≡Keeping your data in order

Because almost everything in these programs depends on it. Three examples with the same cause:

• A campaign reaches the wrong people — because the categories are not right.
• The location map stays empty — because the postcode was not assigned during the import.
• The AI match finds nothing — because roles and skills are thinly maintained.

None of these is a software error. Ten minutes of order save manual work every week.

Five, which together are enough and fit on half a page:

1. Spelling of company names. A variant creates a separate file.
2. Which categories exist and what they mean.
3. Who creates contacts — ideally everyone, otherwise shadow lists appear.
4. What goes in the note and what in a field: anything you filter by belongs in a field.
5. When things are classified — straight after the import, not “later”.

Write it down. An agreement that has only been discussed lasts until the first holiday.

Set a rule and stick to it. What has proved its worth: the way the company writes its own name, with the legal form, without additions.

What you should avoid, because it creates separate files:

• German umlauts sometimes written as in “Müller”, sometimes as in “Mueller”.
• The legal form sometimes with full stops, sometimes without.
• Location additions for some companies but not for others.

If you suspect a problem, search for both variants — the files grow together as soon as the name has been made consistent.

With three habits:

• Search before creating. Part of a word is enough; the search covers your entire database.
• Pay attention to the email address when importing — that is how existing contacts are recognised and supplemented rather than overwritten.
• A spot check after every import. If the number of new contacts is higher than expected, duplicates have usually been created.

Merge any duplicates that have arisen; keep the record with the longer history.

Three, and everything else is a bonus:

• Email — without it, the contact does not appear in any campaign.
• Salutation — if it is missing, the neutral form from the template is used, and the recipient notices.
• A category — without it, the contact is invisible when selecting an audience.

On the overview you can see the difference between the total and the number that can be written to. If the gap is large, it is usually email addresses missing from an old import.

Because the location maps are assigned precisely via the postcode. If it is missing, the contact is missing from the map — not because it does not exist, but because it cannot be located.

The classic case: during the import, “01129 Dresden” stood in one column, and the assignment failed. Postcode and town belong in separate columns.

If a map stays empty, that is almost always the explanation — not an error in the evaluation.

By four signs that the system itself gives you:

• Hard bounces pile up — the address no longer exists, often because the person has left.
• Out-of-office messages naming a stand-in give the new contact person.
• Availability details for candidates that have not changed for months.
• A large gap between the total and the number that can be written to.

All four are occasions for maintenance — and all four come up in everyday work anyway. Use them instead of scheduling a maintenance day that never happens.

One named person for the rules — categories, spellings, structure — and everyone for the content.

The most common mistake is the reverse: if only one person may create contacts, shadow lists appear in private mailboxes, and they are worse than an untidy address book.

What works: everyone may create, one person tidies up and decides in case of doubt.

Before: 1. Header row present, one item of information per column? 2. No intermediate headings and no totals row? 3. Dry run done — do the numbers match what you expect?

After, on the same day: 1. Spot check: open three contacts — are the values in the right place? 2. Classify: set a category or tag. 3. Compare the total with the dry run.

Point 3 of the second list is the one that uncovers errors, and any import can be undone for 24 hours.

By one single test question: who will I want to write to together one day?

What follows from this:

• Category = the role in the business (customer, prospect, supplier). One per contact.
• Tag = everything else (industry, trade fair, origin). Several per contact.
• A few clearly divided ones beat many fine ones. Eight tags work; eighty are a search problem.

And the trap you need to know about: a campaign with an empty category selection does not send to nobody — sending to everyone requires a tick of its own, and the number of recipients in the confirmation prompt is the reliable figure.

For two reasons, and the second is the more important:

• Evaluation: only this way can you tell which source delivers useful contacts and which only noise.
• Proof: if someone asks where you got their data from — or complains — the import history is the answer. Every import remains listed with its result, even one that has been undone.

In practice: import a separate file per source and set a tag straight afterwards. That takes thirty seconds and is invaluable when it matters.

What belongs in them is whatever helps a colleague in the next conversation: “prefers calls in the morning”, “does not decide alone”, “project runs until March”.

What does not belong in them: assessments of people, assumptions about payment habits, anything you would not show the person concerned.

The reason is not politeness: notes can be part of a data access request. And everyone in the team sees what is written here anyway.

Rarely. A scheduled maintenance day usually does not happen, and if it does, the result lasts four weeks.

What works are occasions: classify after every import, check after every bounce, write the note after every phone call. Five minutes where you are anyway beat one day a quarter.

The only case where a block of time is worthwhile: once at the start, when databases from several lists are being brought together. Never again after that.

By four figures, all of which are on the overview:

1. Total versus “can be written to” — how much of your address book can be reached?
2. Bounce rate — how good are the addresses?
3. Distribution across categories — are there groups you can actually write to?
4. Unsubscribes — are the audience and tone right?

If one of these figures gets worse, that points to data maintenance — not to the software. And all four are measured, not estimated.

When there is no longer a purpose and no retention obligation stands in the way. This is not a question of tidiness but a duty: the GDPR does not require a particular date, but it does require a concept.

Practical guidelines:

• Application data from completed procedures: usually six months.
• Candidate profiles without contact and without consent: set a deadline.
• Completed form enquiries: after they have been dealt with.
• Contract and booking data: subject to tax retention periods — do not delete.

And before you delete a contact: notes, tasks, follow-ups and the email history go with it. For “do not write to them any more”, unsubscribing is the right way.