Build vs buy — when a custom CRM or ERP beats Salesforce and HubSpot

The bottom line

Framed as “custom CRM versus Salesforce,” the question loses either way. Rent everything and you bend your process around a vendor’s data model, pay per seat forever, and still end up running the genuinely specific work in a spreadsheet beside the tool. Build everything and you burn a five-figure budget rebuilding contact lists and email logging that HubSpot rents for $20 a seat. The split that works: buy the standard 80% — contacts, pipeline, email, calendars, invoicing — and build only the 20% no vendor models, because that 20% is usually the workflow your margin comes from.

Seat count decides the rest. A per-seat subscription is a meter that runs forever and rises with headcount; a build is a one-time cost that doesn’t care how many people log in. We do that math below with real mid-2026 list prices, and we also mark the cases where the math says don’t build — there are more of those than a custom-software shop is supposed to admit.

Buy the standard 80%

Where your process is genuinely standard, SaaS wins and it isn’t close. A pipeline with stages, a contact record with an activity log, follow-up reminders, email sync — that’s a solved problem, sold at scale, with a decade of other people’s edge cases already fixed. HubSpot’s Sales Hub Starter lists at $20 per seat per month and Salesforce’s Starter Suite at $25. Nobody can build you the equivalent for anything near that, and nobody honest will offer to. The same goes for accounting ledgers, payroll, calendars, and email itself: these are commodities, and building a commodity is paying custom prices for a generic outcome.

The failure of buy-everything isn’t the buying — it’s the over-buying. A workflow gap appears, and the reflex is to go up a tier, bolt on a module, or add seats for people who touch the system twice a month. It compounds quietly: Zylo’s 2026 SaaS Management Index found the average organization uses only 54% of the SaaS licenses it pays for. If your CRM bill keeps climbing while the actual gap — the quoting, the job scheduling, the production handoff — still lives in a spreadsheet next to the tool, you’re not under-tooled. You’re paying standard-software prices to not solve a non-standard problem.

Build the 20% that makes your margin

Every business with a real operational edge has a slice of process no product sells, because modeling it would mean modeling that specific trade. That slice is almost never the pipeline. It’s the quote that has to be derived from geometry, the allocation rule a founder can recite but no settings screen accepts, the commission split that mixes product lines at different rates.

A custom-furniture manufacturer we built for is the clean example. Its standard 80% was never the problem — email, accounting, and messaging were all fine as rented software. The margin leaked in the 20%: quoting a parametric design, computing the cut list from the same model, nesting parts onto tracked offcuts, rolling confirmed orders up into purchase requirements, settling tiered commissions at close. No CRM or ERP vendor models a panel width that is a formula over total width, stile, and panel count, and no amount of configuration gets one there. We built that loop as one system on one database, and the shop keeps renting the commodity layer around it.

That’s the general shape of our custom-systems work: engagements run $20,000–60,000 one-time, an engineer embeds in the business, and the deliverable is a working system you own — code, data, deployment. What made this affordable for SMBs at all is the delivery model: forward-deployed engineering with AI doing the throughput work compressed builds from quarters to weeks, and the price with it.

The seat math, with real prices

Now the arithmetic. Current list prices, per seat per month, billed annually:

HubSpot Sales Hub Professional is $90 ($100 month-to-month), plus a one-time $1,500 onboarding fee; Enterprise is $150 plus $3,500 onboarding. Salesforce’s Pro Suite is $100, Enterprise $175, Unlimited $350 — and partner implementations of Salesforce typically start around $25,000 on top of the licenses.

Take a 15-person sales-and-ops team — a small business, not a micro one. On HubSpot Professional that’s 15 × $90 × 12 = $16,200 a year, about $50,100 over three years with onboarding. On Salesforce Enterprise it’s 15 × $175 × 12 = $31,500 a year — $94,500 over three years, before the implementation partner sends their invoice. Against that, a $40,000 build (the midpoint of our range) breaks even against Salesforce Enterprise in about 15 months and against HubSpot Professional in about two and a half years. At 30 seats, halve both figures. And the build has no meter: the 31st user costs nothing, where the subscription bills them like the first.

