Your Rankings Held. Your Clicks Died. Refreshing That Post Will Not Save It

You open Search Console. Traffic is down. Not off a cliff, just a slow bleed, month after month, on the posts that used to pay the bills. So you do what every blog on page one told you to do. You add 400 words, change the date to 2026, hit update and wait.

Nothing happens. Or it gets worse.

Here is why. You diagnosed one disease and there are three. Most content decay advice was written for a search engine that no longer exists, where losing traffic meant losing rankings. In 2026 you can hold position one and still lose most of your clicks, because Google now answers the question above you and the searcher never scrolls. Refreshing the post does not touch that problem. It is a different problem with a different fix.

This is the diagnosis first, treatment second version. Ten minutes in Search Console tells you which of the three you have. Then you fix the right one.

What content decay actually is

Content decay is a page that used to perform in search and now performs worse, gradually, without you touching it. It is not a penalty. It is not a core update slapping you. It is the slow erosion that happens because the web around your page kept moving and your page did not.

The classic version is simple. You published a guide in 2023. Three competitors published better ones in 2025. Google re-evaluated, they won, you slid from three to nine, and the clicks went with you.

That version still exists. It is just no longer the most common one.

The three kinds of decay, and why the usual advice only fixes one

Every page losing search traffic is losing it for one of three reasons. They look identical in a traffic graph and completely different in Search Console.

Nearly every guide on this topic treats all three as one thing and prescribes the same medicine: update, expand, republish. That works on rank decay. On click decay it is worse than useless, because you have just spent a day making a page longer for an audience that is never going to load it.

How big is click decay, really

Big enough that it is now the default explanation, not the exotic one.

Ahrefs ran the numbers in April 2025 and found that the presence of an AI Overview cut clicks to the top ranking page by 34.5%. They re-ran the same study with December 2025 data and published the update in February 2026. The number had moved to 58%. Same methodology, 300,000 keywords, half with an AI Overview and half without, measured against aggregated Search Console click-through rates.

It is not just position one. Their position by position table is worth pinning to your wall.

Read that again. The better you rank, the more an AI Overview costs you. Ranking first is now the position that bleeds hardest.

And this is not one company with an axe to grind. Ahrefs list the corroborating numbers themselves: Seer Interactive measured a drop between 49.4% and 65.2%, Kevin Indig put it above 50%, Authoritas at 47.5%. Different panels, different methods, same direction.

Zoom out and it is the whole search result, not just AI Overviews. SparkToro and Similarweb looked at US clickstream data for January to April 2026 and found 68.01% of Google searches ended without a single click. In 2024 that figure was 60.45%. Clicks of any kind fell 9.51 percentage points in two years, a 22.9% decline. Ahrefs put AI Overviews on more than 20% of all searches.

So when your page loses half its traffic and the ranking has not moved, stop blaming your content. Your content did not decay. The result page did.

The uncomfortable bit: most of the sites bleeding right now have no content problem at all. They have a business model problem, and a new blog post will not fix it. More on that further down.

Diagnose it in ten minutes with Search Console

You need no tool for this. Free Search Console, five clicks.

Step 1: get the comparison right

Performance report, Search results. Date range, Compare, Custom. Set the last 6 months against the previous 6 months. Not 28 days. Decay is slow and a monthly window is all noise and seasonality.

Step 2: sort by what you lost, not what you have

Switch to the Pages tab and sort by the Clicks difference column, worst first. You now have your patient list. Most sites find that five to fifteen URLs account for the overwhelming majority of the loss. Fix those. Ignore the long tail for now.

Step 3: read the three columns together

For each losing page, look at clicks, impressions and average position side by side. This is the entire diagnosis:

• Clicks down, impressions down, position worse. Rank decay. Somebody beat you.
• Clicks down, impressions flat or up, position roughly unchanged. Click decay. You are being shown and skipped.
• Clicks down and the top queries are not the queries you used to win. Intent decay. The market moved.

Step 4: check the generative AI report

Google launched the Search Generative AI performance report in Search Console on 3 June 2026 and rolled it out worldwide by 31 August. It shows impressions, pages, countries, devices and dates for AI Overviews, AI Mode and generative results in Discover, split out from the classic web report.

It does not give you clicks. It does not give you click-through rate. It does not give you the query. Google has said more metrics will come and has not said when. Use it for what it is: proof that the AI surfaces are showing your page, which combined with flat classic impressions and falling clicks is the signature of click decay.

