Return to overview
CV & Data Automation
Remo Vloet6 min.

Measuring recruitment data quality

Measure whether your team can use the information, without treating a populated field as automatically correct or current.

On this page

A candidate record has every field filled in. The phone number is old, the owner has left and the latest note belongs to another person. Calling the record ‘complete’ says little about whether a recruiter can use it.

Recruitment data quality metrics should serve a specific task. A callback list needs different information from a report on application stages. Start with the decision or action, define which records count and assess completeness, freshness and accuracy separately. Otherwise, filling empty fields can improve a score without helping anyone do the work.

Key takeaways

  • Link every metric to a work question and a defined set of records.
  • A populated field can hold an incorrect or outdated value.
  • Keep the counts behind percentages and make unknown results visible.
  • Use findings to fix causes, rather than merely filling more fields.

Which work question should the data answer?

Choose what you want to do with the information first. ‘Which candidates are waiting for my response?’ requires an outstanding commitment, an owner and a clear next action. A complete education history doesn’t make that list more dependable.

Define the records included in the measurement. Are they active records belonging to your team, everyone contacted this month or a particular import batch? Record the selection rules and measurement date. Without that boundary, two colleagues can calculate different results while each follows their own definition correctly.

The Government Data Quality Framework considers fitness for purpose and distinguishes several quality dimensions. It’s a framework for government data, not a mandatory recruitment standard or evidence for a particular ATS score. Its concepts help keep different questions separate.

How do populated, valid and accurate values differ?

A populated phone field establishes that a value exists. A format check can establish whether it follows the agreed pattern. Neither proves the number belongs to the candidate or remains in use.

Name the check you’re actually carrying out. Call a missing-value check ‘completeness’ and a pattern check ‘format validation’. Reserve an accuracy judgement for cases where you’ve checked the value against an appropriate source.

Unknown is a useful outcome. If nobody has checked whether information remains correct, don’t silently count it as accurate. Equally, ‘not checked’ doesn’t mean ‘wrong’. That distinction prevents a falsely precise quality score.

How do you choose a useful set of checks?

Limit the first measurement to a process where someone will act on the findings. Record the definition alongside the numerator and denominator. These are possible checks, not a standard dashboard every ATS already supplies.

CheckPossible measureWhat it doesn’t prove
CompletenessRecords with the required value divided by records needing itThe value is correct
FreshnessRecords with confirmation within your chosen periodNothing has changed since
FormatValues that follow the agreed formatThey belong to the right person
RelationshipChecked notes attached to the intended recordEvery other relationship is correct
Possible duplicatesFlags requiring investigationRecords can safely be merged
UsabilityChecked records that support the agreed taskThe whole record is error-free

Choose a freshness period suited to the information. An education detail and an availability agreement don’t change in the same way. Explain your choice rather than presenting it as a legal deadline or industry benchmark.

What does a fictional calculation show?

A fictional recruitment team reviews twenty active records for an agreed contact task. Eighteen contain a contact route. Twelve contain evidence that the route was confirmed within the team’s chosen review period.

Presence is eighteen out of twenty, or 90%. Recent-confirmation coverage is twelve out of twenty, or 60%. Those are different findings. The second doesn’t mean the other eight contact routes are wrong. Two are missing; the remaining six lack recent confirmation under this definition.

These numbers are only a fictional calculation. They describe neither an average agency nor Simply, and they aren’t a target for your team. The useful output is the action list: which records lack information, which need confirmation and who will follow up?

Saving a new note may change the record’s modification date. It doesn’t suddenly reconfirm the contact route. For this measure, use the relevant confirmation date rather than automatically relying on the latest technical update.

How should you handle blanks and exceptions?

Agree when a field is actually needed. Requiring a home address for every initial contact could change a completeness score without helping the contact task. Don’t collect extra personal data simply to raise a number.

Distinguish ‘not known’, ‘not applicable’ and a confirmed value where the process needs that distinction. Make those categories checkable. A casually entered dash or zero shouldn’t quietly count as a meaningful answer.

Record exceptions alongside the measure. If some records are excluded, explain why. Silently removing the most difficult records from the denominator doesn’t resolve the underlying problem.

How do you check relationships and duplicates?

Quality also concerns where information belongs. An accurate callback agreement attached to the wrong person’s record is still unusable for the recruiter. Include a content sample alongside technical counts.

