Data hygiene for marketing: rules before dashboards

Aug 22
3 min read

Updated: 4 days ago

Introduction

Reporting fails far more often because of the data going in than the tool showing it. A dashboard built on inconsistent records produces confident, wrong numbers.

The unglamorous work of keeping records consistent is what makes everything downstream possible, and it is almost entirely a matter of agreeing rules before anyone starts typing.

1. Data hygiene for marketing starts with one agreed name per thing

The most common defect in small business records: the same channel recorded four ways.

Facebook, FB, facebook ads, Meta. Google, google ads, adwords, PPC. Referral, word of mouth, friend, recommendation.

Each variant becomes a separate row in every report, splitting one channel's results into fragments that individually look insignificant. Write a fixed list of channel names, keep it short, and use only those.

2. Use dropdowns instead of free text wherever you can

Rules that rely on people typing carefully will be broken, not through carelessness but because typing is fast and checking is slow.

Any field you intend to count should be a dropdown, a set of buttons, or a validated list. Free text belongs only in fields you plan to read rather than count.

This single change removes most inconsistency permanently, and it makes the rule self-enforcing instead of dependent on training.

3. Record the date something happened, not when it was entered

Records entered in batches lose their timing, which quietly ruins every trend.

Each entry needs the date of the actual event — when the enquiry arrived, when the sale closed, when the customer first contacted you. If those get stamped with the day someone updated the sheet, a month's activity appears on one Tuesday.

Keep both dates if it helps. Never keep only the entry date.

4. Decide how to handle duplicates before you have thousands

The same customer will appear more than once: a different email, a nickname, a company name in one record and a personal name in another.

Choose a single identifier now — usually a phone number or an email address — and treat it as the key. Decide what happens when a duplicate is found: merge and keep the earliest first-contact date, or keep the most complete record.

Undetected duplicates inflate customer counts, understate repeat rates, and make lifetime value look lower than it is.

5. Capture source at the moment of contact

Attribution data collected later is invented. Ask when the enquiry arrives, while the answer is known.

That means the question belongs in the form, in the booking flow, or in the first sentence someone says on the phone — with the answer recorded immediately in the agreed vocabulary from the first section.

Reconstructing sources at the end of the month produces a tidy report describing nobody's actual behaviour.

6. Keep required fields to the minimum that matters

Every mandatory field lowers completion, and low completion is its own data problem.

For most small businesses the essential set is small: name, one contact method, source, date, and what they wanted. Everything else can be optional or gathered later.

Long forms produce fewer records, more abandonment, and a habit of entering placeholder values to get past a field — which is worse than not collecting it.

7. Fix inputs, not exports

When a report looks wrong, the temptation is to correct the spreadsheet.

That fixes one month and guarantees the same work next month. Trace the error to where the data was entered and fix the process — the form, the dropdown, the missing rule, the person who was never told the convention.

A useful habit: every time you clean something manually, write down what caused it. After a quarter that list is your process improvement plan, and it is usually four or five items long.

8. Audit quarterly with three checks

A short recurring review keeps records usable.

Count records missing a source. Count entries whose channel name is not on the agreed list. Count suspected duplicates. Three numbers, tracked quarterly.

Rising figures mean the rules have stopped being followed, usually after new staff joined or a busy period. Catching that at 5% is a conversation; catching it at 40% is a rebuild.

Conclusion

Agree one name per channel and enforce it with dropdowns rather than instructions, record event dates rather than entry dates, and settle your duplicate rule and identifier early.

Capture the source at the moment of contact, keep required fields minimal, always fix the input rather than the export, and audit missing sources, off-list values and duplicates every quarter.