The standard way to choose a CRM is to build a comparison grid: features down the side, vendors across the top, ticks in the boxes. It feels rigorous and it is close to useless, for one reason — every vendor writes the grid that they win. The features that distinguish the products are not the features anyone puts in a grid.
What actually determines whether a CRM works is narrower and more boring: whether the people who were not in the buying meeting use it on a Tuesday in July. Twelve questions get at that.
First, name the leak
Before you look at a single product, write one sentence: "We are buying this because ____ is costing us ____ a month." If you cannot fill in both blanks, you are not ready to evaluate anything — you will end up choosing on interface taste, which is the most expensive way to choose software.
In most service businesses the blank is one of five things: leads answered too slowly, quotes never followed up, customer history scattered across phones, invoices sent late, or a schedule only one person understands. Each of those points at a different product. A tool that is superb at scheduling and mediocre at follow-up is the wrong purchase for a business whose leak is follow-up, no matter how the grid scores.
The twelve questions
Four groups. Ask all twelve of every finalist, and write the answers down — the differences are much sharper on paper than they feel in a demo.
- 01Does it hold the whole relationship, or just part of it?
Calls, texts, emails, quotes, jobs, photos, payments — how many of those live in the record, and how many live somewhere else with a link? Every gap becomes a place your team goes back to using their phone.
- 02What does it do without a human?
A CRM that only stores is a database with better fonts. Ask specifically: what fires automatically, on what trigger, and what stops it. If the answer to "what stops it" is unclear, that is a support problem waiting to happen.
- 03What does the field see on a phone?
Not "is there a mobile app" — what can a technician actually do, one-handed, in a driveway, without calling the office? If the honest answer is "view the schedule", the office will keep re-typing everything the field learns.
- 04How long until the first useful thing works?
Ask for a date, not a range. "One automation live in week one" is a real answer. "Full deployment in 8–12 weeks" means you will be paying for months before anything changes.
- 05Who does the setup, and what does it cost?
Implementation fees, data migration, and mandatory onboarding packages are often a larger first-year number than the subscription. Get all three quoted in writing before you compare monthly prices.
- 06What happens to your data if you leave?
Ask for the export format and whether it includes attachments, message history, and notes. A vendor confident in their product answers this immediately; the flinch is the signal.
- 07How many people need to be trained, and for how long?
Multiply the training hours by everyone who touches it. A tool that needs two days per person is not cheaper than one that needs twenty minutes, whatever the per-seat price says.
- 08Does it fit how you already sell?
If your business quotes on site and books by text, a system built around desk-based deal stages and email sequences will be fought every day. Software rarely changes how a trade actually works.
- 09What integrates, and how deeply?
There is a large difference between "integrates with QuickBooks" meaning invoices sync both ways and meaning a nightly one-way CSV. Ask which fields move, in which direction, and how failures surface.
- 10What does month three look like, not month one?
Every product demos well. Ask what customers typically stop using, what the common support tickets are, and what people wish they had set up differently. An honest answer here is worth more than any feature.
- 11Who is on the other end when something breaks?
Support hours against your hours. A service business does its admin at 6am and 7pm; a vendor answering 9–5 in another time zone is functionally unsupported at exactly the moments you need it.
- 12What is the contract, really?
Term length, auto-renewal, notice period, price increase at renewal, and whether the quoted price depends on annual prepayment. This is where the surprise lives.
The demo to insist on
Do not accept the standard walkthrough. Ask for a demo built on one of your real scenarios — a specific customer, a specific job type, a specific leak — and watch for three things:
- How much typing happens. Every keystroke in the demo is a keystroke your team will resent in July. Demos are performed by experts on rested hands; reality is not.
- How often they leave the screen. If the story requires four tabs, your team will need four tabs, and one of them will quietly stop being updated.
- What happens on the exception. Ask them to show a customer who replies "actually, can you come Thursday instead?" mid-sequence. Products differ enormously in how gracefully they handle the case that is not the happy path — and the exception is most of a service week.
Four red flags
| Red flag | What it usually means | What to do |
|---|---|---|
| No pricing on the website | Price varies by how much they think you will pay | Get the full first-year number in writing before demo two |
| Mandatory annual contract on day one | Churn is a known problem | Ask for a monthly term; the answer is informative either way |
| "That is on the roadmap" | It does not exist | Evaluate only what you can see working today |
| Setup fee larger than three months of subscription | The product needs consultants to be usable | Price the consultants in for year two as well |
How to actually decide
Shortlist three, no more. Run the twelve questions, take the demo on your own scenario, then pick the one where the leak you named gets fixed with the fewest clicks by the person least interested in software. That last clause is the whole test — the buyer is never the constraint, the reluctant user is.
And keep the trial narrow. Move your active customers and open quotes only, switch on the one automation that addresses the leak, and give it two weeks against real work. A trial that tries to model the whole business proves nothing and expires before anyone has learned anything.
Common questions
What should a service business look for in a CRM?
Start by naming the leak — slow lead response, unfollowed quotes, scattered history, late invoices, or a schedule only one person understands — then evaluate whether each product fixes that specific thing in a few clicks, on a phone, without the office re-typing anything.
How many CRMs should I shortlist?
Three. Beyond that, the evaluations blur and the decision drifts toward interface taste rather than fit. Run the same twelve questions and the same real scenario through each one, and write the answers down.
What questions should I ask a CRM vendor?
The high-signal ones are: what fires automatically and what stops it, what the field can do one-handed on a phone, how long until the first useful thing works, what setup costs on top of subscription, what the data export includes, and what the contract does at renewal.
Is a free CRM good enough for a small service business?
Free tiers are usually built for desk-based sales rather than field work, and the limits tend to appear exactly where service businesses need depth: automation, texting, job records, and mobile capture. They are a reasonable place to learn what you need, not usually where you stay.