A data quality assessment checks whether an event contact list supports a specific business action. Define that action, inspect the records and their evidence, test a deliberate sample, and assign accept, research, or exclude decisions before releasing contacts for use.
Picture a hypothetical pre-show handoff: the spreadsheet looks complete, company names are familiar, and every row has an email address. Then a rep notices that a supposed attendee works for an exhibitor but has no personal connection to the upcoming event. The file can support account research. It cannot yet support a message claiming that person will attend.
That distinction drives this proposed assessment method. It gives event marketers and sales operations teams a practical way to judge records, document uncertainty, and decide what happens next. It is an original practitioner workflow, not a certified framework or an account of observed customer results.
What is a data quality assessment for event contacts?
An event contact assessment tests whether each record supports its intended use with sufficient evidence.
The assessment unit includes more than a person and an email address. It also includes the company relationship, the particular event edition, the source behind each important claim, and the proposed action. A contact can be correctly identified while their connection to the event remains unknown.
Keep separate decisions for account research, general outreach, and attendance-specific messaging. A company listed in the current exhibitor directory can justify researching that account. Addressing a particular employee as an attendee requires evidence about that person. Neither decision automatically establishes permission to contact them through a particular channel.
Start with a short assessment brief: what the team wants to do, which records are in scope, which facts the action requires, and who can approve release. That brief prevents reviewers from applying different standards to the same list.
Why assess the list before assigning sales work?
Assess the list before committing research time and outreach capacity.
Consider a hypothetical planning example with 600 records. Assume individual preparation takes six minutes per record and staff time is valued internally at $50 per hour. Preparing every record would consume 60 hours, representing $3,000 in staff capacity. These are planning assumptions, not market benchmarks or measured results.
Suppose, within that same hypothetical example, an initial assessment takes eight hours and identifies 150 records that should not enter individual preparation. Avoided preparation would equal 150 × six minutes, or 15 hours, valued at $750. Assessment effort would cost $400, leaving a modeled net capacity value of $350.
The hypothetical capacity return is ($750 − $400) ÷ $400 × 100 = 87.5%. That calculation expresses the value of reusable staff time; it does not establish cash savings, event revenue, or customer acquisition ROI. If the released time is not used productively, the modeled benefit will not materialize.
Use your own effort estimates to decide how much review is justified. Expensive account preparation deserves deeper identity checks than broad market mapping. Also consider the consequence of the error: a wrong phone number wastes a call, while a mistaken attendance claim can undermine the opening conversation.
Which data quality dimensions should you check?
Use separate dimensions to make different defects visible.
The UK Government Data Quality Framework identifies six dimensions. It distinguishes completeness from accuracy and explains that suitability depends on intended use, with potential trade-offs between dimensions. The concise definitions below paraphrase that framework.
For this proposed event workflow, record results by field and action. An email can pass a syntax check while its ownership remains unresolved. A current company name can coexist with an unsupported event association. A complete row therefore needs further inspection before release.
Avoid averaging everything into a single quality score. A critical identity conflict should remain visible even if every optional field is populated. Use dimension results to explain the decision, then apply the release rules established in the assessment brief.
How to assess an event contact list
Move from the intended action to the evidence required to support it.
The following sequence keeps review focused and makes the final decision traceable:
- Define the intended action and required evidence.
- Preserve the original file and profile its fields.
- Inspect person, company, and event relationships.
- Resolve identity conflicts and distinguish missing from wrong values.
- Sample records across meaningful groups.
- Log defects and assign a release decision.
- Recheck repairs and approve a documented version.
1. Define acceptance rules around the intended action
Write acceptance rules before examining attractive accounts or recognizable names.
For account research, require a resolved company identity and a relevant event association. For general outreach, add a usable contact identity, a supported company relationship, and channel eligibility under your organization's rules. For messaging that refers to attendance, require evidence supporting the particular attendance statement.
Treat these as separate release destinations. Someone can be accepted for research while remaining excluded from an attendance-specific campaign. Store the destination alongside the decision so that a later export cannot turn a limited approval into unrestricted use.
Specify blocking conditions in plain language. In this proposed workflow, an unresolved person match blocks personalized outreach. An unsupported attendance assertion blocks attendance-specific wording. A missing optional phone number does not block an otherwise approved email workflow.
