Missing marketing data: find the gap and fix your reporting

Zero and unknown answer different questions

A verified zero means the relevant source was available, checked over the requested period, and returned no qualifying records under a stated definition. Unknown means the measurement process cannot establish whether qualifying records existed. That can happen because data is missing, delayed, inaccessible, or not linkable to an outcome.

The distinction is practical. If a paid channel appears to have no spend, a marketer may pause it. If its account has not synced in ten days, the right next step is to restore coverage and rerun the report—not to make a budget decision from an invented zero.

Find the kind of gap

Not every gap has the same cause or remedy. Naming it makes the next action clearer and prevents teams from treating all incomplete data as a generic warning.

• Connection gap: an account was never connected, authorization expired, or access changed.
• Freshness gap: a source is connected but the latest import is delayed or has failed.
• Coverage gap: the requested date range predates retained data or excludes some campaigns, regions, or channels.
• Identity gap: a touchpoint and conversion exist but cannot be confidently linked to the same journey.
• Definition gap: systems count different events, currencies, refunds, or conversion windows.
• Consent or privacy gap: identifiers are unavailable by design or regulation, limiting person-level matching.

Make the limitation visible near the result

A warning buried in a settings page rarely reaches the person reading a chart. Put source status, period covered, last update, and material exclusions alongside the affected figure. If some channels have full coverage and others do not, say which is which.

Avoid showing an exact blended return when its numerator or denominator is incomplete without an explicit qualification. If the system cannot calculate a trustworthy value, show why it is unavailable and what needs to happen before it can be calculated. A blank with an explanation is more useful than false precision.

Separate recovery from interpretation

First repair the data pipeline: restore access, confirm the account and date range, wait for a successful sync, and check that the expected records arrived. Then reconcile totals and definitions against the source system. Only after that should the team interpret the newly complete period.

Do not quietly backfill a result with a different attribution window or merge sources whose definitions disagree. Record the change and rerun the same comparison with consistent rules. That keeps an operational fix from looking like a marketing performance shift.

Build a useful missing-data routine

A good routine names an owner for each source, sets an expected sync cadence, and defines which reports become partial when coverage is lost. It also spells out how unmatched records are handled and when a period should be excluded from a comparison.

The goal is not to eliminate every uncertainty. Some activity will remain unobservable, particularly across privacy boundaries or offline interactions. The goal is to distinguish what was measured from what was not and make that difference actionable.

A missing account can make return look better

Imagine a fictional retailer with £40,000 of revenue and two ad accounts. The first account supplies £8,000 of spend. The second actually spent £2,000 but failed to sync. If the report silently counts the second account as zero, it divides £40,000 by £8,000 and shows a ratio of 5.0. With both spend sources present, the ratio is 4.0. Missing spend made the result look stronger.

The opposite happens if conversion data is missing while spend is complete: the report can make marketing look weaker. This is why a single warning symbol without an explanation is insufficient. The reader needs to know which side of the calculation is affected, which period is incomplete, and whether the displayed ratio should be used at all.

Triage a gap in a useful order

Start with account identity and access. Is the connected account the one that owns the campaigns? Has the user lost permission, or has authorization expired? Then inspect freshness: a healthy connection can still have a failed import. Finally check historical coverage: a successful sync for yesterday does not prove that the previous ninety days have been backfilled.

If records arrived but reporting still differs, inspect filters and definitions. A campaign excluded by a region filter is not the same as a campaign missing from ingestion. A CRM opportunity counted by created date will differ from one counted by closed date. Work through these questions before treating the issue as a tracking failure.

Assign an owner and a next check. “Google Ads missing for Monday to Thursday; access restored; backfill under review” is actionable. “Some data may be incomplete” does not tell a marketer whether to wait, reconnect, or investigate a definition.

• Confirm the source account and permission scope.
• Check the last successful import, not just connection status.
• Compare the requested dates with imported coverage.
• Review filters, currencies, and conversion definitions.
• Verify a sample of records against the original source.

Measure coverage with a named denominator

A statement such as “80% coverage” needs a definition. It could mean eighty percent of expected accounts are connected, eighty percent of revenue has a matched journey, or eighty percent of days have successful imports. These measures answer different questions and should not be combined into one unexplained score.

For journey matching, report both matched revenue and total eligible revenue under the same outcome definition. If £36,000 of £45,000 is matched, revenue match coverage is 80%. That does not mean the system observed eighty percent of every customer's interactions. Some matched journeys can still be incomplete. Explain what the denominator includes and which gaps the percentage cannot detect.

Handle late conversions and refunds

Recent periods often remain unsettled. A shopper may convert several days after a click, a lead may qualify weeks after a form submission, and a refund may arrive after the weekly report is exported. Set an operational reporting convention: for example, review completed weeks while showing when each source was last updated. The appropriate delay depends on your conversion cycle and source behavior.

When revenue is revised, decide whether a refund adjusts the original order period or appears in the refund period. Keep that choice consistent with the question and the finance definition. A team reviewing acquisition performance may need the original cohort revised; a team reviewing cash movements may need a different period. The important point is to state which view the report uses.

Agree when a report can be used

Define which missing sources make a review unusable and which permit a scoped interpretation. A small unavailable organic source may not block a paid-spend reconciliation, but a missing primary ad account likely does. Base that decision on the question and materiality, not on a universal percentage threshold.

For any partial review, state what remains valid. You may be able to inspect one fully synced account's spend trend without calculating a blended return. Once the gap is repaired, rerun the same dates and settings, compare the corrected totals, and notify people who used the earlier version.