While some core features are shared by many search solutions, certain things make Sajari different. When we set out to build Sajari, the intention was to address the problems inherent in current search technologies. In doing so, we think we have made it uniquely capable in some important areas.
Some Sajari specific benefits include:
- Real-time reinforcement machine learning. To continually and dynamically optimize relevance, Sajari employs reinforcement learning by default for all customers. Within two to four weeks Sajari’s ML system will develop a clear understanding of which pages or records are most relevant to each of your users' queries. What’s more, you can gain visibility on how the system is being trained in the Learning section of your Console. The process can be sped up with ‘manual’ tuning as described below, but it’s the best thing to fall back on to constantly improve relevance.
- Dynamic query boosting. Sajari is engineered to allow users to run multiple algorithms on their data without using caches or seeing any negative effect on performance. Multiple algorithms can be run over the same data set at speed, e.g. one algorithm can be used for site search, while another produces dynamic content blocks tailored to different customer segments. This principle can be applied in any search or matching context, allowing for the creation of much richer interactions.
- Custom score creation, on-the-fly. As results scoring is performed dynamically as queries are run, Sajari allows different blends of the factors and weightings that score results to be used without rebuilding the entire index. This allows Sajari users to tune, optimize and iterate search performance much faster.
- Multi-dimensional indexing factors. Along with keywords, Sajari can take into account a plethora of business data all at once, for more nuanced relevance scoring. For example, in ecommerce search, Sajari customers can use different combinations of stock levels, profit margins, and conversion rates when ranking their search results, boosting product results accordingly. Most competitors only allow the use of these factors one by one to order results.
- Use records as queries. Queries don't have to be limited to keywords or phrases. You can use an entire document, user profile, or product as a query to search for similar items.
- Speed. We've spent a lot of time making our engine as fast and efficient as possible, for better UX. In tests against competitor search technologies, we've recorded about 10x faster indexing and about 100x faster searching.
We're always happy to chat about the benefits of Sajari and how they can help to your business. Drop us a line if you have any questions at firstname.lastname@example.org.