The meta description formula that gets clicks when AI Overviews take the top spot

The search result looked fine — ranked position four, title tag matched the query, meta description said something reasonable about the topic. The click-through rate was 0.8%. Not because the description was bad. Because the AI Overview above it had already answered the question.

This is the new reality for meta description clicks 2026 and beyond. The descriptions that worked in 2023 — the ones that summarised what the page contained — now compete against a generated answer that gives the summary for free. Your meta description isn't fighting other blue links anymore. It's fighting the instinct to stop scrolling entirely.

Why the old formula stopped working

The traditional approach to meta descriptions was straightforward: tell people what the page covers, include the keyword, maybe add a benefit. "Learn how to write effective meta descriptions that improve your click-through rate." Functional. Clear. And now, largely irrelevant.

When an AI Overview answers the query directly, a description that promises to explain the same thing offers no reason to click. The searcher already has the explanation. They're not looking for what you cover — they're looking for what they can't get from the generated answer.

The shift isn't subtle. Organic CTR for informational queries has dropped measurably since AI Overviews rolled out across most search results. The pages still ranking in positions three through ten are getting fewer clicks even when their rankings haven't changed. The meta description best practices from two years ago assumed you were competing for attention against other summaries. Now you're competing against the answer itself.

What click-worthy meta descriptions actually do now

The descriptions that still earn clicks share a pattern: they offer something the AI Overview can't. Not a summary — a reason.

That reason usually falls into one of three categories. First, specificity that generic answers lack. An AI Overview might explain what meta descriptions are and why they matter. A click-worthy description might reference "the exact character count that performed best across 2,400 tested pages" — something the searcher can't get without clicking.

Second, a perspective or framework. AI-generated answers tend toward consensus explanations. A description that signals a distinct approach — "why we stopped writing benefit-focused meta descriptions entirely" — creates curiosity that a summary can't satisfy.

Third, proof of depth. Not "comprehensive guide" depth, which means nothing. Actual signals that the page goes further than a generated answer can. "Includes the three meta descriptions that doubled CTR for product pages last quarter" tells the searcher there's something specific waiting.

The formula that's working for meta description CTR

Here's what's actually earning clicks in search results where AI Overviews dominate the top spot. It's not complicated, but it requires abandoning the instinct to summarise.

Start with what the AI answer didn't give them. Think about what the Overview probably covered — the definition, the general best practice, the surface-level advice. Your description should reference something beyond that. Not "learn about meta descriptions" but "the phrasing pattern that outperformed every other variation we tested."

Include a specificity marker. Numbers work. Timeframes work. Anything that signals this isn't generic advice pulled from the same sources the AI used. "Three templates from pages with 4%+ click-through rates" beats "templates that improve your CTR."

Keep it tight. The character count hasn't changed — 130-150 displayed characters is still the safe range — but every word matters more now. Cut anything that sounds like filler. "In this article, you'll learn" wastes six words saying nothing. Start with the value.

Examples that earn the click

Here's a weak description for a page about meta descriptions: "Learn how to write meta descriptions that get more clicks. Our guide covers best practices and examples to help you improve your SEO."

That description summarises. The AI Overview already did that.

Here's a stronger version: "The 3-part structure that lifted organic CTR from 1.2% to 3.8% on 47 tested pages — plus the one word that tanked performance every time."

The second version promises something specific the AI couldn't synthesise: actual test results, actual numbers, an actual warning about what not to do. It gives the searcher a reason to want the full page.

This matters even more when you're writing meta descriptions for brand-specific pages. Generic industry language sounds like every other result — and like the AI Overview that summarised all of them. When your description references your actual product names, your specific methodology, or results from your particular business, it stands out because it sounds like content from a real business, not a content template.

Writing meta descriptions in the AI era takes different input

The hard part isn't the formula. It's having specific enough material to put in the description. You can't promise "the exact phrasing that worked" if your page contains the same general advice as everyone else's.

This is where the content itself has to change. Meta descriptions are downstream of page quality. If the page is generic, the description will be too — and no formula fixes that.

For writers working with client brands, this creates a research problem. You need to know what makes this business's approach different, what specific results they've seen, what terminology they actually use. Without that, you're writing meta descriptions that sound like every other result in the SERP.

That's exactly the gap BrandDraft AI was built for — it reads the brand's website before generating anything, so the content references actual products and positioning instead of generic industry language.

The test for whether your description works

Read the description after imagining you've just read the AI Overview. Does it give you a reason to click anyway? If your immediate reaction is "I already know that" or "sounds like what I just read," the description fails.

The meta descriptions earning clicks in 2026 aren't better summaries. They're arguments for why the summary isn't enough.