Data enrichment adds attributes from an external source to records you already hold. You match each record to the source by a shared key, such as a domain or an identifier, then append the fields that matched, such as company size, technologies in use or recent announcements.
How an enrichment run works
- Choose the match key and normalise it on both sides, as in domain matching.
- Match, and keep how each record matched.
- Append the fields you need with the date each was observed. Do not overwrite values you already trust.
- Keep records that found no match, flagged, instead of dropping them.
- Refresh at the pace the source changes, and expire a field you cannot refresh.
Choosing fields and a refresh rate
The fields worth adding are the ones your users act on, and each changes at its own pace. Illustrative choices from Fokals datasets:
| Field to add | Dataset | Refresh |
|---|---|---|
| Technologies in use | Technology Stack | Daily to weekly |
| Open postings | Hiring Activity | Daily |
| Intent score for a topic | Intent Scores | Weekly |
| Latest announcement | Company News | Daily to every three days |
Refreshing faster than the source changes adds cost and nothing else, and refreshing slower leaves the value stale. Store the observation time with each value so that a user can see how old it is.
Common mistakes
- Reading a high match rate as a sign of accuracy: check a sample by hand.
- Overwriting a trusted first-party value with a vendor value.
- Showing a stale value as if it were current.
- Forgetting the licence: showing enriched fields to your own users is usually embedding, and letting customers export them is redistribution.
In Fokals data
Fokals enriches company records with firmographic, technographic, hiring and intent fields, so a record in your CRM or data product gains the fields of its company. You can match on the company ID, on a domain, or on listing identifiers such as ISIN. Every record carries the time it was observed, and the data dictionary lists the columns. The use case on enriching company records in a data product works through an example.
Related terms
- Entity resolution: deciding which records are the same company before fields are added.
- Company identifier: the keys an enrichment can match on.
- Firmographic data: attributes that describe a company, often the first fields added.
- Technographic data: the technologies a company runs, another common addition.
Frequently asked questions
What is the difference between data enrichment and data cleansing?
Enrichment adds fields that a record lacks. Cleansing corrects or standardises fields it already has, such as fixing a country name or removing duplicates. They often run together, because a clean, normalised key makes the match behind an enrichment more reliable.
How often should enriched data be refreshed?
At the pace the source changes, and no faster. Hiring and technology data change daily or weekly, so a monthly refresh leaves them stale. Stated headcount changes slowly, so a daily refresh adds nothing. Store the date each value was observed so that users can see how old it is.
What is a good match rate for data enrichment?
There is no universal figure, because it depends on what your records hold and what the source covers. Measure it on your own list, and check the quality of the matches as well as their number by reviewing a sample by hand. A high match rate with wrong matches is worse than a lower rate with right ones.