Explain how you select that sample. Looking only at the tidiest records gives a different view from examining different sources, record ages and entry routes. Deliberately checking likely problem cases helps uncover errors, but their error rate can’t automatically be applied to the whole database.

Treat duplicate signals as questions to investigate. Matching names may belong to different people, while two records can contain different context. Establish whether they concern the same person and which relationships and history must survive before considering a merge.

What should you do with the findings?

Choose a problem that obstructs the daily task and find where it starts. Is an owner missing because nobody enters it, because the import skips the field or because responsibilities are unclear? The same blank cell can have different causes.

Assign someone to the correction and measure again using the same definition. If the selection or rule changes, state that alongside the comparison. Otherwise, a better score may mainly reflect a different calculation.

Simply’s ATS and CRM data model lets your administrator configure objects, fields and relationships. Use that flexibility to make the meaning of your information clear, for example with separate fields for an availability date and when it was confirmed. Then agree who checks and updates an outdated value. The configuration can follow the quality question you’re actually measuring.

For a demo, bring one measurement definition and a fictional error example. Ask how you would find and inspect the affected records. Then decide which checks fit the system and where human assessment remains necessary.

Share this post

About the author

Remo Vloet

Remo Vloet is a co-founder of Simply, the AI Operating System for recruitment agencies: inbox, meetings, sourcing, CV parsing, search and matching, documents and automation in one system. With a background in building complex software, he contributes to the technical vision behind Simply.

LinkedInArticles by Remo Vloet

Frequently asked questions

What data quality percentage should a recruitment team achieve?

There isn't one universal answer. Choose a target suited to the task and the consequences of errors. Record what the metric measures and what remains outside its scope.

Is a complete record a good record?

Not necessarily. Populated values may be outdated, incorrect or unnecessary. Assess the information needed for the intended work.

Is a recently modified record current?

Only if the relevant information has actually been confirmed. A new attachment or administrative update doesn't automatically reconfirm earlier agreements.

Can accuracy be measured entirely automatically?

Presence and format suit technical checks. Meaningful accuracy often needs an appropriate source or human assessment. Keep the scope of each check visible.

Should we clean all historical data first?

Start with a defined task and the data it needs. Fix causes at the point of entry while addressing relevant existing errors. A large one-off clean-up doesn't prevent new problems.

More in CV & Data Automation

Chase less,place more.

Find out how Simply can completely evolve your workflow. No slides, just product.

Cookie settings

Choose which optional cookies we may use; our cookie policy explains each one. Optional categories are off until you turn them on, and you can change this at any time from the footer.

Necessary

Keeps the site secure and remembers your language and these choices. These cannot be switched off.

Always on
What this includes
  • simply_lang · Simply

    Remembers the language you picked in the footer.

    Stores: Cookie, 1 year

  • simply_consent · Simply

    Remembers the choices you make here.

    Stores: Local storage, 12 months

  • Cloudflare Web Analytics · Cloudflare

    Counts page views and measures loading speed, without identifying you.

    Stores: No cookies

  • Cloudflare Turnstile · Cloudflare

    Checks the demo request form is sent by a person, not a bot.

    Stores: No cookies

Analytics

Shows us how visitors use the site, so we can improve it.

What this includes
  • Google Analytics · Google

    Counts visits and shows which pages are used.

    Stores: _ga, _ga_*: 2 years

  • Contentsquare · Contentsquare

    Analyses clicks and scrolling, with form input masked.

    Stores: _cs_id, _cs_c: 13 months; _cs_s: 30 minutes

Marketing

Measures our ads and campaigns, and lets us show relevant ads on other sites.

What this includes
  • Google Ads · Google

    Measures which ads lead to a demo request.

    Stores: _gcl_au: 90 days

  • Meta Pixel · Meta

    Measures conversions and supports retargeting on Facebook and Instagram.

    Stores: _fbp: 90 days

  • LinkedIn Insight Tag · LinkedIn

    Measures conversions and supports retargeting on LinkedIn.

    Stores: bcookie, li_sugr, lidc and others on linkedin.com: 1 day to 1 year

  • Factors.ai · Factors.ai

    Identifies visiting companies and attributes visits to campaigns.

    Stores: _fuid: 400 days

Google Tag Manager loads the Google, analytics and marketing tags above only after you allow them.