What It Costs to Replace Salesforce With a Custom-Built Tool

Most teams pay for a hundred Salesforce features to use six. The pipeline view, a couple of custom fields, a report, an email log, a task list, maybe a forecast — that is the working surface for the average sales desk. Everything else is paid-for capability that sits idle, plus the per-seat fee that climbs with every hire and every renewal. The question that brings people to us is rarely "is Salesforce good." It is "why does the thing we barely touch cost more each year than it did last year."

This note lays out the honest economics of when to replace Salesforce with a custom tool, what a build actually costs, and — just as important — when you should keep the platform you have. We sell custom software, so treat the recommendation to stay put as the more telling one.

The per-seat meter is the real cost, not the sticker price

Salesforce Sales Cloud lists at $175 per user per month for Enterprise and $350 per user per month for Unlimited, billed annually, with most editions requiring a sales conversation rather than a public price (tech.co, 2026). In August 2025 Salesforce applied roughly a 6% increase across Enterprise and Unlimited editions. That last detail matters more than the headline number: the price is not fixed, and you do not control when it moves.

Run the meter. A twelve-person sales team on Enterprise is about $25,200 a year before a single add-on, integration, or consultant invoice. Twenty-five people pushes past $52,000. None of that buys you an asset — it rents access for another twelve months, after which the rate resets upward and the seat count, ideally, has grown. The Salesforce replacement cost conversation usually starts the moment a renewal quote lands with a number nobody recognised agreeing to.

The hidden layer sits underneath. Salesforce Ben's survey of architects found an average Salesforce budget of over $700,000 per year among respondents, with the platform demanding constant hands-on technical work to keep customisations alive. That figure skews enterprise, but the shape holds at every size: the licence is the part you see, and the admin, the integration glue, and the specialist who understands your org are the parts you feel.

What a custom replacement actually costs to build

Here is the comparison that reframes Salesforce per-seat cost vs custom. A focused CRM — your pipeline stages, your fields, your reports, the two or three integrations you genuinely rely on — is a defined piece of software, not an open-ended platform. As a working range, a build like that lands between £15,000 and £60,000 as a one-off, with the spread driven by integration count, data-migration complexity, and how much automation you want on day one. That is an estimate from the shape of the work, not a quote; the only honest number comes after we have seen your actual workflows.

The structural difference is ownership. After the build, there is no per-seat line. Adding your thirteenth or thirtieth user costs nothing in licensing — the marginal cost of a seat on software you own is zero. Hosting and maintenance are real and ongoing, typically a modest monthly figure for infrastructure plus a support arrangement, but they do not scale with headcount and they do not reset upward on a vendor's schedule.

So the break-even is simple arithmetic. If you are spending £25,000–£50,000 a year on Salesforce, a £15k–£60k build that covers your real workflows pays for itself inside one to two years and then keeps paying, because the recurring cost afterwards is a fraction of the annual licence. The cost to replace Salesforce with a custom build is front-loaded; the licence is a tax that never stops. Past a certain spend, owning is cheaper than renting — and you set the roadmap.

Why this calculus only recently flipped

For two decades, "build your own CRM" was a warning, not a plan. Custom software took three to six months, tied up a team, and aged badly. Buying was almost always right. What changed is that one side of the equation collapsed while the other held still. As one engineering write-up puts it, custom development costs fell sharply with AI-assisted tooling while SaaS prices stayed flat, reversing the historical logic of buy-over-build (paddo.dev, 2026).

The market noticed. The same analysis tracks a heavy repricing of SaaS through early 2026, with the sector's forward earnings multiple compressing as buyers reconsidered renewals. You do not need to believe every market number to take the signal: the cost of building a focused tool has dropped enough that the build vs buy hidden costs equation now tips toward building for narrow, well-understood workflows far more often than it used to.

The build revolt is already mainstream

This is not a fringe move. Retool's 2026 build-vs-buy report, surveying 817 builders, found that 35% of teams have already replaced the functionality of at least one SaaS tool with something custom, 78% plan to build more in 2026, and 51% have shipped production software currently in use. Among teams running such tools, about half reported saving six or more hours per week. The pattern of mid-market companies replacing SaaS with their own tools is the headline finding, not an anecdote.

The named examples are concrete. ClickUp built six internal AI tools that, by its own account, cut $200,000 a year in automation software. The startup-data platform Harmonic replaced a $20,000-per-year third-party tool after deciding it was faster to rebuild it than wait on support. These are not heroic engineering feats — they are teams declining to rent narrow capability they could own.

The Klarna story — and why we'd quote it carefully

You will hear that Klarna replaced Salesforce with a custom AI system, and it is half true in a way worth being precise about. Klarna's CEO told investors the company "shut down Salesforce," part of a wider retreat from around 1,200 SaaS platforms ahead of its IPO. But Klarna did not pour its CRM into a language model. It built an internal stack — using a graph database to unify its data — and stitched together other tools to cover what Salesforce had done.

The honest coda: Klarna's CEO later said he was "tremendously embarrassed" by parts of the fallout and was clear he does not think Salesforce is finished or that most companies should follow Klarna's exact path. We mention this because the cautionary half of the story is the useful half. Replacing a platform is not a press release; it is a migration, a data model, and an ownership decision. Klarna had the scale and the engineers to absorb the risk. The lesson for a mid-market team is narrower and safer: replace the workflows you actually use, not the entire idea of having a system.

When to build a custom CRM instead of Salesforce — and when not to

The case to build a custom CRM instead of Salesforce is strongest when three things are true. Your real workflow is narrow and stable — you use a small, well-understood slice of the platform. Your seat count is growing, so the per-seat meter is working against you. And your processes are specific enough that you are forever bending Salesforce to fit, or paying a consultant to. In that situation, a tool that does only your six things, that you own outright, with a custom CRM to replace Salesforce price that is a one-off rather than a renewal, is usually the better long-term call.

The case to stay is just as real, and we will say so plainly. If you depend on Salesforce's deep ecosystem — AppExchange integrations, a mature partner network, compliance certifications you would otherwise have to recreate — the platform is earning its fee. If your processes change constantly, a vendor absorbing that churn may be worth the rent. And if your team genuinely uses the breadth of the product, you are not paying for a hundred features to use six; you are paying for the hundred. In those cases, building is a way to spend money proving a point.

A measured middle path exists, too. You do not have to replace everything at once. Lift the one workflow that costs you the most friction — the speed-to-lead routing, the bespoke quoting flow, the reporting Salesforce makes painful — into a tool you own, and let it run alongside the platform. Often that single move recovers enough cost and time to fund the next one, and you learn what a full replacement would really involve before committing to it.

How we'd cost it for you

The right number is not a range from a blog — it is the output of looking at your actual setup: which Salesforce features your team touches in a normal week, your seat count and renewal trajectory, your integrations, and how much data needs to move. From that we can tell you the build cost, the recurring cost after, and the year your ownership beats your licence. If that crossover sits comfortably in the future, we will tell you to keep Salesforce. If it has already passed, you have been renting an asset you could have owned.