Step 5: actually look at the search result

Open an incognito window and search the query. Is there an AI Overview? Does it answer the whole question? Is your site cited in it? Thirty seconds of looking beats an hour of theorising. We walk members through this exact audit, page by page, inside the Dojo, because the diagnosis is the part everybody skips and it is the part that decides everything else.

Fixing type one: rank decay

This is the straightforward one, and it is still worth doing, because a page that already has history, links and impressions is far cheaper to lift than a blank page is to launch.

The order matters more than the effort:

1. Open the current top three results and list what they cover that you do not. Not word count. Coverage. Sub-questions, comparisons, a calculator, a table, current pricing, a named author who clearly did the thing.
2. Fix the factual rot first. Prices from 2023, tools that shut down, screenshots of interfaces that no longer look like that, statistics with no year on them. This is the single highest ratio of impact to effort in the whole job.
3. Rewrite the opening 60 words as a direct answer to the exact query. Question as the heading, answer immediately underneath, no throat clearing.
4. Add the sections you are missing. One new section that answers a real sub-question beats a thousand words of padding.
5. Re-link internally. Point three or four relevant pages at it with descriptive anchor text. Most decayed pages have quietly become orphans.
6. Only then touch the date. Change the publish date when the content genuinely changed. Doing it when nothing changed is how you teach Google your dates mean nothing.

Fixing type two: click decay

Different disease, different medicine. You are not trying to rank better. You already rank. You are trying to become either the thing the machine quotes, or the thing worth clicking anyway.

There are exactly three plays and you should run all three.

Play one: get cited inside the answer

If the AI Overview is going to eat the click, be the name inside it. That means writing the page in blocks a machine can lift cleanly: a question as a heading, a tight self-contained answer underneath, no dependency on the paragraph before it. Long meandering prose does not get quoted because no single chunk of it stands alone. We covered the mechanics of this in how AI reads one chunk of your page, not the whole thing, and in the AI Overviews playbook.

Play two: change what the page is for

If a machine can answer the query in two sentences, that query was never going to be a business. So stop trying to win it with prose and give the page something an AI cannot hand over in a summary box: a calculator, a real template, a downloadable file, actual proprietary data, a tool, a quote form, pricing.

This is not a hunch. Cyrus Shepard at Zyppy analysed more than 400 winning and losing sites in April 2026 and measured which features correlated with year on year traffic change. The top of the list was not content quality. It was offering a product or service (0.391), letting a visitor complete a task (0.381), and owning proprietary assets (0.357). Tight topical focus (0.250) and brand strength (0.206) came after.

Translation, in plain English: Google is sending its remaining clicks to sites that do something, not sites that explain something. If your decayed page only explains, that is the decay. Give it a job.

Play three: stop counting the wrong number

Sessions from Google is dying as a metric whether you like it or not. The traffic that is replacing it arrives from ChatGPT, Perplexity, Gemini and Copilot, usually pre-sold, usually converting better, and usually filed by your analytics under Direct where you will never see it. Fix your attribution before you conclude a page is dead: here is how to pull AI referral traffic out of Direct in GA4. Plenty of pages that look decayed are quietly bringing in the best leads on the site.

Fixing type three: intent decay

The rarest and the most expensive to get wrong. The query still has volume. The people typing it now want something else.

Look at the Queries tab for that page, compared against a year ago. If the winning queries have changed shape, from "what is X" to "X pricing", or from "best X" to "X alternatives", your page is answering a question nobody is asking any more.

No amount of refreshing fixes that. You either re-point the page at the new intent, which usually means a new angle, new headings and a new primary keyword, or you split it: keep the old page for the old query and build a separate page for the new one. Guessing between those two is how sites end up with three pages fighting each other for the same term.

When to merge and when to delete

Not every decayed page deserves rescuing. Be ruthless. This is the table we use:

Deleting pages feels wrong after you paid for them. Write off the sunk cost. A site of 40 pages that all earn their keep beats 400 pages where 360 are dragging the average down.

How long it takes

Nobody can promise you a position or a date, and anybody who does is selling something. What is reliable is the direction: fixing the right disease moves the number, fixing the wrong one does not.

What it costs

Ranges are what we see in the market, not a quote. Prices vary wildly by country and by how much of the fee is account management.

