What Maintaining a CRM Actually Involves

Every article about CRMs warns that they need maintaining. Almost none says what that means, how long it takes, or who does it — which is unhelpful when the answer to "who" is "nobody in particular".

So here is the actual work, with honest timings, sized for a business where the person doing it has no admin title and no spare afternoon.

The weekly pass — fifteen to twenty minutes

One sitting, same time each week. It is short, and the order matters because the first item is the one that keeps the whole thing honest.

1. Close what is dead

Look at every open deal with no activity for a month. For each: is this alive?

Most are not, and closing them is the single most valuable thing on this list. An open deal that died in March makes your pipeline a work of fiction, and a pipeline that overstates reality is worse than no pipeline — you make decisions on it.

There is real resistance here, because marking things lost feels like admitting failure. It is the opposite: a clean "lost, no response" is information, and a deal left open forever is the failure.

2. Clear overdue follow-ups

Anything whose date has passed gets one of three outcomes: do it now, give it a new date with a reason, or drop it.

What must not happen is leaving it overdue. A list with fourteen overdue items stops being read, and once follow-ups are not read the CRM has lost its main practical use.

3. Log what has not been logged

The conversations from the week that never made it in. Two lines each — what was said, what happens next.

If this step regularly takes more than a few minutes, the problem is not the maintenance routine, it is friction at the moment of logging, and that is a product or configuration issue worth fixing at source. It is the leading cause of CRM projects failing.

The quarterly pass — about an hour

Duplicates

They arrive however careful you are: someone enters a contact that already existed under a different spelling, an import creates a second copy, a person changes email address.

Duplicates are corrosive out of proportion to their number, because they produce the exact experience that starts the distrust loop — you look someone up, find two records, and cannot tell which is current. Merge them quarterly while there are still few enough to do by hand.

Archive

Anything you will not act on: former clients with no prospect of return, contacts at companies that no longer exist, leads from three years ago.

This feels like destroying value and is the opposite. Every dead record makes the live ones harder to see. Archive rather than delete — the data is kept, it just stops pretending to be current.

Check the shape still fits

Look at your pipeline stages and ask whether they still describe how you sell. Businesses change faster than their CRM configuration, and stages that no longer match reality are quietly filled in wrongly rather than corrected.

A stage nothing has sat in for six months should probably go. A stage everything sits in for weeks probably needs splitting.

Check the fields

Any field that is empty on most records is doing nothing except adding visual noise and implying incompleteness. Remove it, or make it genuinely optional and stop looking at it.

What is not maintenance

Worth separating, because these expand to fill whatever time is available and produce much less than the list above.

Adding fields and stages. Feels productive, adds ongoing cost. Only do it when the absence has demonstrably cost you something.

Building automations. Occasionally worth it, always deferrable. Automating an unstable process encodes the instability.

Reorganising. If the impulse is "this would be tidier as…", stop. Tidiness is not the goal; accuracy is, and reorganising usually reduces accuracy for a fortnight.

Who does it, when nobody's job it is

Ownership at this size is not a role. It is a name and a recurring slot.

The right person is whoever most depends on the data being right — usually whoever does the following up, because they feel the cost of a stale record immediately. Not the most senior person, and not whoever is most interested in software.

Two conditions make it stick: it is in the calendar, and it is short enough to actually do. Fifteen minutes on a Friday survives a busy week. "Tidy the CRM" as an aspiration does not survive anything.

How the product choice changes the workload

Maintenance is not fixed — it scales with how much product you have switched on.

A CRM with one pipeline, few custom fields and automatic email logging needs very little attention, because most of the record maintains itself and there is not much to go stale. That is the argument for products like Zoho Bigin's free plan — one pipeline and 500 records is not just cheap, it is a small surface to keep true — and for Less Annoying CRM, where a single tier means no configuration drift between what you have and what you could have.

A platform with three pipelines, twenty custom fields and a dozen automations needs someone thinking about it. That is fine when someone is. It is how CRMs die when nobody is — the reasoning behind choosing for the administration you actually have.

The whole routine, in one place

• Weekly, 15–20 min: close dead deals, clear overdue follow-ups, log the week's conversations.
• Quarterly, ~1 hour: merge duplicates, archive dead records, check stages and fields still match reality.
• Never: add structure because it would be tidier.

That is the entire cost of ownership beyond the subscription — or rather, it is the part people forget. The rest is in what a CRM really costs.