A company decides it needs a CRM. The first meeting is about tools: HubSpot vs Salesforce vs Dynamics vs Zoho vs Pipedrive. Demos get booked, comparison sheets get built, and someone says "Salesforce is the industry standard, let's just go with that."

Six months later, the CRM is live and nobody really uses it. Reps keep their real pipeline in Excel. Managers still ask for reports by email. The company blames the software and starts looking at a different one.

In most of these cases the tool wasn't the problem. Nobody had agreed on what the CRM was supposed to do.

Before I talk about software with anyone, I ask questions like these:

1. How do leads actually reach you today? Web form, email, phone, events, referrals? Where do they land first?
2. What happens after a lead is qualified? Who takes it, and by when? What does "qualified" actually mean?
3. How do you track opportunities? What are your real stages, not the ones from a template?
4. What do managers need to see every Monday morning? If you can't name the reports, you can't design the CRM.
5. What eats the most manual time? That's usually where automation pays off first.
6. Who owns the customer data? Who can edit it, and who's responsible when it's wrong?

Once you have answers, the decision gets much clearer, and it's not always "pick a big brand":

• Off-the-shelf works great when your process is fairly standard. You get a mature product, integrations, and a community. The trade-off is that you'll adapt your process to the tool, and per-seat licensing adds up as you grow.
• Heavily customized off-the-shelf is where a lot of projects get expensive. If you're paying for endless workarounds, custom fields, and add-ons just to force the tool to fit, you're often halfway to a custom build without the benefits.
• A custom CRM makes sense when your sales process is genuinely unusual, when the CRM has to plug deeply into your own product or internal systems, or when you'd otherwise pay for a hundred features to use ten. The trade-off is that you own the maintenance, and it's a bad idea if your requirements are still vague.

That last point is the catch. Custom is the most requirement-dependent option of all. If you can't describe your process clearly, you'll build the wrong thing, just with more code.

So the order matters: process first, requirements second, tool (or custom build) third. I always start with a discovery phase and only then decide whether an existing platform or a custom build fits.

Curious what others have experienced. If you've been through a CRM rollout, did you define requirements first or pick the tool and figure it out later? Has anyone gone custom and regretted it, or been glad they did?

Happy to answer questions in the comments if you're in the middle of an evaluation.