CRM · 6 MIN READ · APR 2026 · Layla Saeed

Build vs. buy: when a custom CRM actually pays back.

The seven questions we ask before recommending a custom CRM build over Salesforce or HubSpot. If five answers are yes, build. If not, fix what you have.

"Should we build a custom CRM?" is one of the most common questions on our first calls. Our default answer is "no" — until it isn't. Here's the framework we use.

Salesforce and HubSpot are extraordinary products

Let's start there. Both have invested billions in features you genuinely need: integrations, mobile apps, permissions, audit trails, security certifications, AI features that are now better than most custom builds. For 80% of teams, one of them is the right answer.

Custom only beats them when one specific thing is true.

The seven questions

We ask these seven questions before recommending a custom build:

1. Is your sales motion structurally different?

Real estate, automotive, healthcare and certain marketplaces have data models that don't fit standard CRMs. If your "lead" is actually three records over four channels and the dedup is the work, custom is on the table.

2. Are you spending more than $200/seat/month on the current CRM?

At a certain seat-count and feature tier, a custom build pays for itself in 18 months on licenses alone.

3. How much customisation are you already paying for?

A Salesforce admin team plus a third-party agency at a few hundred grand per year is often roughly the cost of a custom build amortised over four years.

4. Do you have engineering capacity for ongoing maintenance?

This is the real question. Salesforce never goes dark. A custom CRM does, if you don't fund it. Have you got a 0.5-FTE engineer, internal or external, dedicated to it forever?

5. Is your data model going to keep evolving?

Custom shines when the data model is part of the strategy and changes often. It's a liability when the data model is set and you just need stability.

6. Do you have privacy/sovereignty needs that vendors don't meet?

Some industries (healthcare, government, defence) require data residency or audit guarantees that the major vendors don't always offer in your region.

7. Have you tried fixing the existing CRM properly?

Honestly. Most "Salesforce is bad" verdicts are actually "our Salesforce admin left two years ago and nobody's owned it since." Try a six-week tune-up first.

If five out of seven are 'yes,' build

If you've answered yes to five of these, custom is probably the better economic and operational choice. Below five, fix what you have. Two-of-seven is almost always 'rescue your existing CRM.'

The best custom CRM is the one you didn't have to build.

What 'good custom CRM' looks like

If you do build, the targets we hold ourselves to:

If you'd struggle to hit those, the CRM you'll build will probably be worse than HubSpot. Which means you should buy HubSpot.

LS
Layla Saeed
Founder · Startup13

More from the blog.

Other recent field notes.

why·
AI · 8 MIN

Why most AI integrations fail in production.

Four engineering patterns that separate the demos from the systems.

Read →
the·
AUTOMATION · 5 MIN

The 80/20 of business automation — start here.

Where to begin before adding AI to the mix.

Read →

Want this for your team?

Most of what we write about is something we've already shipped for clients. Tell us your problem and we'll walk you through how we'd approach it.