis. Sort by clicks lost. The queries that lost the most are almost always question-shaped: "how to," "what is," "best way to." Those are the exact queries AI Overviews loves to answer. Your transactional queries ("life coach Toronto," "executive coach NYC") usually held up because Overviews is less aggressive on local commercial intent.

This is the part most coaches get wrong: they see the drop, panic, and start publishing more "how to" posts to win back the traffic. That is the worst possible response. You are now feeding the exact content type that AI Overviews scrapes and re-serves without a click.

Why publishing more SEO content makes the problem worse, not better

The instinct to publish more is wrong because AI Overviews rewards comprehensive answers by quoting them inside the search page and discarding the link. The better your post answers the question, the less the reader needs to click through.

Think about what AI Overviews is actually doing. It reads the top 5 to 10 ranking pages, pulls the cleanest answer, and stitches it into a 3-paragraph response with small inline citations that almost nobody clicks. If your post is the cleanest answer, congratulations, you just got scraped. The reader has their answer. They are not clicking your link.

The old SEO playbook was a positive feedback loop. Rank, get clicks, build authority, rank for more. The new dynamic is a negative one for "how-to" content. Rank, get scraped, lose clicks, lose engagement signals, slip in rankings, get scraped less but also seen less. You are working harder to make Google's product better.

This is why "SEO is dead" gets repeated so often. It is not literally dead. Branded queries still work. Local commercial queries still work. But the bulk of the long-tail traffic that filled coaching websites between 2018 and 2023 is now eaten upstream. I wrote a longer piece on what's actually replacing SEO for buyer discovery if you want the bigger frame.

Picture Priya, a life coach in Toronto working with second-gen South Asian women navigating career and family expectations. Her site ranked on page one for "how to set boundaries with immigrant parents" and used to bring 400 organic clicks a month from that query alone. By early 2026, impressions were higher but clicks had collapsed close to zero. The AI Overview was answering the question in 3 paragraphs and citing 4 sources, hers included, in a small footer link almost nobody touched.

Her response was not to publish more. It was to change where she showed up. Buyers were still researching her topic. They had just moved the research one layer earlier, into ChatGPT, Perplexity, and the AI Overview itself.

The 3 fixes, ranked by effort and impact

Here are the three responses that actually work, ranked from highest impact to lowest. Most coaches do them in reverse order, which is why most coaches are still losing traffic.

Fix #1: Get cited by AI engines directly (highest impact)

This is the move. Instead of fighting AI Overviews for the click, get cited inside the answer so the buyer sees your name at the moment of recommendation. When a buyer asks ChatGPT "who's a good life coach for women in tech," you want your name in the response.

This is what The Citation Cluster Method does. It is the system I used to get BakingSubs (a niche site, no ads, no backlinks, no social) to 144,321 Microsoft Copilot citations across a single quarter. The mechanism: publish a tight group of posts that all answer related buyer questions about one specific topic, structure them so AI engines can pull a clean answer, and back them with clear author signals so engines trust the source.

The shortest version of how to start:

1. List the 8 questions a buyer asks an AI engine before hiring someone like you. Not topic ideas. Actual questions, in the actual words buyers use.
2. Write one focused post per question. Each post opens with a 2-sentence direct answer, then expands.
3. Add a real Person schema block to your About page so engines know you are a specific human with credentials.
4. Wait 6 to 12 weeks. Citation volume builds slowly, then snowballs.

The longer walkthrough is here and the case study breakdown is here. If you want to see where you currently stand, run the free AI Visibility Check. It asks 8 buyer-shaped questions across ChatGPT, Claude, Perplexity, and Copilot and tells you whether you are invisible, mixed, winning, or in an empty niche.

Fix #2: Shift content from "how-to" to "who/which/should I" (medium impact)

The second move is changing what you publish. AI Overviews loves to scrape "how to set boundaries" type posts. It is much weaker at "should I hire a life coach or a therapist," "which type of coach helps with X," "what to expect from a first coaching session." Those queries require judgment, not just information, and engines are more likely to surface a real recommendation rather than synthesize one.

