How to choose a CRM for a small ops team

A CRM is a long-term decision dressed up as a quarterly one. The data you put in is the same data you'll be making business decisions on three years from now — through the next pricing change, the next acquisition, and the next platform pivot of whichever vendor you pick.

This guide is the checklist we walk operators through when they ask. It's twelve questions, organized in four buckets, in the order the answers actually matter.

Before you start: who is this for

A CRM for a 5,000-seat enterprise org is not the same product as a CRM for a 6-person ops team. This guide is the latter. If you're at the former, your problems are different — read Salesforce alternatives instead, or talk to a consultancy.

The kind of operator we're writing for:

• 1–30 people total.
• Sales is a few founders or a small revenue team, not a 50-rep org.
• You don't have a Salesforce admin and you don't want to hire one.
• You'd rather have agents working the pipeline than a bigger team.

Bucket 1 — Ownership

The questions you ask before you ask anything else.

1. Where does the customer list live?

If the answer is "in our cloud," you're renting your customer relationships. Read your contract. The vendor's "your data is your data" line is marketing, not law. The legal reality is what the export clause says.

A good answer here is "on your machine, in a file you can open without us." A workable answer is "in our cloud, but you can export the canonical model in one click, no rate limit." A bad answer is "contact support for an export."

2. What's the exit cost?

Pull up the export documentation right now, before you talk to sales. Is the export the same shape as what you put in, or is it a downgraded CSV that loses your associations, custom fields, and activity history?

The export determines your bargaining position for the rest of your relationship with the vendor. If you can't leave cheaply, every pricing change is a tax.

3. What does the vendor see?

Marketing software that knows the names of your customers' competitors is a security event waiting to happen. Read the data-processing addendum. If the vendor trains models on aggregate customer data — even with anonymization — that's something to weigh.

Bucket 2 — Capability

What the system can actually do, beyond storing contacts.

4. Can agents do work, or just suggest it?

Every CRM in 2026 has "AI." Most of it is a chat assistant that summarizes what you already know. Real agent capability looks like:

• A worker that can claim a task and complete it without you in the loop.
• A separate verifier that checks the work before it lands.
• An audit trail of every action, including the model name and prompt.
• A budget the worker can't exceed without supervision.

If the CRM's AI is a sidebar that drafts emails, that's a copilot. That's fine, but price it as a copilot, not a worker.

5. Does the pipeline auto-progress?

The unglamorous middle of the sales motion — moving a lead from new → qualified → contacted → proposal — is where ops time goes to die. A modern CRM should:

• Move stages automatically on evidence (email reply, meeting booked, call logged).
• Or at least surface "these deals haven't moved in 14 days" to one place.

If progressing the pipeline requires a human in every cell, you're buying a spreadsheet with a logo.

6. Are sequences a feature or an add-on?

Outbound is part of the job. If the CRM charges separately for sequencing, the integrated price is what to compare. If sequences are bolt-ons via Zapier, you're stitching two products together — and you'll find out at the year mark.

7. Does the schema fit your business?

The CRM you buy will outlast the first three custom-field exceptions you ask for. Look at the data model without the UI — most vendors document it. Can a contact have multiple companies? Can a deal belong to two contacts? Does a "lead" become a "contact" by changing a status, or by being created as a new record? The wrong answer here is a long argument with your data later.

Bucket 3 — Cost (real)

The price the vendor advertises is the floor, not the ceiling.

8. What's the year-3 per-seat math?

Plot it. Pull the public pricing page, take the seat count you'll be at in 36 months, multiply. Don't forget:

• The upgrade trigger — sequences in the next tier, reporting in the one above that.
• The mandatory hubs that ship together. Many CRMs will sell you the marketing module the moment a marketer joins.
• The partner ecosystem cost — the consultants you pay to make it work.

If the floor at year 3 is more than your COGS line for ops, you bought the wrong CRM.

9. What's the cost of leaving?

Cost ≠ "is there an export button." Cost = the engineer-time and consulting fees to map your CRM's model onto the next one. If your CRM has custom objects, custom fields, and ten Zapier integrations, the cost of leaving is real money.

The right CRM for you is one where the exit cost is low enough to negotiate from. A vendor who knows you're a hostage will price like it.

Bucket 4 — Operational reality

Day-in, day-out, what life is like.

10. Who actually runs it?

A CRM that needs a certified admin is a CRM that needs a certified admin's salary. For a 6-person ops team, that's wrong. Ask:

• Can a non-admin add a stage to a pipeline?
• Can a non-admin add a custom field?
• Can a non-admin write a report?

If any of those is "talk to your admin," you're buying admin time too.

11. How does the system handle imports?

You will import. Vendors that fail at import — that demand a specific CSV shape, that lose phone formatting, that don't dedupe — fail at it for everyone. Run an import test before you sign.

12. What's the support story when it breaks?

Read the SLA. Look at the public status page for the last 12 months. Search the vendor's name on Hacker News and skim the negative comments — every popular CRM has them, the question is which complaints are deal-breakers for your team.

The honest summary

Most small ops teams pick the CRM their LinkedIn feed has heard of. The right CRM is the one that answers the twelve questions above honestly at the price you'll pay 36 months in.

For the operators we build Excellent for, the answer turns out to be: a local-first CRM where the data is yours, agents do the work, the verifier signs off, and the export is a cp command.

If that's the kind of answer you're looking for, the HubSpot comparison, Salesforce comparison, and the best HubSpot alternatives page apply this framework concretely.