Skip to content

The customer record a small business actually needs

Why most CRMs fail service businesses, what belongs on a contact record and what does not, how duplicates really get made, and the four fields that earn their place.

Frontiva · · 5 min read

Most CRMs were built for sales teams with long pipelines and named accounts. A dental practice, a med spa and a plumber have none of those things. They have a lot of people, each of whom interacts briefly and repeatedly over years, across whatever channel was nearest to hand.

Used as designed, a sales CRM makes that worse. This guide is what a customer record should actually hold for a business like yours, and the two things that decide whether it stays useful.

The problem is not storage, it is identity

You already have the data. It is in six places: the phone's call history, a text thread, an Instagram DM, an inbox, the booking software, and somebody's memory.

The same person appears in all six under different identifiers. A mobile number in one, an email in another, an Instagram handle in a third. Nothing joins them, so nobody can answer the only question that matters at the front desk: has this person dealt with us before, and what happened?

A CRM earns its place by answering that in one screen. Everything else it does is secondary.

Why duplicates get made, and it is not carelessness

Duplicate contacts are usually blamed on staff. They are almost always caused by inconsistent normalisation.

An inbound text arrives from +15551230000. A receptionist types the same number as (555) 123-0000. If the software stores what it was given, those are two different strings, so they become two different people. The customer texts, gets a thread with no history, books an appointment, and now has two records with half a story each.

The fix is that every identifier is normalised on the way in, by one piece of code, whatever door it came through. Phone numbers to a single international format. Email lowercased. Same rule for the webhook, the manual form and the import.

This is worth testing before you commit to a system. Add a contact by hand with a prettily formatted phone number, then send a text to your business from that number. If you end up with two records, you will be cleaning up duplicates forever.

What belongs on the record

Four things earn their place. Most CRMs offer forty.

Identity. Every way of reaching them, joined: phone, email, social handles. Plural, because people change numbers and keep email.

Timeline. Every interaction, in order, across every channel, including the ones the AI handled and the appointments they attended and missed. One list. The value is entirely in it being one list: a timeline split by channel is six timelines and nobody reads six.

Consent. What they have agreed to be contacted about, on which channel, when they agreed, and the exact wording they saw. Not a checkbox. The wording, stored, because the question a complaint asks is "what did they agree to" and a boolean cannot answer it.

A few notes a human wrote. Nervous patient. Pays by card. Prefers mornings. Brother of another customer. This is the part that makes your team look like they remember somebody, and it is the part no automation generates.

What does not belong

Lead stages borrowed from software sales. Prospecting, qualification, proposal, negotiation. A service business has an enquiry, a booking, and a visit. Three states. Forcing seven on it means five are always empty and the pipeline view lies.

Lead scores nobody can explain. A number between 0 and 100 with no visible reason is ignored within a fortnight, correctly. If you use scoring, the record has to say why: replied within an hour, asked about a specific service, has attended before. A score you can read is a score somebody acts on.

Custom fields nobody fills. Every unfilled field makes the record harder to read. Add one only when you can name the decision it changes.

The rule that keeps it useful

Nothing is entered twice.

The moment your team has to write down something the software already knows, the record starts rotting, because that step gets skipped on a busy Tuesday and nobody notices until the data is half-real.

In practice this means the contact record is written by the things that already happen: a text arrives and the contact is created or matched, an appointment is booked and it lands on the timeline, the AI answers and that is on the timeline too. The only thing a human should type is the thing only a human knows, which is the notes.

If a demo shows you a lovely record and every field on it was typed by a person, ask who types them on a Tuesday afternoon with four people waiting.

Merging, and why it needs to be reversible

You will have duplicates from before, and imports will make more. So merging matters, and the detail that matters most is that a merge can be undone.

Merges are done quickly, by someone fairly sure, and being wrong means two people's histories are now one person's. An irreversible merge turns a small mistake into a permanent one. Ask what happens if a merge was wrong, and treat "be careful" as the wrong answer.

About importing your old list

Two things to know before you import a spreadsheet.

Consent does not come with it. A contact imported from your old system has, from a legal standpoint, agreed to nothing you can evidence. Software that marks imported contacts as consented is inventing permission on your behalf. Expect imported contacts to start unable to receive reminders, and expect to capture consent as people come in. It reads as a nuisance and it is the honest state of your list.

Import the people, not the noise. Old lists carry dead numbers, people who moved away and duplicates. Importing everything gives you a database that is 40% wrong and a team that stops trusting it. If you cannot tell which half is real, import the last two years and leave the rest in the spreadsheet.

The short version

A customer record for a service business exists to answer one question at the front desk: have we dealt with this person, and what happened. That needs joined identities, one timeline across every channel, real consent with its wording, and a few human notes. It needs everything else to be written automatically, duplicates to be hard to make and easy to undo, and imported contacts to arrive honest about what they have agreed to. Pipelines, scores and custom fields are what you add later, if you can name what they change.

Start free. Be live this week.

Built for dental, med spa, home services, and the businesses that live on the phone.