Score CRM Source Attribution Completeness for Local Leads

A dashboard can display a neat channel breakdown while many CRM records have blank, conflicting, or manually guessed sources. That creates false confidence: the chart looks complete even when the underlying lead history is not. A source-completeness audit measures how much of the lead population can be classified under a documented rule, identifies where information is lost, and avoids pretending that an assigned label always proves causation.

Define a lead record and source vocabulary

Before calculating completeness, define what one lead means in the CRM: a person, an enquiry, a company, or an opportunity. Create a controlled list of source categories, such as organic search, paid search, partner referral, repeat customer, offline campaign, and unknown. Keep first-known source separate from latest touch if both matter. Do not let every staff member invent a new label for the same channel.

Separate missing from unknown

A blank field means the system has no stored value; 'unknown' means the record was reviewed but the origin could not be established. 'Other' should be reserved for a defined category that is not covered by the standard list, not used as a hiding place for missing data. This distinction makes remediation measurable. Record the source of the value—automatically captured, imported, customer-reported, or manually entered—when that information is available.

Calculate coverage transparently

A basic completeness rate is the number of eligible records with a valid, documented source divided by all eligible records in the same population, multiplied by 100. State exclusions, such as test records or spam, and keep the denominator fixed when comparing periods. A higher completeness rate means the dataset is more classifiable under the chosen rule; it does not necessarily mean attribution is more accurate or that one channel is more successful.

Trace gaps to their origin

Sample records with missing sources and follow the workflow backward: landing page, form hidden fields, call intake, staff handoff, CRM integration, or import job. Identify the point where information disappeared. Do not simply make the sales team fill every blank from memory, because a guess can be worse than an explicit unknown. Assign a technical or process owner and test the correction using a small set of new records.

Report coverage and reliability separately

Show source completeness beside a data-quality note that explains how sources are captured and verified. A field populated from a default value may count as nonblank while still being unreliable. Periodically sample records and compare the label with available evidence. If a new integration changes the capture process, annotate the date and consider treating the later series as a different measurement phase.

Worked example

Hypothetical example: a local training institute has 200 eligible enquiries, 150 with valid documented source labels, 25 marked unknown after review, and 25 left blank. Report 75% labelled coverage if the documented-valid category is 150/200, while separately showing the blank and reviewed-unknown counts.

Sample for correctness, not just presence

Select a set of records with populated labels and compare each label with available evidence such as campaign tags, referral details, or the original intake note. Report the share that appears supported under the review rule. This check complements completeness and can reveal a system that fills every blank with a default channel.

Make staff workflows easy to follow

A required source field can improve coverage only if staff know what the options mean and can complete it without guessing. Use concise definitions, examples, and an unknown option. Review recurring errors with the process owner instead of blaming individual staff for a confusing form.

Frequently asked questions

Is a filled source field always accurate?

No. Completeness measures coverage; a separate validation sample is needed to assess reliability.

Should unknown records be removed from the denominator?

Usually not when measuring population coverage. State any exclusion rule and apply it consistently.

Can a CRM source prove which channel caused a sale?

Not automatically. It records attribution under the organisation's rules and available evidence.

Decision rule

Decision rule: publish a channel-performance comparison only with the source-coverage rate and a plain explanation of how the labels are obtained.

## Review the source field after process changes

When a form, call system, or CRM workflow changes, sample fresh records before assuming the completeness rate improved. A new default can fill blanks while quietly assigning the wrong channel to every enquiry. Check whether the recorded source is supported by a tag, referrer, intake answer, or other defined evidence. If a label was manually chosen, preserve that distinction. The goal is not a full column; it is a source classification that can withstand a reasonable review.

Evidence to retain

For source-quality work, retain the denominator, blank-field count, reviewed-unknown count, and validation sample so completeness is not mistaken for accuracy.