This is the content type that still earns clicks. It also happens to be content that converts, because the reader asking "should I hire a coach" is much closer to buying than the reader asking "how to set boundaries." The full breakdown for life coaches is here and the same approach adapted for health coaches is here, where the credibility bar is higher because engines weight health-topic signals harder.

Fix #3: Diversify off-Google traffic sources (lower impact, but worth doing)

The third move is reducing your dependence on Google entirely. Not because Google traffic has no value, but because if 80 percent of your pipeline comes from one source that just got eaten by its own AI feature, you have a concentration risk. The 5 client-getting channels that still work for coaches in 2026 covers the realistic options. The short version: podcast guesting, referral partnerships, and being cited by AI engines (which is fix #1 again, because it is that much bigger than everything else).

I list this third because it takes the most effort for the smallest near-term return. It matters. But fix #1 is 10x the impact for the same hours of work.

What this looks like for someone who is not a coach

The same pattern is hitting consultants and expert-led ecommerce too, because the mechanism is the same: AI eating informational long-tail queries.

James is a workplace mediator in Manchester. He used to get a steady trickle of organic traffic from posts like "how to handle a colleague who undermines you in meetings." That traffic mostly evaporated in late 2025. What replaced it: when HR leaders ask ChatGPT "who handles workplace mediation in the UK for senior teams," James now shows up in the answer because he restructured his site around 8 buyer-shaped questions and added clear author signals. The mechanism that worked for Priya works for James. It also works for a Shopify founder running a specialist tea brand whose category posts got scraped: they shifted to "which loose-leaf is best for X" recommendation content and started showing up in Perplexity answers when buyers compared brands.

The mechanism does not care what you sell. It cares whether AI engines can find a clear, specific answer attached to a credible human source.

Frequently asked questions

How do I know if my traffic drop is from AI Overviews or a Google penalty?

Check Google Search Console for the last 12 to 16 months. If impressions are flat or up while clicks fell, that is the AI Overviews signature. If impressions dropped along with clicks, that is more likely a ranking issue or a penalty. AI Overviews leaves you "visible" in the results page but answers the query above your link, so the reader never clicks through.

Will publishing more SEO blog posts win the traffic back?

Usually no, and often it makes things worse. AI Overviews scrapes the cleanest answer it can find on the page and serves it without a click. The better your "how to" post, the more likely it gets quoted in the answer rather than clicked. Shift to "who/which/should I" content that requires judgment, and put your real work into getting cited by AI engines directly where the recommendation moment actually happens.

How long does it take to start getting cited by ChatGPT or Perplexity?

For most coaches starting from zero, the first citations usually appear in week 4 to week 8 after publishing a focused cluster of 6 to 10 posts with clean author signals. Volume builds slowly at first and then snowballs as engines crawl deeper into the cluster. BakingSubs hit 144,321 Microsoft Copilot citations across a single quarter, but that came after roughly 12 months of building.

Is SEO completely dead for coaches in 2026?

No, but the long-tail informational SEO that fed coaching sites between 2018 and 2023 is largely gone. Branded queries, local commercial queries, and "should I hire" decision queries still drive clicks. The longer answer is here. The bigger shift is that buyer research has moved one layer earlier, into AI engines, so the work has to move there too.

Do I need backlinks or a big social following to get cited by AI engines?

No. BakingSubs has no meaningful backlink profile and zero social presence. What AI engines look for is a clear topic focus, clean answer structure, and credible author signals (a real Person schema, named credentials, a consistent identity across the site). Those things matter more than domain authority for citation work.

What to do this week

Open Google Search Console and confirm the pattern. If impressions held but clicks fell, you are not broken and you are not penalized. You are running the right playbook for the wrong year. Run the free AI Visibility Check to see how you currently show up across ChatGPT, Claude, Perplexity, and Copilot for the 8 questions buyers actually ask before hiring someone like you. The result will tell you whether the work is "fix what's already there" or "start a fresh citation cluster." Either way, the next move is no longer guessing.