CRM Implementation Cost: Salesforce vs Custom (2026)

⚡The honest answer up front

CRM implementation cost is really three separate numbers that vendors prefer to quote one at a time: the licences, the implementation work, and the migration. A buyer who budgets for the first and discovers the other two in month two is the norm, not the exception. So here are all three, as labelled US market ranges we observe, for a company with 20 to 60 CRM users.

On the Salesforce path, expect licences at $25 to $330 per user per month depending on edition, implementation partner fees of $10,000 to $75,000 for a properly scoped rollout, and data migration at $8,000 to $40,000 depending on how messy the source data is. On the custom path, expect $45,000 to $150,000 to build the system, $5,000 to $20,000 for the same migration work, and then only hosting and maintenance — typically 15 to 25 percent of the build cost per year — with no per-seat fee at any headcount.

The ranges are wide because scope variance is real. A Salesforce rollout with two integrations and clean data lands at the low end; one with a CPQ setup, a marketing automation sync and a decade of spreadsheet history lands at the high end. Treat any single number quoted before a scoping conversation as a guess.

CRM development services — build and implementation→

💳Salesforce licence tiers, priced honestly

Salesforce pricing is public, and the trap is not the list price — it is which edition your requirements silently force you into. Starter Suite at roughly $25 per user per month is a genuine product, but it caps automation, API access and customisation in ways a growing team hits within months. Most serious sales organisations land on Professional (around $80), Enterprise (around $165) or Unlimited (around $330) per user per month.

The forcing functions are specific. API access for integrations — the thing you need to connect your CRM to your product, your billing system or your data warehouse — is effectively an Enterprise-edition feature. Advanced automation, fuller sandbox environments and heavier customisation limits push the same direction. This is why "Salesforce costs $80 a seat" is true on a demo call and false on the invoice.

Add-on clouds multiply the bill: CPQ, Marketing Cloud, Service Cloud and Data Cloud are each priced separately, and a quote that starts as "Sales Cloud, 25 seats" can arrive as four line items. None of this is hidden — it is all on the price sheet — but it means licence cost is a function of your requirements, not your seat count alone.

🤝What implementation partners actually charge

Almost nobody self-implements Salesforce past the Starter tier, and Salesforce itself will introduce you to its partner ecosystem. Implementation partners bill in one of three shapes: fixed-price for a bounded rollout ($10,000 to $40,000 for a clean SMB setup), time-and-materials at $150 to $300 per hour for evolving scopes, or a managed-package engagement for larger rollouts that runs $50,000 to $150,000 and up.

The partner market is stratified. Boutique consultancies with ten certified people quote lower and staff senior. Large global integrators quote higher and often staff a blended team where the people who sold the work are not the people configuring your flows. Neither is automatically wrong — but ask who will actually do the work, their certifications, and how many of them are on your account.

A scope pattern worth knowing: partners earn margin on configuration hours, so the incentive gradient points toward more configuration. A good partner tells you which of your requests should be handled by an out-of-the-box feature you are already paying for. A bad one builds everything custom inside Salesforce, and you pay for the build and then pay again to maintain it. The interview question that separates them is simple: "What are you going to talk me out of?"

📈The costs that arrive after go-live

Go-live is the midpoint of the spend, not the end of it. A Salesforce org of any real complexity needs an administrator — part-time at the low end, a full-time hire at $70,000 to $120,000 a year in US salary once you pass roughly 30 to 50 users or run multiple clouds. Some companies outsource this to a managed-services partner at $2,000 to $8,000 a month; either way, the line exists.

Then there is the AppExchange tax. The core platform rarely covers deduplication, document generation, advanced forecasting, routing or e-signature the way a sales team wants, and each gap is filled by an app at $10 to $75 per user per month. A stack of five apps on 25 seats adds $15,000 to $60,000 a year, and it creeps — each app is added to solve one problem and never removed.

Finally, price increases. Enterprise SaaS vendors raise list prices regularly, and your renewal is negotiated from a position where your data and processes already live inside their product. The switching cost is the moat, and it is priced into your renewal. None of this makes Salesforce a bad choice — it makes the 5-year number, not the year-1 number, the right basis for the decision.

