Custom CRM vs Salesforce: which one is the cheaper mistake

Both choices are mistakes for somebody. Salesforce is a mistake for a forty-person distributor whose quoting rules will never fit a standard object model. A custom CRM is a mistake for a six-person consultancy that needs a pipeline and nothing else. The useful question is not which tool is better — it is which mistake your company can least afford.

We build custom CRM systems for a living, so read the rest with that in mind. We also talk clients out of it several times a year, which is the only reason this article is worth anything.

The five-year maths, done properly

Most comparisons stop at the per-seat price. That is the smallest number in the sum. A realistic five-year total for a subscription CRM at forty users looks like this: licences at $80 per user per month is $192,000; implementation partner, $15,000 to $40,000; paid add-ons for the things the core product does not do, $10,000 to $30,000; an internal administrator at part-time, which nobody counts but everyone pays.

The custom side: a $45,000 build, $1,200 a month of managed support for five years at $72,000, and hosting at roughly $150 a month. Call it $126,000, with an asset on your balance sheet at the end instead of a renewal date.

Those numbers cross over somewhere between eight and twelve users for most businesses. Below the crossover, the subscription wins clearly and we will tell you so. Above it, the argument stops being about money and starts being about fit.

When Salesforce or HubSpot is plainly right

Buy the subscription when your sales process is genuinely conventional — leads, opportunities, stages, close — and you have no rules that a configuration screen cannot express. Buy it when you need to be running next month rather than next quarter. Buy it when your team is under ten people and likely to stay there. Buy it when you have no one internally who can own a system, and no appetite to pay someone to.

And buy it when the integrations you need are on the vendor's marketplace already. That is a real advantage and an expensive one to reproduce.

When custom wins

Build when your process is the thing that makes you money. A distributor whose pricing depends on contract tier, branch, volume band and a manager's override is not going to fit a standard object model — and the workarounds that make it fit become the system you are actually running, without the documentation.

• Your quoting or pricing logic is genuinely unusual, and getting it wrong costs real money.
• You need to integrate with something local — a Pakistani or Gulf accounting system, a regional payment gateway, a government portal — that no marketplace covers.
• Per-seat licensing punishes you for growing, because you want operations, finance and field staff in the system too.
• You have data-residency obligations, or a client contract that says where the data lives.
• You have tried configuring a platform and the configuration has become its own maintenance burden.

The cost nobody prices in: configuration debt

A heavily configured subscription CRM is custom software. It is custom software written in a proprietary configuration language, by consultants, without tests, that you cannot move. Every workflow rule, every formula field, every managed package is a thing that will break when the vendor ships a release, and a thing you must pay someone who knows that platform to fix.

This is the failure mode we get called about most. Not "Salesforce was too expensive" — "Salesforce was configured by three different partners over six years and nobody can tell us what it does any more." Recovering from that costs more than either original choice.

The test is simple. If the platform will do what you want out of the box, buy it. If you are going to configure your way to a bespoke system anyway, you are choosing between custom software you own and custom software you rent.

The hybrid that usually works

You do not have to pick one for everything. Several of our clients run a subscription tool for marketing — where the ecosystem genuinely is better — and a custom system for the part that is theirs: quoting, contracts, field operations, service history. The two talk over an API. It costs a little more to maintain and avoids forcing the whole business through one vendor's idea of a sales process.

What moving off a subscription actually involves

The export is trivial and everyone worries about it. The real work is archaeology. Over six years somebody built workflow rules, validation rules, formula fields, approval processes and probably a trigger or two, and the documentation is in the heads of consultants who have moved on.

Before you can rebuild a process you have to know what the current one does — including the automations that quietly fire on save, and the report finance trusts without knowing where its numbers come from. We budget four to twelve weeks on top of the build for this, and most of it is interviews and tracing, not coding.

Two practical notes. Keep the old system read-only for a quarter rather than cancelling on cutover day; the licence cost is trivial next to needing a record you no longer have. And migrate open work plus current master data, not every closed opportunity since 2017 — history you never open is history you are paying to move.

The question of who maintains it

This is the fair objection to building, and it deserves a straight answer. A subscription CRM is maintained by the vendor: security patches, uptime, browser compatibility, a support line. A custom system is maintained by whoever you pay to maintain it, and if you pay nobody, it rots. Not immediately — it works fine for a year, and then a dependency has a vulnerability and there is no one whose job it is to notice.

So price it in from the start. Managed support with an agreed response time runs from $1,200 a month and covers patching, monitoring, backups and small changes. If that number makes the five-year comparison stop working, the subscription was the right answer and you have just saved yourself a project.

How to decide in a week

Write down the five things your sales team does that a competitor does not. Then book a demo and ask the vendor to show you those five, on their standard product, with no code. What they cannot show you is your build scope. If the list is empty, buy the subscription today and spend the saved money on people.

If the list is long enough to worry you, a two-week discovery sprint will price the alternative properly — and you keep the backlog and prototype whichever way you go.