Who this is for, and who it is not

Do this first if: you have 20 or more pages with search history, your traffic peaked at some point in the last three years, and you have pages that once ranked. Refreshing beats publishing for you, every time. A page with history is a running start.

Do not start here if: your site is under six months old with no impressions to compare. There is nothing to decay. Go and build the pages and the links first. Equally, if your business is local, your problem is more likely the map pack than your blog, and that is a different job entirely.

The tools you actually need

• Google Search Console. Free, and it is the only source of truth for what you lost. Everything else is an estimate.
• The Generative AI performance report inside Search Console. Free. Impressions only, but that is enough to confirm the AI surfaces are showing you.
• GA4 with a proper AI referral segment. Free, and most people have it configured wrong.
• Ahrefs or Semrush, optional, for the SERP feature filter so you can see which of your keywords trigger an AI Overview at scale.
• An incognito window. Free, underrated, and it beats every dashboard for working out why a specific page stopped getting clicks.
• Our tools for the on-page side once you know what you are fixing.

You will notice how much of that list costs nothing. The expensive tools tell you what happened. They do not tell you which of the three diseases you have. That is a judgement call, and it is learnable.

Five mistakes that make decay worse

1. Changing the date and nothing else. Users notice, Google notices, and you have burned the one signal that was supposed to mean something.
2. Adding length to fix a click problem. If an AI Overview is taking the click, a longer page gives it more to summarise. You are feeding the thing that is eating you.
3. Refreshing everything at once. Thirty pages in a week and you will never know which change did what. Batch of five, wait, measure.
4. Judging it after three weeks. Under 90 days you are reading weather, not climate.
5. Refreshing a page that should be merged. Two pages half-ranking for the same term do not add up. They cancel out.

The alternatives, honestly

Refreshing is not the only move, and it is not always the best one.

Publish new pages instead. Sensible if your site genuinely has gaps. Slower and more expensive per unit of traffic than fixing pages that already have history.

Chase striking distance keywords. Often better value than a refresh programme. Positions four to fifteen move with small pushes.

Give up on the click and play the citation game. Increasingly the right answer for informational content. You are optimising to be named, not visited.

Go and get mentioned elsewhere. Third party mentions and earned coverage feed the models far more than your own blog does. Your homepage bragging counts for nothing. What other sites say about you is the whole game.

Buy the traffic. Sometimes correct. Just do the maths honestly rather than using ads to paper over a content problem.

The risks

Refreshing is low risk but not no risk. The three things that actually go wrong:

You rewrite a page that was working for a query you did not know about. Check the full query list before you change the angle. Pages often earn from terms you never targeted.

You consolidate badly. Merging without proper 301s loses the links and the history. Map the redirects before you delete anything.

You optimise for a surface that changes. AI Mode was only 0.34% of searches in the SparkToro data for January to April 2026, yet Google said at I/O 2026 that it had passed a billion monthly users with queries more than doubling every quarter. Do not rebuild your whole site around one feature. Build pages that answer clearly and give people something to do. That works everywhere.

Where this goes next

The direction is not subtle. Zero-click went from roughly 45% a decade ago to 60.45% in 2024 to 68.01% in early 2026. Ahrefs track Google traffic share across more than 75,000 domains and it fell 8 percentage points between June 2025 and May 2026, around a 22% drop, for sites with professional marketers actively trying to stop it.

Planning for that to reverse is not a plan. What survives it is content that either gets quoted by name, or gives the visitor something they cannot get from a summary. Most of the web does neither, which is precisely why the window is open right now for anybody willing to do the boring diagnostic work first.

And this is the part worth sitting with: the fix for decay in 2026 is barely a content job. It is a positioning job. Give the page a function. Put your name on it. Make the answer liftable. Then let the machine quote you.

Do this today

Open Search Console. Last 6 months against the previous 6. Pages tab. Sort by click difference. Take the worst five URLs and label each one rank, click or intent. That is one coffee's worth of work and it will tell you more than any tool you were about to pay for.

Then fix the right one.

If you would rather not learn this the slow way, the full audit process, the decision tree, the refresh templates and the tracker all live inside the Dojo, alongside the rest of the system. Join The Dojo. $67/mo. Cancel whenever you like.

Want it done for you instead of learning it? That is what Done For You is for.