Why Your CRM Is Only as Good as Your Habits

Most abandoned business systems were not badly built. They were configured optimistically, used properly for a month, and then gradually stopped reflecting reality — not through any decision, but through a hundred small omissions that each seemed reasonable.

Understanding why the omissions are reasonable is the only route to fixing them, because the usual response — reminding people to keep the system updated — has never worked anywhere.

The asymmetry that causes everything

Recording information costs the person doing it. The benefit goes to somebody else, later.

A salesperson logging a call gains nothing; they already know what happened. The benefit accrues to whoever picks up the relationship next month, or to the person trying to understand why deals stall. That is a real cost paid now for a diffuse benefit paid to others.

People are not being lazy when they skip it. They are responding accurately to the incentive in front of them. Any fix that does not change the asymmetry is asking people to be virtuous indefinitely, which is not a plan.

The three ways it degrades

Selective recording. The good news gets entered; the ambiguous does not. Deals that are progressing get updated because updating them is satisfying. Deals that have gone quiet do not, because there is nothing to say and saying nothing feels like admitting something. The system therefore develops a systematic optimism bias, and any forecast built on it is wrong in a consistent direction.

Deferred recording. Everything gets updated on Friday, from memory, in one session. The facts survive; the timing does not. Since much of the value of the record is in sequence and interval — how long between contact and reply, how long a deal sat — reconstructed timing quietly destroys most of what the data was for.

Parallel recording. The real state of play lives in someone's notebook, inbox, or head, and the system holds an approximation maintained for management. Once this happens, the system is a reporting obligation rather than a working tool, and it will get exactly the care that implies.

Why more required fields make it worse

The instinctive response to poor data is to require more of it. Make fields mandatory, add validation, enforce a next step on every record.

This reliably backfires, and the mechanism is worth understanding. Mandatory fields do not produce accurate data; they produce whatever passes validation. A required "next action date" gets a date two weeks out, chosen because it is far enough away to avoid a nag. The field is now full and means nothing, which is worse than empty — an empty field is honest about being unknown.

The discipline that works is the opposite one: fewer fields, each with a person who reads it and a decision it changes. That is the same rule that keeps any record trustworthy — not collecting things speculatively "in case it is useful later" [1]. A system with four fields that are always right beats one with twenty that are usually approximately right, because the first can be acted on without verification.

Make the record a by-product

The only durable fix is to arrange things so that doing the work produces the record, rather than the record being a separate task performed afterwards.

If the email is sent from the system, the email is logged. If the quote is produced there, the quote is attached. If the call is dialled there, the call has a timestamp nobody typed. The person does their job; the history accumulates behind them.

Every place where recording is a separate action is a place where the record will eventually diverge, and no amount of process discipline holds that indefinitely. This is why integration into the actual working tools matters more than feature depth: a system people must visit to update is a system that gets updated late, in batches, by someone reconstructing the week.

Two habits worth the effort

Not everything can be a by-product, and two things are worth doing deliberately.

Write down why something stopped. One sentence when a conversation ends. It is the only field that distinguishes "we are too expensive" from "we were too slow to reply", and those have opposite remedies.

Write internal notes you would be content to show the customer. This is a practical rule, not a moral one. Customer records are personal data, and the people described have the right to receive it "in a structured, commonly used and machine-readable format" and to take it elsewhere [3]. A complete export — for us, a JSON export of all your records [2] — includes the notes. A team that has internalised this writes better notes anyway: more factual, less speculative, and considerably more useful to the colleague who reads them next month.

The manager's habits set everyone else's

There is one factor that predicts whether a system stays current better than any feature, training programme, or policy: whether the people who ask for things use the system to find their answers.

If a manager wants to know how a deal is progressing and asks the person directly, they have communicated something precise — that the record is not where answers live, and that maintaining it is an administrative obligation separate from the real work. Everyone updates it accordingly, which is to say minimally and late.

If the same manager looks it up first and only asks about what is genuinely not there, the message reverses. The record becomes the thing that represents your work to the rest of the business, and keeping it accurate becomes self-interested rather than dutiful.

This costs nothing and is harder than it sounds, because asking a person is faster in the moment and the record is often incomplete at the start. The first few weeks involve looking things up and not finding them. That is the transition, and shortcutting it is how organisations end up with a well-configured system nobody trusts.

What to do about a system that has already drifted

Do not attempt to bring it up to date. A clean-up campaign restores accuracy for a fortnight and changes nothing about why it decayed.

Instead, cut the fields to the ones people actually use. Move the recording into the tools where the work happens. And pick one habit — the closing note is the highest-value candidate — and do only that one until it is automatic.

A small system that is true is worth considerably more than a comprehensive one that is out of date, because the first can be acted on and the second has to be checked against reality before anyone dares use it.