⚖️When custom CRM wins on 5-year TCO

The crossover is arithmetic, not ideology. Salesforce cost grows with seats; custom CRM cost is mostly fixed at the build and then grows with maintenance. That means headcount is the pivot variable: below roughly 15 to 25 users, Salesforce (or HubSpot) is cheaper over five years no matter how you slice it. Somewhere between 40 and 80 seats — depending on edition and add-ons — the cumulative licence and admin spend passes the cost of a custom build plus its maintenance, and the gap widens every year after.

The second pivot is process fit. If your sales process maps cleanly onto lead-contact-opportunity with standard stages, you are paying a custom price to build what a platform already does, and you should not. If your "CRM" is really a workflow system — quoting logic, multi-party approvals, industry-specific data models, compliance workflows — you will pay to bend Salesforce into that shape through configuration and AppExchange apps, and the bending is often the majority of the implementation bill. At that point building the shape directly is frequently cheaper and always cleaner.

The honest caveats run both ways. Custom CRM means you own the roadmap and the data model, but you also own uptime, security patches and the bus factor of whoever built it. Salesforce means a vendor carries the platform risk, and you carry the licence risk. Neither is free; they are different shapes of cost.

📦Migration: the line item everyone underestimates

Moving into a new CRM is a data project wearing a software costume. The visible part — exporting records and importing them — is the cheap part. The expensive part is everything around it: deduplicating a decade of contacts, mapping fields that do not correspond, preserving activity history and attachments, re-pointing integrations, and retraining a team whose muscle memory lives in the old system.

As labelled ranges we observe: a clean migration from one structured system (say, HubSpot to Salesforce) with under 100,000 records runs $8,000 to $20,000. A messy one — spreadsheets plus an old CRM plus email history, duplicates everywhere — runs $20,000 to $40,000 and up, because the work is data cleaning and reconciliation, not transport. Migrating off Salesforce to a custom system adds a reverse-mapping exercise: years of accumulated custom fields, flows and reports have to be inventoried and either rebuilt or deliberately retired.

Two practices cut migration cost more than any negotiation. First, clean the data in the old system before moving it — deduplication is far cheaper where the data already lives. Second, run both systems in parallel for one sales cycle rather than doing a big-bang cutover; the parallel month feels expensive and is consistently cheaper than the revenue damage of a botched cutover.

🧾A worked example: 40 seats, five years

Abstract crossovers only go so far, so here is the anatomy of one comparison — a composite illustration, labelled as such, not a quote. A 40-person revenue team, standard-to-moderately-custom sales process, two integrations (billing and a product database), migrating off spreadsheets and a legacy CRM.

Salesforce path: Enterprise licences at 40 seats run roughly $79,000 a year at list. Implementation with a mid-market partner lands around $35,000 to $55,000. Migration of the messy data, $20,000 to $30,000. A part-time-to-full admin, $40,000 to $80,000 a year. Two or three AppExchange apps, $10,000 to $25,000 a year. Five-year total: roughly $500,000 to $800,000, before any price increases.

Custom path: a build covering pipeline, accounts, quoting, the two integrations and role-based reporting runs $90,000 to $150,000. The same migration, $20,000 to $30,000. Hosting and maintenance at 15 to 25 percent of build, roughly $15,000 to $35,000 a year. Five-year total: roughly $185,000 to $335,000. The custom path is not cheaper because engineers are cheaper than licences — it is cheaper because at 40 seats the per-seat annuity has become the dominant term. At 12 seats the same arithmetic favours buying, and we would say so.

🧭How to make the call

Decide with a spreadsheet, not a demo. The inputs are knowable before you sign anything: your realistic seat count at year 3, the edition your integration and automation requirements force, the add-on apps your process needs, the admin cost, and the migration state of your current data. Price both paths on those inputs over five years and the answer is usually obvious enough that the interesting work is checking the assumptions.

Talk through the build-vs-buy math with our CRM team→Build vs Buy CRM: Custom CRM vs Salesforce (2026)→