What are the advantages and disadvantages of building search functionality in-house and using SaaS solution via their API? Joel Kall, our co-founder and Sr. Engineer answers this question here and on Quora.
As a SaaS-solution for e-commerce search, I’m going to reflect a bit on what you gain by choosing a SaaS solution instead of Open Source Software (OSS).
From my experience, the hardest thing about deploying a search solution for e-commerce is tuning it for relevance. Most OSS search engines have many tools for managing document scoring, synonyms and language features such as stemming, pluralization of words, spelling correction and other necessities that improve the end user experience.
With all this tooling, tuning the engine to provide relevant search results sounds like an easy task, but in reality it can be quite difficult. It is made harder by things such as complexity of language (words can mean different things in different context, or the same thing can be described by multiple words), individuality of users (many words have different meaning to different people) and bad quality of data (products are missing attributes, are described inconsistently, or simply has the wrong data). On top of this you have to take into account variations such as seasonal variations, product inventory variations and trends in language and demand.
What often happens when you set up an OSS search engine deployment is that you tweak its relevance to provide good results for the top 10–100 most common search queries. This means adding synonyms, tuning stemming and pluralization settings and make sure you have coverage for the most common spelling errors. But in many e-commerce stores, the top 100 search queries only make up a few percent of the total search volume, so most queries will not be optimized. So you need to dedicate a lot of resources to keep optimizing more and more queries, and while you’re doing this your inventory changes so you need to do it all over again. It is not uncommon for retailers to have at least one person full-time committed to tuning search relevance.
So while OSS search engines are good because they are flexible and have great community support, you often miss out on a lot of sales because the search engine is simply not made to optimize for conversion. For instance, they don’t have features such as learning from user behavior, product up-sell/cross-sell, personalized search results, conversion-aware spelling correction and autocomplete, related products etc.
If you opt for an e-commerce-focused SaaS solution instead, you get an engine with all these features and get to buy into a team of search experts that make sure that the engine is performing optimally when it comes to end user experience and sales. Plus you get benefits such as reduced maintenance and professional support, should you need it.
To learn more about the differences between an OSS and hosted search solution, check out this great Q & A on Quora. Or, alternatively, download our spec sheet to see what a hosted search solution can do.