Vlad and Gus, forensic web analysts from the Web Detective Agency, are having a sandwich in the office and doing some online Christmas shopping.
Gus: It's that time of year I love when I spend all this money online buying stuff; shame I am not spending it on myself. You know I always end up buying presents for people that I really want for myself. Like last year, I got my sister a pair of wearable speakers. She wasn't impressed, I could tell.
Vlad: Hmmm.. Unlike you I hate buying presents, I never get it right, and I never get anything I want. I really don't see the point.
Gus: Well, you're cheerful! Why don't you see it as a competition, to see if you how close you get to the perfect present. What about Ineke ? What would she like ? Has she dropped any hints?
Vlad: She has, but I just can't find it online.
Gus: What does she want?
Vlad: It's something like a Daniel Fusterbucket dress.
Gus: And you saying you can't find it online? Well, have you thought visiting an actual store?
Vlad: Ineke is adamant that you can buy it at Bijenkorf , and she loves the packaging and the whole shop, in fact. But whenever I go into the shop I can never find anything, and on their website it just says "No matches found".
Gus: They must be annoying so many people with that "no matches found" message. You would think they would do something about that.
Vlad: Come to think of it, one of our clients, Hema, had the same problem. Before they invested in improving their search efficiency, they wanted to understand how much "failed searches" or those with 'no matches found' was costing them.
Gus: How do you calculate that?
Vlad: It's straightforward. Hema were initially quite sceptical, but in the end found the whole process of putting a value on a failed search very useful.
Gus: But how can a failed search have a positive value?
Vlad: That's the thing, you can't. Quite simply, a failed search is an example of a 'negative goal' - it's something you don't want to happen because each time it does, the business loses money.
Gus: Go on then, show me how you put a value on a failed search.
Vlad: OK, let's say Bijenkorf has 35,000 searches made on their website each month. 20% of these searches result in 'no match found' or are failed searches. We know from our analytics tool that the conversion rate for those with failed search is 0.5% and those with successful search, where a match is returned, the rate is 2%. Can you work out how much money they lose for each failed search?
Gus: Not without knowing their Average Order Value ?
Vlad: AOV is 125 euros.
Gus: OK ..if there were no failed searches we would have a revenue of 35,000 x 2% x 125, is 87, 500 euros. Right?
Gus: And the revenue from failed searches is 35,000 x 0.5% x 125, so that is 21,875.
OK, wait a minute. If I take 21, 875 away from 87,500 I am left with the extra revenue I would have got if all searches had been successful that month.
That's 65,625 euros.
Then if I take 65,625 and divide it by the number of failed searches that month, 7,000 (or 20% of 35,000), I can see how much failed search loses them.
And the answer is 9.375 euros.
Vlad: That's right, if these numbers were real, Bijenkorf would be losing over 9 euros for each failed search.
Gus: Wow, and that's discounting the repeat orders they would be getting from visitors who had used search successfully and had gone on to buy.
Vlad: Good point.
Gus: But what could can you do to reduce failed searches?
Vlad: There are several things you could do. The first is to benchmark the % of searches that fail, and then to identify which search terms are causing no matches to be found. Whilst you are doing your analysis on these search terms, the least you can do is to show your bestsellers where no match is found. Here, have a look at this sketch to see what I mean.
Once you know which search terms are causing the problems you need to update your CMS to capture misspellings and synonyms.
Gus: There might be items visitors who are searching for items that you don't stock anyway. I suppose if the volume of searches is pretty high then you might consider stocking these on a trial basis.
Vlad: I hadn't thought of that. Absolutely.
Secondly, you might think about using a dedicated search systems like SLI Systems, Nextopia or Celebros. They do clever matching on search terms in the background and then returning the most relevant matches based on previous visitor searches.
Gus: Interesting. But hold on, how do you capture the search terms that appear in failed searches. I don't think Google Analytics does that.
Vlad: Not at the moment, but it wouldn't surprise me if they introduced in their next release. Here's a link to a blog post I found on the Analytics Talk that explains what you need to do get failed search terms into a GA report.
Gus: Perhaps we could approach Bijenkorf and see if we could help them with their searches.
Vlad: Good idea. Let's finish our Christmas shopping first.
Not sure what website goals you should be setting and think you need help?
Read our ebook below for 8 questions you must ask to find, hire and get great results from CRO professionals.