Custom CRM vs HubSpot: when building your own pays off

If your sales process is a standard pipeline and most of what you need is email, forms and marketing automation, stay on HubSpot. A custom CRM pays off when the process itself is what you sell, when people outside your team need to log in, or when the integrations are the job. On licence maths alone, building rarely wins below a few dozen seats. The workarounds are where the real cost hides.

Desses sells custom builds, so every argument for building below moves money toward me. I've written the case for staying as hard as the case for leaving, and I'll make the same call on a first call.

How much does HubSpot cost over three years?

For 25 sales users on Sales Hub Professional, about $82,500 over three years before add-ons, by HubSpot's own published prices.

I checked HubSpot's Sales Hub pricing page on 10 October 2026. Professional "starts at" $90 per seat a month with annual billing, shown against a struck-through $100, so read $90 as the floor, possibly a promotion. Enterprise starts at $150 per seat a month. Both carry a required one-time onboarding fee: $1,500 for Professional, $3,500 for Enterprise. Starter starts at $7 per seat (struck through from $20), and the free plan covers up to 2 users.

Seats × list price × 36 months, plus onboarding. At the struck-through $100, Professional runs about 10% higher.

That table is the floor. It leaves out HubSpot Credits beyond the included allowance ($0.010 each), other Hubs, and the partner you'll likely pay to configure it. View-only seats are free, which helps if half the company only needs to look.

What does a custom CRM cost to build?

At Desses, a first working module starts at $8,000 (€7.000) and usually lands in three to four weeks. Most teams then keep building on a monthly retainer from $4,500 (€4.000).

What you pay for depends on five things, and none of them is the number of users:

• The data model. How a lead becomes a customer, a customer becomes a project and a project becomes an invoice. Get this wrong and every later module pays for it.

• Roles and permissions. Sales, finance, clients, partners, installers. Each sees their own slice, enforced on the server, never only hidden in the UI.

• Integrations. Each system you connect (accounting, email, calendar, an existing CRM) is its own small project, with its own failure cases.

• Migration. Moving years of records out of the old tool, deduplicated, with owners and history intact. Messy data is the line item people underestimate most.

• Maintenance. Dependencies age, APIs change, the process changes. Software you own needs someone looking after it.

The running costs for the stack I build on are per developer, not per user. Today's list prices: Vercel Pro $20 a month per developer seat, Convex Professional $25 per developer a month, Clerk Pro $25 a month with 50,000 monthly retained users included. All three bill usage on top, so a busy system costs more. Your 25th salesperson costs nothing extra.

Is a custom CRM cheaper than HubSpot?

Usually not in year one, sometimes by year three, and only at higher seat counts.

Note on hosting: the custom columns add the per-developer plans above plus usage. For a 25-user internal tool the usage part is usually small, but it depends on data volume, so I quote it per project rather than print one number.

At 10 seats, HubSpot Professional wins on every column. At 25 seats, a custom build that stops after a year of active development is cheaper on paper. A build you keep developing for three years costs about double HubSpot, and the only reason to do that is that it's doing work HubSpot can't.

So I wouldn't build on licence savings alone. Build when the CRM you'd buy needs three workarounds and a spreadsheet to match how you sell, because those workarounds have a cost too: people retyping data, quotes built outside the system, reports nobody trusts.

When does building your own CRM pay off?

When the CRM stops being a contact list and starts being the way the business runs. Four signals, and two are usually enough:

Your process is the product. Quotes, installations, approvals or onboarding follow steps a standard pipeline only models with custom fields everywhere. If your team explains the workaround to every new hire, the tool doesn't fit.

Your customers need to log in. Clients want to see their orders, documents or status. Partners need their own view. A portal on the same data as your team removes the "can you send me the latest version" email.

The integrations are the job. Accounting, email and calendar need to stay in sync with sales, and right now someone copies data between tabs every Friday.

Per-seat costs scale badly for you. Large field or operations teams who need light write access get expensive on a per-seat model. On a custom build, adding a user is a row in a table.

When should you stay on HubSpot?

When your pipeline looks like everyone else's pipeline. Leads come in, get qualified, move through five or six stages, close. HubSpot has modelled that for years and will do it better than anything you build in a quarter.

Stay too when marketing automation is the main use. Email sequences, lead scoring, landing pages, forms and attribution are HubSpot's home ground, and rebuilding them is money spent catching up. And stay if nobody on your side can own a piece of software. A custom CRM without an owner decays.

Can you keep HubSpot and build the rest?

Yes, and it's what most teams should look at first. Keep HubSpot for marketing, build a custom portal or operations layer for the part it can't model, and sync the two.

Marketing Hub Professional lists at $800 a month on annual billing, with 3 core seats, 2,000 marketing contacts and a $3,000 onboarding fee, according to HubSpot's own pricing post updated 14 September 2026 (the pricing page itself wouldn't load for me on 10 October, so check it before you budget). That's about $31,800 over three years for the marketing side, and sales and operations can live in the custom layer.

Same logic I apply to websites: the marketing site and the operational app are different jobs, and they don't have to share a tool. I wrote about that split for sites here.

What does a custom CRM look like in practice?

ChargeBlock, a Dutch battery storage company, runs its sales-to-service cycle on a platform I designed and built in Next.js, Convex and Clerk. A custom CRM sits at the core, with 22+ pages on the business side.

A 4-step quote wizard generates branded PDFs the moment a quote is built. Customers sign digitally on a PIN-secured public page. Invoicing runs automatically on a 40/40/10/10 milestone split tied to project progress. Clients, partners and installers each log in to a portal with role-based access, so each side sees what they need and nothing else. Project management runs on kanban in the same platform.

It's hybrid too. The platform integrates with Outlook, Exact Online and Pipedrive, so the existing toolset stays in play. Nothing about that setup is a standard pipeline, which is why building made sense there.

What are the risks of a custom CRM?

The real ones are ownership, scope and continuity.

Ownership: you need someone who decides what the system does next. Scope: a CRM is never finished, so build one module, prove it earns its keep, then build the next. Continuity is the one to ask any small studio about directly, me included. Desses is one person. The answer I give is that code, data and documentation live in your accounts, on a mainstream stack any Next.js developer can pick up. If a builder can't give you that answer, don't sign.

More on how I scope and price this on the custom CRM development page and the broader AI and custom development work.

Who this is for, and who should not hire me

This is for B2B teams whose process has outgrown a standard pipeline, who want design and code from the same person. It's not for teams whose process fits HubSpot with a few custom fields, teams looking for a HubSpot implementation partner, or projects under €7.000.

If you want a straight read on build, buy or hybrid, book an intro call. Sometimes the answer is to keep HubSpot.