By Charles Barsley — Senior Digital Analytics & Optimisation Manager at Selfridges
I’ve used Attraqt Fredhopper now for over three years, in two different companies. Perhaps uniquely, my roles have covered being both the Fredhopper “Subject Matter Expert”, and my company’s Digital Analytics Manager. Fredhopper asked me to discuss how I have brought these two roles together. My aim is to promote the use of analytics for the optimisation of onsite search with Fredhopper.
In this blog, we’ll explore advanced tagging.
In my previous blog I mentioned several techniques for optimising search. Most ‘Out of the Box’ analytics tools will allow you to pull reports for all of the techniques I mentioned. However, Fredhopper offers several unique search functions whose effectiveness you can only measure with custom tagging. If you feel that you’re on top of all the basic search optimisation techniques, then you may be ready to try these unique functions. And if you do, you’ll want to analyse how they are working and what effect they have on the customer experience.
If you redirect a search term (e.g. ‘DVDs’ to a category landing page), is this term still tracked as a search? Often with many systems, the user’s search is no longer available on the redirected page, so the analytics tool doesn’t record that a search took place. One simple way to overcome this is to add a unique parameter to the redirect URL in Fredhopper (e.g. redirect=dvds).
In your Web Analytics tool, create a URL report filtered on contains “redirect=” and this will give you a list of search terms that are being searched but redirected. This can be as important to check as your top search terms, since websites are never static. Over time, landing pages and campaigns change. So are these the most relevant, best experiences or is this a landing page from three years ago? Also, if you have redirected many of your top searched-for terms, the wider business may only be receiving the standard search report and believe that customers aren’t searching for these key terms.
On a related note, if there is only one match for a search result, do you use Fredhopper’s Redirect to Product functionality to directly take the customer to the product page? By default, this could be missing from your onsite search report, as often the analytics search tag doesn’t fire from the product page. But it shouldn’t be a problem getting it added in. Once you do have it, it’s worth reviewing the top searches with just one result. It may be that some products are getting high traffic but are not what the customer was searching for at all. For example, they may be searching for a brand, but you only sell a book about the history of that brand, this is going to leave the customer disappointed and perhaps confused.
I’m a big fan of the Fredhopper Spell Correction engine! This saves merchandisers hours of time creating synonyms for every possible misspell. However, do you track what it does? Check your analytics to see. If you search for a term, do you track what you typed, or the spell correction that Fredhopper presented to the customer?
If you can, set up a new custom analytics attribute which contains “searchedterm|correctedterm”. Being able to see these side-by-side might shock you. Probably 95% of the time, you will get perfect spell corrections, but the remaining 5% is when the search really should have zero results. It might be a brand name you don’t sell or a trend that you don’t mention in your copy, but Fredhopper will attempt to spell correct these, which may result in a different brand name or word entirely. For this, you should consider what strategy you should use for zero results. You could just redirect these terms to a “No Results page”, or if there is enough volume, create specific landing pages to “Go in-store” or “Try these brands instead”.
If you use the Search Suggestions function and show search terms / products / categories / brand suggestions as the user types, do you record clicks on these? Ultimately, they still result in a search. So you would have them in your main search result report. But tracking these with a separate attribute to highlight the search method can be helpful. Occasionally, I see searches spike for a term for which we have weak results. Tracking this back, I can identify that this is because we are surfacing it as a suggested term after the user started typing. Blocking this from the Search Suggestions list will improve customers’ search experience, as they won’t be distracted to click on a search term that you don’t have a strong product range to back. They will instead focus on their search. An example of this is the case in which a brand name appears in the suggested search terms, but you only sell a single, small piece of their jewellery. It may be better to stop the brand name from appearing and only show “brand-name jewellery”, so that users don’t feel they are misled by your site.
I hope there are a few nuggets in here that will help you in your regular review and analysis of your onsite search. For best results, build a search strategy that works for your business and check incrementally and often.
Want the basics? Read my blog about Search Analytics.
