When the business asks for a CRM, engineering usually gets asked whether to build one. Building ties up engineers you probably need on the product, while buying means a licence bill of roughly $60,000 to $117,000 a year for fifty users at list price, so the decision is worth getting right.
Buy an off-the-shelf CRM when your sales and service process looks like most companies' in your industry, which is most of the time. Build a custom one when the workflow the CRM would run is the thing your customers pay you for, or when the licence bill for the users you need costs more over five years than building and running the narrow tool they actually use.
Between those two sits the answer most mid-market companies end up with, and it's usually the right one. They buy the CRM, then have their engineers build the pieces around it that make it fit the business.
What off the shelf costs in 2026
Per-user list prices, billed annually, as published on the vendors' pricing pages on 11 October 2026.
| Product | Edition | Per user per month |
|---|---|---|
| Salesforce | Free Suite | $0 |
| Starter Suite | $25 | |
| Pro Suite | $100 | |
| Core | $195 | |
| Advanced | $395 | |
| Max | $550 | |
| Microsoft Dynamics 365 Sales | Sales Professional | $65 |
| Sales Enterprise | $105 | |
| Sales Premium | $150 |
One thing in that table has changed since this article first ran in 2024. Salesforce's sales pricing page now lists its editions as Core, Advanced and Max, so a comparison written against the old edition names is out of date.
The gap between the cheap tier and the one with the features you need is also wide. Fifty users on Salesforce Pro Suite cost $60,000 a year at list. The same fifty on Core cost $117,000. On Dynamics 365 Sales Enterprise they cost $63,000.
List price is where the conversation starts. Enterprise agreements are negotiated, and the bigger costs often sit outside the per-seat line. Think of the implementation partner, add-ons for CPQ, telephony or AI features, extra storage, and the admin who keeps it configured.
The questions that decide it
Is the workflow standard? Leads, opportunities, accounts, cases, a pipeline with stages. If your process can be described with those nouns, packaged CRMs have had decades to get it right, and building it again is spending money to arrive where you started.
Is the workflow the product? If what the CRM would manage is the core of what you sell (matching a borrower to a lender, routing a repossession assignment, pricing a dealer's inventory against a customer's equity), the CRM's data model becomes your product's data model. Bending a general-purpose CRM to fit it means custom objects, custom code and a licence per user for software you've largely replaced.
How many people need it, and how much of it? Per-seat pricing punishes broad, shallow use. Two hundred warehouse or field staff who need to see one status screen cost the same per seat as a salesperson using everything. That's the case where a narrow custom app reading from the CRM can be far cheaper than licensing everyone.
What does it have to connect to? Count the systems: ERP, ecommerce, billing, marketing, telephony, your own product. Every packaged CRM integrates with the big names. The cost lives in the systems specific to you, and that integration work exists whether you buy or build.
Who will own it in year three? A packaged CRM needs an administrator. A custom one needs engineers who know its code. Neither is free, and the one with no owner is the one that decays.
Switching costs nobody demos
Leaving a CRM is harder than choosing one, and the reasons are predictable.
- Custom objects and code. Fields can be exported. Validation rules, flows, Apex classes or plug-ins can't, and they encode business logic somebody will have to rediscover.
- History and attachments. Exports flatten activity history, email threads and files into separate tables or archives that have to be re-linked on the way in.
- Integrations. Every system wired to the CRM's API needs rewiring to the new one, usually all on the same weekend.
- Habits. Reports people rely on, built by people who've since left.
Before signing, export a full copy of your data in the first month, load it somewhere else, and see what you'd lose. If the answer frightens you, that's the lock-in you're buying, priced before you need to know it.
Three patterns that work
Three shapes recur in our published case studies.
Buy, and wire it into everything else. AmTab runs Dynamics 365 with a nopCommerce store and Business Central. Power Automate keeps the store, the CRM and the ERP holding the same records, so orders placed on amtab.com reach Business Central without anyone re-keying them. The CRM is standard. The value is in the connections.
Buy the CRM, build the engine beside it. Auto Approve needed an automated approval engine for vehicle refinancing, not a different CRM. Building that engine doubled monthly loans processed and replaced more than four manual workflows without adding agents. At iLending, lead and loan officer assignment, rate sheet processing and stipulation documents were automated alongside improved CRM integration.
Build it into the product. TotalLoop started as a cloud-native platform on Azure with its own CRM and a sales dashboard that matches customer equity to inventory on the lot. There, the CRM is part of what the dealership is buying, so owning the data model is the point.
A decision you can make this month
- Write down your sales and service process in one page, using your own words. Circle anything a packaged CRM's standard objects wouldn't hold.
- Count users by how much of the system each one needs, not by headcount.
- Price five years of licences at list for that count, then add the quoted add-ons, the implementation partner and an administrator's time.
- List every system the CRM must talk to, and mark the ones only your company uses.
- If little is circled and few systems are yours alone, buy. If the circled part is your product, build that part, and consider buying for the rest.
If the answer is "buy, then connect it properly", or "build the part that's yours", that's the work our custom software development team does. Talk to us about what you're weighing.
Frequently asked questions
Does the per-seat bill change if only some of my users need the full CRM?
Yes, because every seat costs the same per month whatever the person does in it, so the headcount you license decides the bill, not the features used. Counting users by how much of the system each needs gives you two groups: the few who need the full CRM, and the many who need one screen. The second group is where a narrow custom app reading from the CRM starts to compete with buying more seats.
How do I find out what leaving the CRM would cost before I sign?
Export a full copy of your data in the first month and load it into another system. What fails to load or loses its links is the cost of leaving, and you learn it while you can still negotiate. Check validation rules, flows and code by hand too, since those don't export and the business logic in them has to be rediscovered.
Is the list price what I'll actually pay for a CRM?
No. List price is where the conversation starts. Enterprise agreements are negotiated, so your seat price may differ from the table. The implementation partner, add-ons and the administrator belong in the five-year number as well, and we'd price them before comparing against a build.
If I buy a CRM, do I still need engineers on it?
Yes, though for a different job than building one. A bought CRM still needs an administrator, and the integrations to your own systems exist whether you buy or build. Engineers end up on the connections and on any engine built beside the CRM, which is where the value sat in the AmTab and Auto Approve examples.