The rule of thumb: divide the build cost by your true annual license bill at the tier that actually fixes your problem — which is usually a tier or two above the one you’re on, because that’s how vendors price the fix. Under a year to break even, the math argues loudly for building. Over three years, it argues for buying, and workflow fit alone has to carry the decision.

Two honest corrections before you run off with that number. First, a build isn’t free after delivery: hosting for a typical internal system on serverless infrastructure rounds to noise — the stack we deploy scales to zero when idle — but changes cost engineering time whenever you order them, while a SaaS vendor ships improvements continuously on their own dime. If the vendor’s roadmap is your roadmap, that’s real value the break-even math doesn’t capture. Second, at five seats on Starter tiers the license bill is $1,200–1,500 a year, and no build on earth pays back on license savings against that. Small teams build for one reason only: the workflow gap costs a multiple of the license bill in wasted hours, wrong quotes, or lost jobs. That happens — a spreadsheet-run production process usually costs more than all the software around it — but then you’re buying back margin, not saving on seats.

Side by side

	Buy (HubSpot / Salesforce)	Build (custom, one-time)
Upfront	$0–3,500 onboarding; Salesforce implementations from ~$25k	$20k–60k, delivered in weeks
Ongoing	$20–350 per seat per month, forever	Hosting near zero; changes billed as work you order
Fit	The standard 80%; the edges live in spreadsheets	Shaped to the 20% that makes your margin
Time to first value	Days	Weeks
Improvements	Continuous, on the vendor’s roadmap	Only what you ask for
Adding seats	Linear cost, every month	Free
Ownership	Rented; data export on exit	Code and data are yours

When not to build

Our own service FAQ says it plainly and it stays true here: if a standard subscription genuinely covers your needs, take it — it’s cheaper, it’s faster, and you don’t need us. Concretely, don’t build when:

• You can’t name the margin-making workflow. If the pain is a vague “our CRM is clunky,” an admin cleanup or training usually beats both a tier upgrade and a build. A build needs a target you can describe in one sentence.
• Your edge isn’t operational. If you compete on brand, price, or product and your sales motion is standard, standard software is the right container for it.
• The seat count is small and the tier is cheap. Five seats at Starter prices never closes the arithmetic above. Build only if the workflow leak dwarfs the license bill.
• The gap is integrations, not workflow. If the tools are right but disconnected, glue is far cheaper than a platform — that’s automation work, not a system build.
• Vendor certifications carry your compliance load. In some verticals the audited, certified SaaS stack is most of what you’re paying for, and re-carrying that yourself costs more than the seats ever will.

How to decide in an afternoon

Three numbers settle most cases. Write down the one workflow where hours or margin visibly leak, and what a month of that leak costs — wrong quotes, scrapped material, deals lost to slow follow-up; be rough, but be numeric. Get your real all-in SaaS quote at the tier that would actually fix it, at your real seat count, onboarding and add-ons included. Then divide a build quote by the annual difference. If the leak is bigger than either number, the decision was never really about seats — fix the workflow with whichever side fixes it, which for genuinely trade-specific work means building.

And if all three numbers come out small, close the tab and keep the subscription. The honest end of this comparison is that most businesses shouldn’t build most things. The ones that should already know exactly which workflow it is — they’ve been fighting it in a spreadsheet for years.

Sources

1. HubSpot Sales Hub pricing — Starter $20/seat/month (monthly; lower annually), Professional $90 (annual billing; $100 monthly), Enterprise $150; one-time onboarding $1,500 (Pro) / $3,500 (Enterprise). Accessed July 2026.
2. Salesforce Sales pricing — Starter Suite $25, Pro Suite $100, Enterprise $175, Unlimited $350 per user per month, billed annually. Accessed July 2026.
3. Cargas: 2026 Salesforce Pricing Guide — edition prices cross-checked; partner implementations start around $25,000 (Salesforce implementation partner, 2026).
4. Zylo: How much is wasted on SaaS spend — 2026 SaaS Management Index: the average organization uses only 54% of its SaaS licenses. Accessed July 2026.