The Smart Search feature allows users to find companies based on phrases, concepts, sentences and even complete paragraphs. This is a move away from keyword search and a drastic improvement to the discoverability of companies.
Introducing Smart Search
The Smart Search feature uses a Large Language Model (LLM) based approach of semantic search, matching the meaning and context of the search query to the meanings and context of entire web pages. It then selects the top (up to) 500 companies (where available) with web text that is most similar to the input search phrase.
Functionality
Once a list has been created using a Smart Search, it can be filtered down using the same filtering options as are available on the Explore Page. Or the list can be taken through to the Explore page itself to be further filtered.
When a list is filtered in-place (within the Smart Search page), a fresh “top 500” companies will be selected to the original list with the filters applied. For example, if a location filter is applied, such as “Leeds”, the list will now rank all companies based on their similarity to the search phrase, it will filter for “located in Leeds”, and then select the top 500 companies from this list. As such, applying a filter will not, necessarily, reduce the number of companies in the resulting list, as might be expected.
How is it useful?
This is especially useful when looking for companies in emerging sectors where the hot topics are constantly developing, and buzz words come and go. Or even in identifying sectors which use phrases synonymously to convey the same thing (think “recruitment”, “head-hunting”, “candidate resourcing”, “talent acquisition”, etc.).
It also could be used as another resource in the initial phase of ML list building, offering as an alternative to a keyword taxonomy when seeking to identify relevant companies in the development of training sets.
Best use cases
The tool is very new, and we are still exploring best use cases.
From our experience so far, longer search phrases generate “better” results.
The more descriptive text relating to the kind of companies that the user is looking to identify generally serves a better chance of identifying more of the target companies.
It is not, at this stage, recommended to include terms in the search which would be better applied as filters later, such as location identifiers “AI companies in Leeds”. Nor would we recommend including growth indicators in the search “Fastest growing AI companies”, since these terms are unlikely to be found in the company’s web text. These markers are better applied post-search using the filters.
Good examples can be found underneath the search box as a guide.
See this supporting blog post for more suggestions.