Improving Search Performance
Ensuring we offer good search performance is a continuous process of improvement and refinement.
As our database of part/model numbers grow and as we welcome more customers to our website, the number of database queries increase and the complexity of the queries increase.
We aim to make our search as simple to use as possible. This means that the search has to work even if your search question is written with extra words that could easily cause your search to return no results. To achieve this. we have to query multiple database tables and columns of data through millions of rows to find you what you are looking for.
For example, searching for "RB403N4BC1" is very different to searching for "RB403N4BC1 handle" and "handle for RB403N4BC1" is a really difficult search query. Natural language searching is complicated to get right.
However, in each search case above, you are presented with the correct search results for that model.
We could make the search very targeted and therefore super fast, but if you don't type your model number exactly as we have it stored, no results are shown. However, that isn't always possible, what if your model number has been partially worn away from the sticker or as above you search with words like "door" or "hinge".
To give you the flexibility to search for partial part numbers and model numbers is a very complicated problem to solve without search performance becoming painfully slow.
You can throw money at any problem, we could have multiple dedicated servers to deal with search and then load balance between them, however, being a family run business, we are unable to pour vast resources at any problem we face.
Search is something that we have improved over time and today we have made another step forward.
We have improved search speed by 40% on average, which is a huge gain, when typically improvements are measured in the millisecond range.
We hope this speed up in search results being displayed makes your experience on the website better and more productive.