We recently shared our work improving the coverage and accuracy of our French company to domain matches. The results were impressive. We more than doubled our coverage and cut our error rate by 40%. And then we asked you where we should focus on next.
The biggest demand was for us to the look at the USA, and on a trip to Sunderland, ancestral home of George Washington, thinking about the demise of the Windows Phone, a product of the fine Seattle company Microsoft, I realised that we should look at the state named Washington.
Company Domains
We can’t classify what a company does without the text from its website. And to get that web text we need to know a company’s domain. This is one of the hardest things we do at The Data City.
We’ve worked for years to make this process better in the UK. And we’ve worked for a month now to make this process better for France.
After ten days of focused effort on Washington, we’ve made huge progress in the USA.
For companies with at least 100 employees we’ve increased the number of companies we match by a factor of six while decreasing our error rate by 30%.
We now estimate that we have a correct domain name for 86% of Washington State companies with over 100 employees.
Our improvements are not as big, but still substantial for smaller companies with our coverage more than doubling and the number of correct domain names for smaller companies increasing by over 80%.
What next?
The best news about our improvements for Washington State is that we did nothing to find company domains that shouldn’t work for the whole of the USA. Modelling these improvements suggests that the number of companies with a matched domain name in our global product will increase from 11 million to 23 million.
We still have one hard challenge for the USA that we’re working on. Interstate commerce and company registration are not easy there. Companies often register in every state that they operate in. Companies like Starbucks often register each café too. But our users want to know that Starbucks has 16,000 cafes, with a corporate headquarters in Seattle, and regional headquarters in other places.
We’re learning as fast as we can and we’re trying to figure out how to display multiple locations and registered entity of a single company in a single grouped view.
While our team work on getting our new company domains for the USA we’re looking for the next territory for us to look at. Let us know if you’ve got one you’d really like us to work on.