Agree who resolves exceptions. Sales can explain why an account matters; the reviewer should still document whether the required evidence exists. Commercial priority can justify additional research without changing what counts as a supported claim.
2. Preserve the source and profile the whole file
Keep an unchanged source copy before normalizing or enriching records.
Assign a stable record identifier and capture the import source, batch, and receipt date. Preserve original values beside edited values, or maintain a change log that links them. Reviewers need to distinguish a supplier's statement from a later correction made by your team.
Run structural checks across the full file where practical. Look for missing required fields, malformed addresses, unexpected categories, repeated identifiers, conflicting event editions, and dates that cannot support the intended sequence. These checks identify candidates for review; they do not prove that the underlying facts are correct.
Report denominators carefully. Required-field completeness should use the records for which that field is required. Do not penalize account-only records for lacking a personal phone number when the approved use does not need one.
Keep structural inspection separate from factual verification in the report. State whether a result came from inspecting all rows or manually researching a subset. A complete formatting check must never be presented as a complete accuracy check.
3. Trace the person-to-event evidence
Every attendance-related statement needs evidence connecting the person to the relevant event edition.
Build that connection explicitly. Record the person, employer, event, edition, relationship type, source location, and observation date. Also capture when the underlying source was published or updated, where available. A page retrieved today can still describe an earlier event.
Use relationship categories that preserve what the evidence actually says. An exhibitor directory supports a company-level relationship. A speaker page supports a named speaking association. Registration supports registration status, while direct confirmation supports the person's stated plans at the time of confirmation. Predictions belong in their own category.
None of those categories should silently become proof of physical attendance. If the planned message says that someone visited your booth, the record needs evidence of that interaction. If it says their company is exhibiting, company-level evidence may be sufficient.
For the underlying source distinction, see exhibitor lists versus attendee lists. During assessment, retain the narrower supported claim even when broader wording would make outreach easier.
Pro Tip: Add a field called “supported statement.” Write the strongest sentence the evidence permits, such as “Employer appears in the current exhibitor directory.” Review that sentence alongside the proposed message.
4. Separate missing, wrong, and unresolved values
Missing information and contradictory information require different repairs.
Use distinct statuses for absent, malformed, contradicted, unverified, and supported values. An empty employer field needs research. An employer field contradicted by a current authoritative source needs correction and a check of related fields. A plausible employer with no supporting evidence should remain unverified.
In a hypothetical example, a contact has a correctly formatted email at their former employer. The address can pass a formatting rule while the employer relationship fails factual review. Adding a phone number would make the row more complete without resolving the defect that blocks the intended action.
Record “unknown” deliberately when research cannot establish a fact. Do not replace uncertainty with a guess solely to satisfy a required field. If the receiving system requires a value, change the routing or use an explicit unresolved category instead of inventing information.
Also distinguish review status from truth status. “Not checked” describes work still to do. “Contradicted” describes evidence already found. Mixing them makes the research queue harder to prioritize and can hide known errors among ordinary gaps.
5. Resolve identity before merging records
A merge should preserve a supported identity and its separate relationships.
Use combinations of evidence when matching people: name, employer, role context, source references, and contact details. Treat matching email addresses as useful evidence that still needs interpretation, particularly for shared inboxes. Matching names alone should create review candidates rather than automatic merges.
Separate people from their roles and event associations. A person can change employers, hold multiple responsibilities, or appear in several event editions. Repeated person details may represent legitimate relationships rather than duplicate records that should be deleted.
In a hypothetical conflict, two rows share a name but show different employers and unrelated roles. Keep them separate until stronger evidence resolves the match. Conversely, spelling variations can belong to the same person when independent details align.
Preserve the original identifiers when merging, and record which values survived and why. If the task also requires company normalization, the trade show exhibitor list cleanup workflow provides a related account-level process. The assessment here determines whether the resulting identities are adequately supported for release.
Pro Tip: Treat merge uncertainty as its own issue category. A confident-looking combined record can conceal more uncertainty than the separate source rows did.
How do you distinguish freshness from accuracy?
Track when evidence was checked separately from what it establishes.
For this proposed workflow, retain a checked date for each critical claim rather than relying only on a row-level update timestamp. Enrichment of a phone number should not make an old event association appear newly verified. Record the source date, check date, and any unresolved qualification together.
