Precision testing for optimising search: road-testing the new Tuition Fees site
When you launch a new website, you can’t help but feel a sense of protective affection for it, and an anxious wish that it will do well in the world. Here’s how we measured the new Tuition Fees site for search optimisation – and what we found.
Top tasks and search results
Shortly after launching the new Tuition Fees website, I looked at how well the site was doing in the University’s search results. I wanted to see if the top tasks we’d identified as being important to the prospective student experience were being supported by the results found using common fees-related search terms.
I was especially curious to see how my investigation would compare to the audit we did previous to the site launch.
Sprint 7 review: a new dedicated fees site structure
Google vs Funnelback
Thankfully, the University no longer uses Google to power its search engine. Since Google’s algorithms for what floats to the top tend to be capricious at best and unknowable at its darkest, this wasn’t particularly easy to work with when we wanted certain pages to do well. These days the University uses Funnelback, a tool which can be managed to direct search engine results more precisely.
So, how do you measure search performance?
It’s actually pretty easy.
- To Google Analytics! Look at the top 100 or so search terms people are using within the relevant space.
- Make a small list of the search terms relevant to the area your site is addressing. (I looked for terms like ‘fees’, ‘tuition fees’, ‘tuition’, etc.)
- Type in your search term to the University’s search bar and see what comes up. Look at the top five results.
Learn more with in-house Google analytics training
Testing your results for relevance
Your top five results may vary wildly in relevance. Users might land on:
- Your top task pages.
- Pages you don’t manage but which nevertheless pertain to the user task.
- Bonkers pages that shouldn’t be ranked as highly as they are.
Our team uses John Ferrara’s precision mapping technique which he outlines in the blog, A List Apart. We use the spreadsheet he generously provides which is pre-programmed with calculations showing how the metrics for the results work out.
A List Apart: Testing Search for Relevancy and Precision
I mapped my results using this score:
- Relevant – Completely relevant, hooray!
- Near – Not a perfect match, but reasonable for it to be ranked so highly.
- Misplaced – It’s got words in the content that are bumping it up the search rankings, but it shouldn’t be ranked so highly.
- Irrelevant – Absolutely bonkers.
What do I do if my search results are wild?
First, check the content itself. Your page should have:
- A clear title
- A summary which describes what is in the page using keywords
- Sub-headings which describe what is in each subsection, with one clear idea per paragraph
- Content which uses clear, plain language and helps your reader to achieve their intended goal
Sometimes results come to the top of the University’s search because they are excellently written – not at all helpful when they’re not relevant and they’re pushing your pages further down the rankings.
Watch the Effective Digital Content videos on Media Hopper for more about writing
My pages are well written – so why are they still ranking poorly?
If your pages are written well for search optimisation, you might want to get in touch with the Website and Communications team, which manages Funnelback and might be able to help manage where less relevant pages than yours appear in the rankings.
If one of my baby’s (cough) web pages falls down in the rankings and I can’t work out why, that’s precisely what I’ll be doing.
Contact the Website and Communications team
Hey, the Tuition Fees site works! (What I found)
When I mapped our search results for relevancy and precision, this is what I found – the new Tuition Fees site was absolutely fine. The majority of search terms returned relevant pages; a couple of slightly left-field results but nothing to be too concerned about. Not a single irrelevant result. The site was ok. We heaved a sigh of relief and patted ourselves on the back for writing our content well to begin with.
The takeaway
Write. Good. Content.
Always write good content. If you do this, and if you regularly check in using the technique I’ve just shared with you, you will probably never need to worry about tweaking the search engine itself. Good search optimisation is about good writing. There is no big mystery, and I don’t know about you but I find that very reassuring.
