CRM Marketing: Five Key Metrics to Measure Data Quality | Celsius GKK International
February 15, 2017 Antoni Chumillas

CRM Marketing: Five Key Metrics to Measure Data Quality

Database Quality

B2B marketing databases are, by their very nature, complex. You can try to measure everything, but you’re probably doomed to failure if you do. At Celsius, we suggest that five key metrics will give you a pretty good view on how ready your database is for effective marketing. And if you run these processes on top of a real time dashboard you can track whether your database is improving or deteriorating over time. Here they are:

1.What to measure: Address postal promotability.

Proposed standard: At least 95% of your addresses should be postal deliverable, using local post office standards

How to measure it: Run all your addresses against a postal standards database once per quarter. Online services like Address Doctor (, German-based so they understand European addresses well, can provide the kind of service you need.

2. What to measure: e-Mail address deliverability.

Proposed standard: no more than 10% of your e-mail addresses should be ‘hard-bounces’ in any given e-mailshot

How to measure it: Test all your e-mail addresses once per quarter using your own EHLO script or via a cloud provider like StrikeIron (

3. What to measure: Firmographics.

Proposed standard: 95% of your records should have a valid business activity code, such as NACE or SIC, and a valid number of employees. For business activity, make sure you exclude codes like 9999, holding companies (6420 in NACE, 6719 in SIC) and households (97XX in NACE, 8811 in SIC), which are completely useless when using this measurement. And similarly, for employee numbers, a zero value should count as an invalid value.

How to measure it: implement a report which lists (a) all records with a missing activity code or one not in official list of codes (b) all with a blank, zero or ‘unknown’ number of employees. Run the report either every quarter or as an online dashboard.

4. What to measure: Validity of contact names

Proposed standard: Where named contacts exist in your database, no first name or last name should contain invalid strings, spurious names (for example, “Donald Duck”) or just descriptions.

How to measure it: implement a system which checks for ‘junk’ of this kind. IBM’s NameParser (part of their Global Name Analytics suite) will do the job, and much more besides. We at Celsius also have available such a process.

5. What to measure: Segmentation of contacts by function

Proposed standard: 70% of named contacts should be selectable by function (and ideally by level, as well).

How to measure it: implement codes to classify ‘business card’ job titles by function (e.g. general management, sales, production, purchasing) and level (‘C’ level, director, manager, executive, secretarial/clerical, operative, etc.).  If you have a large number of uncoded job titles in your database, use parsing software from companies specialising in CV processing – or we at Celsius can help.

More information on data quality and CRM data integration in general at the Celsius Blog or contact us

Share this:

Tagged: ,