How to use BayEstimator for Best Match

Best Match takes into account many factors, but this article looks purely at how the “BayEstimate” value of keywords are calculated.

For those that haven’t noticed the BayEstimate tool has been reinstated, but with varying size dots to represent the positive or negative impact of words in your title (instead of percentages). Even if you’ve used the tool before it’s worth reading the instructions and FAQ as it’s changed yet again since it’s relaunch two days ago on the 21st April.

For the purposes of the article I decided to use the Flip Video Camera I’ve just received.

How to generate a keyword rich title using BayEstimate

The Flip is not a product that I’m familiar with and I’ve done no research on good keywords so it was a great way to show BayEstimator tool at its best.

  Starting off with the product name “Flip Video” and the colour “Orange” I entered them into the BayEstimator. That gave four search keyword options, so I picked the most relevant to my title which was “Flip Video”. The BayEstimator then presents an ordered list of Keywords buyers like/dislike when searching for “Flip Video” shown left.

Instantly without knowing too much about the product I can quickly add in keywords to build the title “FLIP VIDEO ULTRA DIGITAL CAMCORDER 2GB 60 MIN ORANGE”.

Ultra is the model of the camera and although I missed that BayEstimate didn’t. Interestingly “60 MIN” is a better keyword selection that “60 MINUTES”, which is the record time for this model. This can be shown by the title I’ve generated with BayEstimate and the second title which is a live listing on eBay.

Using the word “Digital” in favour of “Minutes” makes a difference to the title’s overall attractiveness to Best Match. It should be realised that as Best Match learns which keywords are associated with the search term, in this case “Flip Video”, it will start to disadvantage titles which contain “Minutes”, even though it’s a valid descriptor.

Another interesting point is that the word “New” is more important than the colour “Orange”. Using a gallery picture buyers can see that your product is orange so omit it from your title and use new instead.

Finally if you are going to use keywords that will disadvantage your title in search, BayEstimator will clearly indicate which words you should omit from your titles.

How BayEstimate and Best Match work

Each keyword in eBay item titles is given a “desirability percentage”, although BayEstimator now only displays different sized/coloured dots indication rather than a percentage. Reading eBay’s patent application for Best Match (officially known as “Determining relevancy and desirability of terms.”) reveals more about how Best Match desirability is calculated.

The desirability percentage is calculated based on the most clicked keywords in item titles in relation to the search term (e.g. for the purpose of the exercise above it should be assumed that the buyer searched for “Flip Video”.) Having been presented with the search results Best Match then determines which additional keywords in the title are most likely to best describe the item the buyer is searching for.

When a user searches eBay notes the words in every title of every listing returned in the result and especially the words in the title that users tend to click on. For example if in 1000 searches for “Flip Video” titles including the word “Camcorder” were clicked on 900 times then “Camcorder” would have a desirability of 90% when coupled with “Flip Video”. Equally the product comes with a case, if the word “Case” was in just 50 titles that buyers clicked on then it would have a desirability of just 5%.

On a rolling basis eBay can keep track of which keywords are considered by buyers as highly relevant when coupled with search terms. When scanning a page of titles if buyers eyes are drawn to one particular word (and tend to click on titles containing that word) then it will be considered highly relevant. Equally if a keyword is present in a large proportion of titles on the page, but is selected much less frequently then it will be considered undesirable.

The effect of this would be when searching for “Flip Video”, titles including “Records for 60 minutes” would probably not appear at the top (and maybe not even on the first page) of search results.

Conclusions

The return of the BayEstimate tool has given sellers the ability to fine tune their titles and under Best Match this becomes more important than in the past. Previously regardless of title your listings would rise to the top under ending soonest. With Best Match your listings will live or die with judicious keyword selection. Also, as buyers will automatically modify Best Match desirability by clicking on search results, sellers will need to modify titles on an ongoing basis to ensure they’re remain at the top of the relevancy tables.

Sellers should also remember that keyword selection in item titles is just a part of the Best Match algorithm. Feedback, feedback DSRs, the length of time left to run on the listing, geographic proximity to the buyer, the number of bids already made, Auction/Buy It Now, INR disputes and even pictures can affect your items ranking.

Best Matched title keywords are a part of the picture, but if your titles are better matched than competitors, all other things being equal, then your listings will float to the top above theirs in search.