Key takeaways
- Start with failure modes — mailbox, post-sale, channel, deal value, AI permissions — not a feature matrix.
- Adoption in month nine predicts success better than ticks in the demo.
- The advertised seat price is often under half of three-year total cost.
- A two-week pilot with real data beats a six-week spreadsheet evaluation.
How to choose a CRM
- List your failure modes. Mailbox, post-sale work, the channel customers reply on, how deal value is set, and whether AI respects permissions.
- Build a three-year cost table. Seat licence, campaign module, WhatsApp connector, AI usage, implementation, internal admin and migration.
- Weight adoption highest. Score adoption 40%, revenue motion 30%, three-year cost 20%, feature breadth 10%.
- Run a two-week pilot. Give each finalist three reps and real data. Ban the spreadsheet. Count unprompted activity.
Most CRM evaluations follow the same path. Someone builds a feature comparison spreadsheet. Three vendors demo. The one with the most green ticks wins. Eighteen months later the system is half-abandoned and the pipeline review still runs off a spreadsheet somebody maintains by hand.
That is the standard answer to how to choose a CRM, and it fails for a consistent reason: the feature matrix wasn't wrong, it was measuring the wrong thing.
How to choose a CRM: start with the failure modes, not the features
Every CRM in serious consideration will do contacts, deals and stages. Nobody loses that comparison. What actually decides the outcome is whether your team uses it in month nine.
Ask these first.
1. Where does a rep's email live?
This is the single strongest predictor of adoption. If the CRM requires a rep to work in Outlook and then log something in the CRM, they will do the first and skip the second — not out of laziness, but because duplicate data entry always loses to a full calendar.
There are three answers a vendor can give:
- A plugin or sidebar. Works, sometimes. Breaks on Outlook updates. Reps still have two windows.
- A sync. Emails are copied in after the fact. Threads fragment; replies arrive somewhere the CRM can't see.
- The CRM is the mailbox. Read, reply, file and template from inside it. This is what Qivity does — the thread on a deal is the actual thread, not a copy.
2. What happens to a customer after "Won"?
Roughly half of CRM buyers have renewals, maintenance contracts, service visits or subscription true-ups — and roughly none of the mainstream CRMs handle them without an add-on module or a second product.
If that's you, ask specifically: where does the AMC live? Who gets told when it expires? Can I raise an invoice against it? A CRM whose data model stops at the close date will push the second half of your revenue into a spreadsheet. (This is exactly why we built contracts in.)
3. Which channel do your customers actually reply on?
In India, the Gulf, South-East Asia, Africa and much of Latin America, the honest answer is WhatsApp — a channel with over two billion users. If your CRM treats it as a marketplace integration billed by a third party, you are one connector outage away from losing your primary channel — and the conversation history is sitting on someone's personal phone in the meantime.
4. Where is the deal value coming from?
If it is a free-text number a rep types in, your forecast is a creative writing exercise. Look for a product catalog that computes the value, and stage gating that makes it mandatory before a deal reaches the board pack.
5. Can the AI see things the user can't?
Ask them to demo it. Log in as a junior rep and ask the assistant about the VP's pipeline. Permission-scoping an AI layer is hard, and a surprising number of products bolt the assistant on top of an unrestricted data view.
What the demo will hide
Three things, reliably:
Import. Demos run on clean seeded data. Ask them to import your export, live, with its duplicate records and inconsistent country fields. The vendor who agrees is the one whose import actually works.
The tenth screen. Demos show five screens. Ask to see the ugly ones — settings, user management, permissions, bulk edit. That is where your admin will spend their life.
Mobile. Ask them to do the whole demo from a phone.
The total cost nobody quotes
Build this line by line before you sign:
| Line | Frequently forgotten |
|---|---|
| Seat licence | The advertised number |
| Marketing / campaign module | Often a separate hub |
| WhatsApp or SMS connector | Third-party, metered |
| AI usage | Metered separately on most platforms |
| Implementation partner | Weeks of consulting on enterprise platforms |
| Internal admin time | Often 0.25–1 FTE, permanently |
| Data migration | Once, but rarely small |
The advertised per-seat price is frequently under half the real number. A cheaper licence that needs a full-time admin is not cheaper.
A scoring model that works
Weight the evaluation like this rather than counting ticks:
- Adoption likelihood — 40%. Mailbox integration, mobile, speed, number of clicks to log a call.
- Covers your actual revenue motion — 30%. Including whatever happens after Won.
- Total cost over three years — 20%. Everything in the table above.
- Feature breadth — 10%. Yes, really. It is the least predictive input, and it is the one every spreadsheet over-weights.
Run a two-week pilot, not a six-week evaluation
Pick your two finalists. Give each three real reps and two weeks with real data and a rule that the spreadsheet is not allowed. Then look at how many activities were logged without anyone being nagged.
That number tells you more than any matrix.
If you want a straight answer about whether we fit your motion — including "probably not, here's who does" — talk to us. We would rather lose the deal than lose the renewal.