How do you choose a CRM for a B2B company? a selection method, not a tool ranking

01

Why write the process down before you watch the first demo?

A vendor demo is built to look good: three stages in the pipeline, one salesperson, no duplicates, every field filled in. Your sales does not look like that. Until you have it written down, you are comparing somebody else's example with somebody else's example and picking whichever clicks more nicely.

The description does not have to be a formal process map. One page per stage is enough: what triggers it, who performs it, what has to be true for the deal to move on, and where the data comes from. Do it with the person who actually sells, not the person who reads the report at month end. That page later becomes the list you check every candidate against.

• Pipeline stages and the condition for moving on, written as a sentence, not as a stage name
• Who plans the next step and how they know it needs planning
• The tools standing next to the CRM: email, quotes, invoices, tickets, delivery
• Objects tied to a customer: contract, device, subscription, site, project
• Every place where somebody retypes data from one screen into another

02

Which requirements are must-haves and which are only nice to have?

An undivided requirements list is useless, because every system meets some part of it and the vendor will show you exactly that part. Split it into two columns. Must-have covers the requirements whose absence stops work or pushes it into a spreadsheet beside the CRM. Nice to have is everything else, including whatever makes the biggest impression in a demo.

One rule: you settle the choice on the must-have list alone. If two systems meet it equally well, then and only then do you look at the nice-to-haves and at the price. Below are six requirements that B2B companies most often file in the wrong column.

The middle column matters more than the first. You recognise a must-have by the manual work or the second register that appears beside the CRM when it is missing. With a nice-to-have, nobody can say who will use it in the first quarter.

03

What should you ask the vendor, and what should you ask yourself?

The vendor conversation is worth having only once the two lists exist. Then you ask closed questions with checkable answers instead of listening to what the platform can do. The three vendor questions below remove half the candidates in the first week.

The other three you ask yourself, and those usually decide whether the rollout survives a year. They are about who owns the system, what happens to the data on the way out, and what exactly stops being done by hand.

• For the vendor: which tier carries each item on my must-have list?
• For the vendor: what does a full export contain, and does it include activity history and custom fields?
• For the vendor: what is inside the implementation quote and what appears as a separate invoice line?
• For yourself: who maintains this system a year from now, and how many hours a week do they have?
• For yourself: what happens to the data when we want to leave, who takes the backup and how often?
• For yourself: which task stops being done by hand, and what number will show that it did?

04

How do you test a CRM on your own data instead of the sample set?

A trial account takes minutes to set up and it is the cheapest part of the whole decision. Load a hundred real records from the last few weeks, and make sure the ugly ones are in there: a company with no VAT number, a duplicate left over from a merger, a deal with no organisation attached. A demo shows the clean case. Your data shows the truth.

Then walk one deal from enquiry to signature the way a salesperson does. Count the clicks and the required fields. If entering a single lead takes more than a minute, the team will start avoiding the system, and you will find out a quarter later from an empty report.

Run the same test on an integration, because that is where the gaps the interface hides usually surface. How we wire a CRM to invoicing, email and fulfilment is described under CRM implementation.

• A hundred real records, at least five of them broken, loaded by import rather than by hand
• One deal taken from enquiry to signature, with the lead entry timed
• One report you genuinely use, rebuilt in the new system
• One integration wired up as a trial, even if it only reads from the invoicing system
• A full export at the end of the test, opened as a file to see what actually came out

05

What does a CRM really cost beyond the licence?

The vendor page shows one line: price per user per month. The first-year bill usually carries six. Not because anyone is hiding costs, but because five of them arise on your side and nobody invoices them separately.

We do not quote vendor rates here, because they change several times a year. Current numbers sit on the HubSpot, Pipedrive, Salesforce and Zoho price lists, linked in the sources at the bottom of this page. Work them out on your target headcount for a year, and on the tier that actually carries your must-haves.

A cheaper system can end up more expensive. Three tools bolted on beside it, plus the hours spent retyping data between them, can cost more than one higher tier, except that cost has no single invoice and nobody sees it at a budget review.

06

What should you migrate when you switch systems, and what should you leave?

The default answer is „everything, just in case”, and that is the trap. Migrating everything carries the mess across with the data: duplicates, free-text fields instead of picklists, contacts at companies that stopped existing four years ago. The new system then shows you the same disorder at a higher price, and the team concludes the choice was wrong.

Migration is the one moment when cleaning data has a natural sponsor, because somebody has to touch it anyway. Use that. Move the core and leave the rest in the old system's archive, or in a file you open once a quarter.

• Move: active companies and contacts, open deals, the last 24 months of history
• Move: the fields somebody actually filters on, converted from free text into picklists
• Archive: deals closed years ago and contacts without a single activity
• Merge before import: duplicates matched on VAT number and company domain, not on name
• Agree before import: the technical field names, because some systems let you set them only once

07

How do you make this decision in two weeks?

A CRM choice can hang for quarters, because both sides have arguments and nobody wants to sign off on a mistake. Two weeks is enough if, instead of comparing features, you test the candidates against your own process and your own data. That is the order we run it in on the process workshop.

1. 01

   Write the process down before you open a price list

   Stages, conditions for moving on, step owners, the tools beside the CRM and the places where somebody retypes data by hand. You do this with a salesperson, not with the person who reads the report.

2. 02

   Split the requirements into must-have and nice to have

   Must-haves are the ones whose absence creates manual work or a second register beside the system. The must-have list should fit on one page, otherwise it is not a must-have list.

3. 03

   Ask six questions: three of the vendor, three of yourself

   Vendor: the tier carrying the deciding feature, the scope of the export, the edge of the implementation quote. Yourself: who owns the system in a year, the exit plan for the data, the task that stops being manual.

4. 04

   Test on a hundred of your own records

   Trial account, import with the ugly data included, one deal from enquiry to signature, one real report and a full export at the end. Two days per candidate is enough.

5. 05

   Cost it over three years and decide

   Licences at target headcount, implementation, migration, training, the tier above and the tools bolted on beside. The person who will answer for the system signs the decision.