CTR Underperformers: Which Titles to Rewrite

Full transcript

1,250 words · script draft, under review.

Intro

Two pages. Same position on Google. Same kind of search. One gets clicked all the time. The other is shown just as often, and almost nobody clicks it.

You did the hard part. You earned the ranking. And then you lose the visitor in the last half second, on a line of text you probably wrote in a hurry.

In this episode I'll show you how to find those pages with your own data, using one of my own sites as the example, how to rule out the causes a title can't fix, and how to rewrite and measure the ones that are worth it.

Build your own CTR curve

Click-through rate, or CTR, is clicks divided by impressions. The obvious question is: what's a good CTR at position three? You'll find industry charts that answer it. I don't use them.

Every site is different. The mix of queries, the brand, the features on the results page. All of it moves the curve. So instead, I build the curve from the site's own data.

It's simple. Take all queries in the top ten. Round each one's average position to a whole number. For each position, add up all the clicks and all the impressions, and divide. That gives you a weighted average CTR per position, for this site only.

Then the flag. Any query that gets less than half the CTR the site normally gets at that position is a candidate. Half is deliberately crude. You're not looking for small differences. You're looking for results that are clearly being skipped.

And set an impressions floor, so you don't chase a query with twelve impressions and zero clicks. Small numbers swing wildly.

Real data: under half the average

A quick note on the data. It's real, from one of my own sites, pulled from the Search Console API. I've anonymized it: the domain, the URLs and the search terms are replaced, so we can stay focused on the principles and not on the keywords. The clicks, impressions and positions are exactly what Search Console reported.

The dashed line is this site's own average click-through rate at each position. I'm not using an outside benchmark, because every site is different. Each dot is a query that shows up in the top ten with at least eighty impressions.

The amber dots get less than half the click-through rate the site normally gets at that position. Nine of twenty-seven. Same position, same kind of page, far fewer clicks. Something about how these results look is not working: usually the title, sometimes the description.

The labels are placeholders, so I won't name the queries. Look at the pattern: the amber dots cluster below the line, and the biggest one, number one, had two hundred and eighty-seven impressions and only seven clicks. Treat that as a place to look, not as a verdict. Other things can pull clicks away, and with a sample this small the curve is rough. But it tells us which titles to open first.

Check the results page before you rewrite

Before you rewrite anything, check which page each flagged query lands on, and look at the actual results page. A low CTR has several causes, and only some of them are about your title.

On my site, four of the nine flagged queries land on the home page, including the biggest one. That's worth checking first. A home page shown for a specific search is often the wrong kind of page for it.

So I pulled a live results page for all nine. One desktop snapshot from one day, compared with a twenty-eight day average from Search Console. Keep that difference in mind.

The biggest one, number one on the chart, averaged close to position six in Search Console. In today's snapshot it's eighth among the regular results. Above it: an AI-generated summary, and a box of questions people also ask.

Across all nine, an AI summary sat at the top every time. A box of questions appeared on six. Those boxes take up room: the snapshot only reached about nine regular results per query.

And on three of the nine, my own site was cited inside the AI summary itself. So the searcher may have got the answer, quoted from my page, without clicking. That's a lost click a title can't win back.

Then features. When a summary, a box of questions, a video row or a map sits above you, your result may land below the fold. It's shown, and counted as an impression, but few people scroll that far. A new title won't fix that.

Intent. Maybe the searcher wants something your page isn't. They want one specific answer, and you show them an overview. They want a tool, and you have a table. The title isn't the problem. The page type is.

Brand. If the query contains another company's name, people are looking for that company. Low CTR is normal there. We'll split brand from non-brand properly in episode fourteen.

And one more: check what Google actually displays. Google sometimes rewrites titles. The text in your HTML may not be the text people see. The live results, by hand or through the API, show you the real one.

Rewriting the title

If the results page looks normal and the intent matches, now it's the title. On screen is the shape of a typical fix, as a template rather than a real title.

Use the searcher's own words. Look at which queries bring impressions to this page, and make sure the main phrasing is in the title. Not repeated, just there, and near the start.

Make a specific promise. What will they get on this page that the result above doesn't offer? A full list, a current number, a comparison, a calculator. If it's true, say it.

Keep it short enough not to be cut off. And look at the description too. It isn't a ranking factor, but it's part of what people read before they click.

One more rule: don't promise more than the page delivers. A title that oversells may win the click, but the visitor comes straight back to the results and picks someone else. You've gained a number in a report and lost the person. And a title that doesn't match the page is exactly the kind Google tends to rewrite anyway.

Change one thing at a time, on a handful of pages, and log every change with the date. That log is what makes the next step possible.

Re-measure the right way

After a few weeks, compare. But not just the raw CTR.

If the page also moved in position, its CTR would change anyway. So compare the query against the site's curve, before and after. Say it was at forty percent of the expected CTR, and now it's at ninety. That's the title working.

Use equal periods, and the same queries. And keep a few flagged pages unchanged as a control. If the control pages improve just as much, the change you made isn't the reason.

And if nothing changed? That's information too. Go back to the results page. The cause was probably one of the things a title can't fix, and the next step is the page itself, not another title.

The full method for before and after measurement comes in episode forty-six.

Wrap-up

So: build your own curve, flag what's under half, check the results page, then rewrite and re-measure. The CTR curve script and the title change log are linked in the description. Next episode: when two of your own pages compete for the same search, how to detect it, and how to fix it.