✚Data maintenance as a routine

Because everything a CRM can do depends on it.

Specifically:

• A campaign is only as good as the selection. If the categories are not maintained, it hits the wrong people.
• An evaluation is only as good as the database. Nobody believes figures from an untidy database — and rightly so.
• An access request is only as good as the record of origin. If you do not know where a contact came from, you can neither provide information nor justify why you are writing to them.

The point that weighs most: a database nobody trusts is no longer used. Then everyone goes back to their own lists, and the introduction was in vain.

This is not a question of tidiness but the very basis of the system.

Three moves, about a minute each:

1. Create a new contact immediately. Do not collect them, do not jot them on a scrap of paper, do not “enter them later”. Later is never.
2. A note after the conversation. Two sentences, while you still have the phone in your hand.
3. Set a follow-up, if something remains open.

On top of that, a habit that costs nothing: when opening a file, briefly check whether the details are correct. Wrong phone number, outdated contact person, new company name — you notice these exactly when you are looking anyway.

Whoever does these four things only needs the quarterly maintenance for fine-tuning. Whoever does not will never catch up.

One hour, always the same five steps:

1. Find and merge duplicates. They arise with every import and with every contact who writes from a new address.
2. Go through contacts without a category and classify them. This list is the best measure of how well the data is maintained.
3. Sort out hard bounces. Addresses that no longer exist do harm with every further send.
4. Look at contacts without an email address — can they be completed, or do they only belong in the phone directory?
5. Go through overdue follow-ups: complete, postpone or cross off. Do not ignore them.

Fix the date; do not go by feel. Data maintenance with no date does not happen.

What you only need once a year, and what is legally due:

• Apply deletion periods. Application documents, expired consents, contacts with no purpose at all.
• Check user accounts. Who still works here? Which account is no longer needed?
• Reconsider categories. Do they still fit the business? Is there one nobody uses any more?
• Look through templates. Are the details, prices and contact people correct?
• Spot-check records of origin. If you cannot say for ten random contacts where they came from, you have a problem that is still small now.
• Update the record of processing activities.

Two hours in January, and you are able to answer access requests for the rest of the year.

Merge rather than delete — otherwise you lose the history of one of the two entries.

How duplicates arise:

• Import without matching or with a different spelling.
• A second email address for the same person.
• Two colleagues create the same contact.
• A change of name — marriage, change of employer.

When merging, you decide which detail applies. Rule of thumb: the most recently confirmed, not the most recently entered.

Check before you merge: is it really the same person? Two people with the same name at the same company do exist — and a wrong merge is hard to undo.

And as prevention: use the dry run when importing. It shows duplicates before they arise.

By one question: would I write a different mail to this group?

If so, it is a category. If not, it is a tag or nothing at all.

What distinguishes good categories:

• Few. Three to seven. Nobody maintains more.
• Unambiguous. Everyone in the team understands the same thing by them. Write down one line of explanation per category.
• Complete. Every contact has exactly one — otherwise it drops out of every selection.

What distinguishes bad categories:

• States that change (“written to”, “waiting”) — that is what campaigns and follow-ups are for.
• Combinations (“Prospect trades Saxony”) — those are three characteristics.

If 80% of your contacts end up in “Other”, the division is wrong.

Check first, then decide — and the decision is rarely “delete”.

Step by step:

1. Is there a reason to keep them? A former customer is not a dead entry but a warm contact.
2. Is the data still up to date? A call or a short email clarifies that, and is at the same time a reason to get in touch.
3. Does an email bounce? Then the address is dead and should be sorted out.
4. No purpose any more? Then the obligation to delete applies, not just a love of tidiness.

What you should not do: send a “Are you still there?” campaign to old contacts once a year. That is advertising and needs the same legal basis as any other.

The most useful sorting for this: by last contact, oldest first.

For three reasons, and the first is the compelling one:

1. Duty to provide information. The origin is expressly part of the information under Art. 15 GDPR. If you do not know it, you cannot provide the information.
2. Legal basis. Whether you may write to someone depends on how you obtained the address. Without the origin, that cannot be proven.
3. It tells you what works. After a year you can see which source brings the contacts that turn into business — and which only creates work.

The effort: three words when creating. “Leipzig trade fair 2026”, “Recommendation Krause”, “Website form”, “Research”.

This is the detail that cannot be reconstructed afterwards, and therefore the only one where carelessness is irreversible.

That a colleague can understand it in six months without having to ask. Four components:

1. Occasion — call, meeting, reply to a campaign.
2. Content in keywords, not a retelling.
3. Agreement — who does what by when.
4. Next step, ideally as a follow-up straight away.

What does not belong in it:

• Judgements about people. They can be requested in a subject access request, and then appear in a letter from your business.
• Special categories of data — health, ethnic origin, religion. Not even in passing.
• Abbreviations only you know.

The test: write as if the person concerned were reading along. They may well do so.

Through the responses that come back — they are the most reliable source, and they arrive by themselves:

• Hard bounce: the address no longer exists. Sort it out, and if the contact is important, follow up by phone.
• Out-of-office message saying “has left the company”: worth its weight in gold. It often names the successor straight away.
• Reply from a different address: enter the new one, check the old one.

Plus two active moves:

• Have it confirmed briefly during a phone call: “Can I still reach you at this address?”
• Check company names when a business has been renamed or taken over.

What you can leave out: a systematic check of the whole database. That is never finished. The responses take care of most of it by themselves — you just have to deal with them.

Separated by type:

• Notes and follow-ups: everyone for their own matters. That cannot be delegated.
• Master data quality — duplicates, categories, spellings: one named person.

The second point is the one usually missing. Data maintenance that nobody is responsible for does not happen — regardless of how convinced everyone is that it would be important.

It does not have to be a big role: one hour a quarter is enough for most databases. But it has to be someone, and it should be in the calendar.

In small businesses, that is the person who works most in the system anyway. In larger ones, it belongs in a job description.

By five signs — one is enough to take countermeasures:

1. Someone says “I’d rather look in my own list”. The first and clearest sign.
2. Campaigns find fewer recipients than expected. Usually because categories are missing.
3. The number of contacts without a category grows.
4. The bounce rate rises across several campaigns.
5. Notes become rarer. You only notice this when you open a file and nothing has been written in it for eight months.

What helps then: do not tidy up everything at once. That is the reflex, and it leads to a big project that is never finished.

Instead: name one thing, set a date, get it done. Next quarter, the next one.

Not with the whole database — with the part you need next.

The procedure that works:

1. Filter out the audience of the next campaign — usually a few hundred contacts.
2. Tidy up only these: category, address, duplicates.
3. Run the campaign.
4. Next time, the next group.

After half a year, the part of the database you actually use is clean, and you never missed the rest.

What does not work: a clean-up project across the whole database. It is never finished, and after three weeks nobody is doing it any more.

And the decision that saves the most along the way: whatever you have not touched in two years, you do not tidy up — you sort it out.