Skip to content

Finding keywords to prioritise for improvement

If you run search advertising campaigns, you’ll probably be looking at what the visitors you’ve purchased are doing on your site. If you advertise against searches on ‘blue widgets’ and people click through at 10p a time, it’s cheaper than advertising against searches on ‘red widgets’ and getting people to click through at £10 a time – but it’s not better if the first group do nothing and the second group all make an enquiry. So can we do the same analysis with natural search, i.e. the product of our SEO efforts?

Once upon a time it was easy, as Google Analytics had a ‘keyword’ section which allowed you to track visitor actions based on the search term they’d typed into Google. That was removed several years ago for privacy reasons, and it’s much missed. But we can still get some interesting pointers in that direction.

One exercise starts with looking at the pages on the site which are ‘converting’ the best. Now, ‘converting’ means different things to different people: it could be making a sale, but for most of us it’s more likely to be reaching the ‘contact us’ page …or even just spending time on the site. That’s up to you. On one site I was just looking at, we set up a ‘segment’ of people who spent time on the site or looked at more than one page (i.e. showed at least some degree of interest). Then we looked at the ‘Landing Pages’ report in Google Analytics, filtered to show just Google Search visits, to compare the proportions of visitors starting at each page who showed interest in the site. We could see that for every 100 visits entering the site on page A, only 25 entered the site on page B. However, 90% of the visits entering the site on page A didn’t go any further, whereas most of the visits entering the site on page B went further, so the absolute numbers of good visitors were higher for page B.

Then we looked at the two pages in Google Search Console, and discovered that in each case, the vast majority of visits from Google Search had been generated by a basic search (different for each page, of course). So we now had a reasonable link between a search term and the ‘quality’ of visit; page B’s search term was better than page A’s. Also, we could see there were more searches for page B’s search term than page A’s – we were only getting fewer clicks because we were appearing much lower down in the results.

So from that analysis, we could predict that a concentrated effort on doing better in the search results for page B’s search term could produce some significant results – more clicks, and ones which were likely to result in better quality visits too. The data was all there, we just had to know where to look.