Set review triggers around the action. Recheck a person's stated attendance before sending a message that depends on it. Review an employer conflict before assigning the contact to an account owner. A freshly downloaded file should still undergo these checks if its underlying evidence lacks dates.
Avoid imposing one universal expiration period on every field. A historical record of a past conversation serves a different purpose from a current claim about someone's role. Define review rules by claim type and business consequence, then document the rationale in the assessment brief.
Where a source provides no publication date, say so. A reviewer can record when the page was observed without implying that its contents were updated then. That distinction gives the release owner a clearer basis for accepting uncertainty or requesting more research.
How should you sample an event contact list?
Sample across meaningful groups and state exactly what the review can support.
Divide the list into strata before choosing records. Useful groups include source, import batch, evidence category, region, and identity-conflict status. Choose divisions connected to plausible failure modes. A file assembled from different sources should not be judged only by its easiest records.
In a hypothetical sampling plan, a 600-record list contains 300 organizer-sourced records, 200 enriched records, and 100 predicted event associations. Reviewers randomly select 30 records from each group. Those sample sizes are illustrative choices, not a statistically justified universal recommendation.
That hypothetical design gives the smaller group more scrutiny relative to its size. Report findings separately for each group. Pooling all reviewed rows into one percentage would overrepresent the smaller group because the sampling fractions differ.
If the team needs a list-wide estimate, it must account for the group sizes, selection method, and sampling uncertainty. A weighted estimate can address unequal sampling fractions; it cannot repair biased selection or unreliable verification. Obtain statistical support when an estimate will determine a material purchase or program commitment.
Add targeted reviews for high-consequence records, such as executive invitations or ambiguous identity merges. Label these separately from random selections. A deliberately difficult review set can reveal failure modes, but its defect rate does not estimate the prevalence of those defects across the file.
Record unresolved checks alongside confirmed defects. The absence of accessible evidence does not automatically prove that a value is wrong. Equally, a sample with no observed errors does not certify every unreviewed record. Document which fields were tested, which sources were consulted, and what remained unknown.
Pro Tip: Freeze the sample before research starts. If a record is difficult to verify, retain it and mark the outcome unresolved rather than replacing it with an easier record.
How do you turn findings into accept, research, or exclude decisions?
Release decisions should identify the permitted action and the remaining uncertainty.
Use the following proposed decision table to keep review outcomes consistent. Apply it at record level, and at batch level when a shared defect affects an entire import or source group.
Acceptance is scoped. A record accepted for account mapping can remain excluded from attendance-specific outreach. Exclusion also need not mean deleting the underlying history. Preserve enough information to explain the decision and prevent the same rejected row from re-entering through another import.
Maintain an issue log containing the record or batch identifier, affected field, defect category, evidence, business consequence, owner, status, and next action. Include the release destination so another reviewer can understand why the issue matters.
Prioritize by consequence and reach. A mislabeled event edition across a batch deserves attention before optional missing job titles. If a sample reveals a possible shared defect, pause the affected group and investigate the pattern before releasing it on the strength of unrelated passing checks.
A hypothetical issue note might read: “Employer participation is supported; individual attendance remains unverified. Accept for account research, hold attendance-specific messaging, and request person-level evidence.” That note tells the next owner what is known and what work remains.
What should the assessment handoff include?
The handoff should make the release decision reproducible.
Name the assessed file version, intended use, required fields, blocking rules, and reviewer. Summarize full-file checks separately from sampled factual checks. Include each sampling group's size, selection method, findings, unresolved cases, and limitations. Attach the issue log and identify who accepted any remaining uncertainty.
After repairs, retest the affected checks. If a transformation caused the defect, inspect other records touched by that transformation. Confirm that merges preserved source references and that repaired values reached the actual release file. An updated research sheet does not establish that the sales export was corrected.
Keep accepted, research, and excluded records in clearly controlled destinations. The event data management guide addresses the broader handoff into operational systems. Carry assessment decisions and supported statements with the records so the receiving team can respect their limits.
Choose an upcoming list and write its acceptance brief before adding more contacts. The useful outcome is a release you can explain: which records support which actions, where evidence is missing, and who owns the next decision. Use that standard when evaluating event intelligence from Lensmor or any other source.
