How missing data happens
Analytics only knows what a tag manages to send. Anything that stops the tag from loading, from firing or from reaching the platform creates a gap, and the gap looks exactly like a period when nothing happened.
The blunt causes are the common ones: a redesign that drops the container, a plugin update that removes the snippet, a thank-you page renamed so its trigger never matches, a developer deploying without the measurement code. Subtler ones follow. A visitor declines cookies, so no event is stored. An ad blocker removes the request. A script error earlier on the page halts everything after it. A redirect strips the campaign parameters, so the visit arrives with no source. A filter written to exclude staff quietly excludes a whole city. On phones over slow connections — most of the traffic in Nepal — a visitor can arrive, read and leave before the tag has loaded at all. Platforms add their own gaps: old data ages out of retention windows, and rows can be withheld when a group is small enough that individuals could be identified.
Why missing data matters
A gap is not neutral, because it will be explained. Somebody will read the flat line as a collapse in demand, cut a budget, blame a channel or credit a change that had nothing to do with it. The absence of numbers gets treated as a number.
The lasting damage is comparison. Every future report that includes the broken period is wrong: the following month appears to boom, the year-on-year figure is meaningless, and any bidding algorithm that trained through the gap learned from an incomplete picture. And unlike most faults, this one cannot be repaired later — analytics does not record retrospectively, so the period stays empty permanently.
Common mistakes with missing data
The first is not noticing. Nothing alerts you when a tag stops, and a chart with a gap still draws confidently. The second is explaining the dip with a story — an algorithm update, a competitor, a bad season — instead of first checking whether the recording broke.
The third is quietly comparing against the gap anyway, which manufactures growth that never happened. The fourth is confusing missing data with data that exists elsewhere: a conversion absent from analytics may sit safely in the inbox, the phone log or the payment gateway, and those sources can prove what really occurred even though the chart cannot be mended.
How to act on it
Prove it first. Compare the suspect period against a record kept outside analytics — enquiry emails, the call log, orders in the shop admin. If those held steady while the chart fell, the problem is measurement, not demand.
Then contain it. Write an annotation covering the exact dates so nobody misreads the gap next quarter, exclude the period from comparisons or label it wherever it appears, and note in the report that the figures are understated rather than adjusting them upwards with a guess. Afterwards, reduce the chance of a repeat: re-test critical events after every release, watch for a sudden fall in a metric that should be stable, and treat prevention as part of ongoing data integrity rather than as an emergency.