Frequently Asked Questions

Are you an ETL tool?

No - in fact, we’re here to solve a problem faced by many companies that use ETL tools (or, for that matter, their own in-house tools and scripts) to migrate and convert data. That’s the challenge of normalizing reference data before it’s converted.

Too often, somebody is stuck manually converting hundreds or thousands of product, company, or location names in an Excel file so that the data is clean and ready for the ETL or migration scripts.

RecordLinker simplifies that step and relieves the burden of data normalization. Our output can be fed into any ETL or in-house solution you use.

Are you on-premise or cloud-based?

We’re cloud-only. But don’t worry - we don’t need access to your systems.

Our process is based on you sending us your reference data. Since you choose what to share, there’s no need for a multitude of internal approvals.

Do you support [insert system name here]?

We can work with any system you have! As long as you can provide us with the input records you need to normalize (either by uploading an Excel or sending the data via API), that’s all we need.

Do I have to manage my own canonical records in RecordLinker?

That’s completely up to you. If you prefer, you can create your canonical record set manually in our system. Alternatively, you can upload the canonical records from your existing system of records, and we’ll use that model and follow its changes.


Sometimes, your data team will prefer a hybrid approach - to import your canonical records into RecordLinker, and later update or even create a new canonical record in RecordLinker - we'll know which ones got changed or manually created, and can provide those back to you via API or Excel download so that you can update your system of records appropriately.

Do you only work with insurance companies?

No! We’ve worked with organizations across a number of different industries, and our portfolio is constantly growing. Reach out to us to learn more about how we can solve your data challenges.

How good is your machine learning? How much training does it need?

Machine learning is an amazing thing, but it’s only ever as effective as the guidance (and the training data that powers it).


We recommend that you provide us with at least 5-10 sample linkages for each of your canonical records. Don’t worry if you don’t have these ready now; you can easily build up your training set via our user interface. It’s very fast and surprisingly fun!

Scroll to Top