Custom CRM vs. Off-the-Shelf: The Build-vs-Buy Math

We get one question from owners at that stage more than any other: when does building our own stop being crazy?

Sometimes never. More often than most CRM vendors will tell you, the answer is sooner than you think. This is the framework we use to answer that question with math instead of instinct.

What an off-the-shelf CRM actually costs

Published prices are the honest part of the bill. Current list pricing for the two platforms every owner shortlists:

Platform	Tier	List price
Salesforce Sales Cloud	Starter Suite	$25 per user/month
Salesforce Sales Cloud	Pro Suite	$100 per user/month
Salesforce Sales Cloud	Enterprise	$175 per user/month
Salesforce Sales Cloud	Unlimited	$350 per user/month
HubSpot Sales Hub	Professional	$100 per seat/month, plus $1,500 mandatory onboarding
HubSpot Sales Hub	Enterprise	mandatory $3,500 onboarding, seat price quoted

Ten seats on Salesforce Enterprise is $21,000 a year. Five years, $105,000, before a single consultant invoice. Ten seats on HubSpot Professional is $12,000 a year plus the $1,500 onboarding fee every new customer pays.

Then the bill grows quietly:

• Support tiers. Salesforce's Premier Success plan costs 30% of your license fees, every year. On a $21,000 contract that is another $6,300 annually for faster answers.
• Integration rights. On Salesforce's lower tiers, API access is an extra $25 per user per month. Paying for the right to connect your own tools is a strange line item, and a common one.
• Admin time. Someone has to run the thing: a part-time internal admin or an outside consultant, and platform consultants typically bill $100 to $250 an hour.
• App sprawl. HubSpot advertises more than 2,000 marketplace apps. Read that number differently: it measures how often the core product does not do the thing a business needs.

List price is the entry fee. The real cost lives in those four lines.

The ten-person tipping point

Under about eight seats, with a standard pipeline, off-the-shelf wins. Every time. We do not argue with that math and neither should you.

The pattern changes between eight and fifteen seats. We have watched it arrive at enough companies to name the signals before a prospect finishes describing them:

1. Your process bends to the tool. Field names get repurposed. A dropdown labeled Industry quietly means something else. Someone maintains an unofficial document explaining how the team fudges the pipeline stages.
2. The spreadsheets move back in. Somebody spends five to ten hours a week copying data between the CRM and an ops spreadsheet because the two refuse to talk.
3. Every gap becomes a new subscription. Quoting lives in one tool, forms in another, reporting in a third. Each has its own per-seat fee and none of them share data cleanly.

One signal is noise. Three is the tipping point. Past it, every new requirement gets priced in consultant hours, because you are paying specialists to make a general-purpose product act specific.

The five questions that settle it

Score yourself honestly. These are the questions we walk through before anyone writes a line of code.

1. Does your data model fit their template? A standard B2B pipeline fits anywhere. Freight brokerage, field service dispatch, property management, clinical referrals: none of these fit a template built for software sales reps.

2. How many systems does the CRM have to talk to? Each connection on a SaaS platform is an app purchase, an API fee, or a fragile automation chain. In a custom build the same connections are code you own.

3. Who maintains it, and at what rate? SaaS maintenance is admin time plus annual subscription increases. Custom maintenance is a support retainer. Compare both over five years, not one.

4. What is the five-year seat math? Seats times tier times 60 months, plus onboarding, plus add-ons, plus admin hours. Set that total next to a build quote plus a retainer. It is the column most owners never add up.

5. Where does your data live, and how do you leave? The longer you run on someone else's schema, the more your reporting, your automation, and your team's habits depend on it. Exit costs grow every quarter.

Zero or one "that's us": stay where you are. Three or more: the math has flipped, and pricing the build is worth an afternoon.

When a custom build pays for itself inside 18 months

License versus license, a build rarely wins fast. A $50,000 build against $12,000 a year in seats breaks even in year four, and nobody should pretend otherwise.

The payback window collapses when you count the four buckets the seat price hides:

• Retired subscriptions. When the CRM natively handles quoting, forms, and reporting, the standalone tools go. Consolidating $800 to $1,500 a month of overlapping subscriptions is typical in the builds we have shipped.
• Reclaimed hours. Ten hours a week of spreadsheet reconciliation at a loaded $40 an hour is about $20,000 a year. That one clerk-shaped cost can cover a third of a build by itself.
• Deals that stop falling through. Automated follow-up and a pipeline the owner can actually see. For a ten-person team, recovering two or three forgotten deals a quarter pays for a lot of software.
• Growth you already planned. Adding five seats next year costs $6,000 to $15,000 a year, forever, on a subscription platform. On software you own, the marginal cost of a seat is zero.

Run it end to end: a $55,000 build, $38,000 a year recovered across the four buckets. Payback lands around 17 months. That is the shape of the deal where we tell an owner to build.

When you should not build

The honest section, and the one a custom shop is least likely to write:

• Under five seats with a standard pipeline: stay on SaaS. The math is not close.
• Your process is still changing monthly. A custom build freezes process. Freeze chaos and you get faster chaos.
• Nobody inside the business owns it. The strongest predictor of a dead custom CRM is not bad code. It is the absence of an internal owner who cares whether the data is right.
• The requirement is "Excel, but prettier." If the underlying process is a spreadsheet, automate the spreadsheet first and revisit in a year.

Any shop that never talks you out of a build is a shop to avoid. We would rather lose the project in the first call than rescue it in year two.

What a scoped build actually looks like

Custom does not mean two years and a vanished budget. A scoped v1 for a ten-to-twenty-five person company runs eight to fourteen weeks:

1. Discovery, one to two weeks. Map the real process, including the spreadsheet workarounds. The workarounds are the requirements.
2. Core build, six to ten weeks. Contacts, pipeline, and the single automation that saves the most hours. Nothing else.
3. Ship in slices. The sales team gets v1 with the fields they already use, not a forty-field form. Adoption is a design decision, not a training problem.
4. Integrations and reporting second. Once the core is in daily use, wire in accounting, phones, and email. Then the dashboards.

The build order matters more than the technology. Ship the boring core first and the rest of the project funds itself out of saved hours.

Run your number before your renewal does

If the three tipping-point signals sounded familiar, your next renewal letter is about to price this decision for you. We put the five-year math on one page with owners before recommending anything, and custom CRM development starts with that page, not a build quote. Sometimes the math ends with "stay on HubSpot." We say that too, in writing, before anyone spends a dollar.