Custom CRM versus Salesforce or HubSpot
The honest trigger is rarely features. It is seat cost at scale, or a bought CRM your team has already stopped updating.
Custom build
Your process, no per-seat licence
Licensed platform
Vast ecosystem, immediate start
Most companies should license a CRM. The platforms are mature, the ecosystems are enormous, and the monthly cost is genuinely reasonable at small and mid headcount. Building one to save money at twenty seats is a bad trade.
There are real cases where building wins, and they are more specific than 'we need customisation'. The most common one is not about features at all: the licensed CRM was bought, configured, and quietly abandoned by the sales team, so the pipeline report is now fiction.
Where they actually differ.
| Dimension | Custom build | Licensed platform |
|---|---|---|
| Cost at 20 seats | Build cost dominates. Almost never worth it at this size. | Modest monthly cost, immediate start. Clearly the better answer. |
| Cost at 200+ seats | One-time build, then maintenance. Often materially cheaper over five years. | Per-seat licence becomes a significant recurring line, and rises with tier upgrades. |
| Process fit | Your stages, your qualification criteria, your approval thresholds, and nothing else. | Highly configurable, but the object model has opinions you will eventually fight. |
| Adoption | Designed around what reps already do; every field earns its place or is removed. | Strong if your process resembles the default. Adoption fails when it does not. |
| Integrations | Built as needed, properly, with reconciliation. Each one is work. | A marketplace with thousands of prebuilt connectors. A real and large advantage. |
| Reporting | Governed metric layer, one definition each, reads from your warehouse. | Powerful out of the box; heavy customisation makes reports increasingly fragile. |
| Time to live | 10–16 weeks for a first usable version. | Days to start, weeks to configure properly. |
| Customisation upkeep | It is your codebase; you upgrade on your schedule. | Heavy customisation carries an ongoing retest cost at each platform release. |
- Large seat count where per-seat licensing has become one of your bigger software line items.
- You have already failed at adoption on a licensed platform, and the diagnosis is process mismatch rather than training.
- Your sales motion is genuinely unusual — complex quoting, unusual territory or commission logic, a multi-party approval chain.
- The CRM needs to be part of a larger custom system, sharing a data model with your ERP or operations platform.
- Data residency or sector rules rule out the platforms available to you.
- Fewer than roughly fifty seats. The economics are not close.
- You need to be running this quarter, not next.
- You depend on the integration marketplace — dialers, enrichment, sequencing, marketing automation.
- Your process is conventional, or you are willing to adopt a good default rather than encode current habits.
- You have no engineering capacity, and no intention of retaining any.
Our view
Below about fifty seats, license one. Above a couple of hundred, model it properly over five years including maintenance, because the answer often flips. And if a licensed CRM has already failed on adoption, fix the process question first — building a custom version of a process nobody follows just produces a more expensive empty database.
Follow-ups.
Can you migrate our existing CRM data?
Yes, and the cleansing is the slow part rather than the moving. Deduplicating accounts and contacts, reconciling owner history and closing out stale opportunities takes longer than the migration itself. We run it in parallel and validate before cutover.
How do you avoid the adoption failure that killed the last one?
By interviewing reps before designing anything and cutting required fields to those that drive a decision. Adoption is a design outcome. If someone must enter eleven fields to log a call, they will not log calls, and no amount of management pressure changes that.
Could we build only the parts that do not fit?
Frequently the best answer. Keep the licensed CRM as the contact and activity system of record, and build the quoting engine or commission calculation alongside it. Cheaper than either extreme and lower risk.
Tell us what you are building.
A short conversation is usually enough to tell whether we are the right firm for the problem. If we are not, we will say so and point you somewhere better.