⚡The verdict up front
This article does one thing our broader build-vs-buy piece deliberately does not: it puts the five-year math on the table. The decision framework — process fit, speed to value, strategic differentiation — lives in the companion article. Here we answer the narrower question a CFO actually asks: what does each path cost, all-in, over five years, and where is the break-even?
The commitment: Salesforce is priced per seat per month, which means its cost scales with your headcount, while a custom CRM is priced mostly as a fixed build plus a maintenance line, which means its cost scales with complexity, not headcount. That single structural fact drives everything below. The more seats you add, the worse Salesforce’s math gets relative to custom; the faster you need to be live, the better Salesforce looks regardless of the math.
So the verdict has two clauses. If you are under roughly 25 users, or you need a working CRM this quarter, buy Salesforce and spend your energy on adoption. If you are between roughly 40 and 100 users with workflows that bend standard CRM objects out of shape, model a custom build seriously — the five-year saving is commonly six figures, and it grows every year you stay.
| Team size | Likely 5-year winner on cost | Why |
|---|---|---|
| Under 25 users | Salesforce | Custom build cannot amortize below this scale |
| 25–40 users | Model both | Crossover zone; depends on workflow complexity |
| 40–100 users | Custom CRM (usually) | Per-seat license and admin costs compound past the fixed build cost |
| 100+ users, standard processes | Salesforce (on speed/ecosystem, if not pure dollars) | Ecosystem, hiring pool and certifications justify premium |
| 100+ users, non-standard workflows | Custom CRM | You would be paying per seat AND paying consultants to fight the data model |
🧾What Salesforce actually costs: the full anatomy
The license is the number everyone quotes and rarely half the cost. As of writing, Sales Cloud list prices run from a starter tier around $25 per user per month up through Professional and Enterprise tiers into the $165-plus band, with Unlimited tiers above $300, following increases announced in 2025 — verify current list prices, because Salesforce revises them regularly and enterprise negotiated discounts of 10 to 30 percent off list are common but never guaranteed. AI features are increasingly packaged as add-ons or consumption-based credits on top of the seat price, which is a real and growing line item.
Then the operating costs. A Salesforce org does not run itself: a certified administrator in the US market commands roughly $90,000 to $140,000 a year, and an org past fifty users typically needs at least one, with part-time or fractional coverage possible below that. Implementation and ongoing consultant work at labelled market rates of $150 to $300 per hour adds a setup project — commonly $15,000 to $75,000 for a mid-market rollout — plus a recurring stream of change requests, because every process change is billable configuration work.
Finally, the ecosystem tax. The AppExchange is genuinely useful and genuinely metered: most serious apps charge per user per month, typically in the $5 to $100 range, and a mature org accumulates several. Add sandbox environments, extra storage, premium support tiers and integration tooling, and a common pattern emerges from real budgets: the all-in annual cost of a Salesforce org lands at roughly 1.5 to 2.5 times the license line alone. Model with that multiplier and you will not be ambushed.
Ask any Salesforce quote one question: "what is our all-in annual cost at our three-year headcount — licenses, admin, consultants, apps — not year one?" The year-one number is the marketing number.
🏗️What a custom CRM actually costs: the honest anatomy
A custom CRM has two costs, and the first is the one everyone focuses on: the build. In labelled market ranges, a focused CRM for one team with standard objects — contacts, companies, deals, activities, reporting — runs $60,000 to $150,000. A multi-role platform with quoting, approval workflows, customer portals and deep integrations runs $150,000 to $400,000 and up. That range is wide because scope is wide; any narrower pre-discovery number is a guess wearing a suit.
The second cost is the one vendors of custom software undersell: ownership. Maintenance — dependency updates, security patches, small fixes — runs 15 to 25 percent of the build cost per year. Feature evolution is on top of that and is the part you control. And someone must own the system: not necessarily a full-time hire, but a named engineer or a retained partner who answers when it breaks. A custom CRM with no owner is not cheaper than Salesforce; it is a slow-motion outage.
What custom does not have: per-seat fees, per-app fees, consultant bills for configuration changes, and price increases at renewal. Your fiftieth user costs nothing in licensing. Your hundredth workflow change costs engineering time at a rate you chose. This is why the five-year math tilts: Salesforce costs compound with headcount, custom costs compound only with ambition.
⚖️Break-even modeling by team size
Here is the crossover logic, stated as a model rather than a fake-precise calculator. Salesforce’s five-year cost is approximately linear in seats: licenses plus apps scale per user, admin scales in steps (roughly one admin per fifty to a hundred users), and consultants scale with how much you change. A custom CRM’s five-year cost is approximately fixed: build, plus 15 to 25 percent of build per year, plus whatever feature work you choose. A line and a flat-ish curve cross exactly once, and where they cross is the seat count you are looking for.
Run the model with mid-range inputs: Salesforce at an Enterprise-tier seat price with the 1.5 to 2.5 times all-in multiplier versus a $120,000 custom build at 20 percent annual maintenance. The lines cross somewhere in the 30 to 50 seat band. Raise the custom build to $250,000 because your workflows are genuinely complex, and the crossover pushes out toward 60 to 80 seats. Drop Salesforce to a lower tier with minimal consulting, and it pulls back toward 25 to 35. The band, not the point, is the honest answer — but the band is narrow enough to make a decision with.
Two factors move the crossover more than any price assumption. The first is workflow fit: if your process forces heavy Salesforce customization, you pay per seat and pay consultants to fight the data model, which drags the crossover down toward 20 seats. The second is time: if being live in eight weeks instead of eight months is worth real revenue, that value belongs in the model too, and it pushes the crossover up. Cost is only half of TCO if the cheaper option ships a year later.
| Cost line (5-year view) | Salesforce path | Custom build path |
|---|---|---|
| Licenses / build | Scales per seat; $25–$330+ per user/month list tiers | Mostly fixed; $60K–$400K by scope |
| Administration | $90K–$140K/yr per admin, stepped by seats | Named owner or retained partner; fraction of a salary at small scale |
| Consultants / changes | $150–$300/hr; every process change is billable | Engineering time at your chosen rate |
| Apps / extensions | $5–$100 per user/month per app | Built into the codebase or not at all |
| Maintenance | Included in subscription (platform level) | 15–25% of build cost per year |
| Price increases | Renewal risk; list prices revised periodically | None — you own the asset |
📊The honest 5-year TCO table
The table below models three team sizes with mid-range assumptions, and it is a model — every line is a labelled range, not a quote, and your numbers will differ. Its purpose is to show the shape of the comparison: how the Salesforce column grows with seats while the custom column barely moves, and how the ratio flips between the first and third rows.
Read it directionally. At 10 users, custom is not even competitive — you would spend build money to recreate what a starter subscription gives you this month. At 50 users, custom is typically 30 to 50 percent cheaper over five years and the gap widens every renewal. At 200 users the dollar gap becomes enormous, but the decision gets harder rather than easier, because at that scale the ecosystem factors in the next section — hiring, compliance, integrations — carry board-level weight that a spreadsheet cannot dismiss.
If you take one action from this article, take this: build this table with your real seat count, your real workflow complexity and a negotiated Salesforce quote rather than list prices. An afternoon of arithmetic has saved more companies more money than any framework debate in this industry.
| 5-year TCO (labelled ranges, mid assumptions) | Salesforce path | Custom build path |
|---|---|---|
| 10 users | $150K–$300K (licenses, fractional admin, light consulting) | $110K–$250K (build + maintenance) — rarely justified on cost or speed |
| 50 users | $600K–$1.1M (licenses, one admin, consulting stream, apps) | $150K–$450K (build + maintenance + modest evolution) |
| 200 users | $2M–$4M+ (licenses dominate; admin team; ongoing consulting) | $300K–$900K (larger build, dedicated ownership, evolution) |
| Winner on pure dollars | Under ~25 seats | Above ~40 seats, growing yearly |
💪What Salesforce genuinely does better
Honesty requires this section, because the TCO math above could be misread as "custom always wins past forty seats." It does not. Salesforce ships in weeks what a custom build ships in months, and for a company standing up its first real sales process, that speed is worth more than any five-year saving — you cannot compound a pipeline you do not have. Time-to-value is Salesforce’s strongest card and the one custom cannot match.
The second strength is the ecosystem as labor market. Salesforce administrators, developers and consultants are a global, certified, hireable pool; when yours leaves, you hire another. A custom CRM concentrates knowledge in the people who built it, and while good documentation and clean code mitigate that, the bench depth is not comparable. Related: the AppExchange means your eleventh-hour requirement — e-signatures, territory management, commission tracking — is usually a purchase, not a project.
The third strength is inherited trust. Salesforce carries compliance certifications, security reviews, uptime track record and procurement familiarity that enterprise customers and auditors already accept. A custom system earns that trust from zero with every customer, auditor and security questionnaire it meets. For companies selling to regulated enterprises, that inherited credibility is a real, monetizable asset that belongs on the Salesforce side of the ledger.
🧭Decision rules you can act on this quarter
If you have no CRM discipline today and under 25 users, buy Salesforce (or a lighter CRM) and do not look back — your problem is adoption, not software economics. If you are between 40 and 100 users, your workflows have outgrown standard deal objects, and you are paying consultants to hold a customized org together, get a scoped quote for a custom build and run the five-year table with real numbers; the answer is usually obvious within one afternoon.
If you are above 100 users with genuinely standard processes, Salesforce’s premium is often the rational price of ecosystem, speed and hiring depth — negotiate hard on licenses instead of rebuilding. If you are above 100 users with non-standard workflows — the CRM is really an operations platform wearing a sales hat — the dollar case for custom becomes overwhelming, and the question shifts from whether to build to how to migrate without breaking the quarter.
One final rule that overrides the others: whichever path you choose, model the five-year number with someone who has built both sides of this comparison and has no license resale incentive. The worst CRM decisions we see are not wrong answers — they are answers given by a party who only sells one of the options.