What Replacing HubSpot Actually Costs
Every "custom CRM cost" article on the internet says the same thing: $50,000 to $150,000, six to twelve months, hire an agency. They're not wrong for the projects they're describing. But they're describing a process that doesn't apply to every situation, and the range is so wide it's useless for making a decision.
We replaced HubSpot for a 25-person commercial real estate SaaS company. Not a toy. Not a prototype. A deployed, authenticated, security-hardened CRM with MFA, deal pipeline, event management, contact segmentation, and the one thing HubSpot couldn't do at any price tier: full-funnel attribution that actually tells you which marketing channel produced which revenue.
Five days. 456 tests. Deployed at a custom domain. Monthly cost: hosting and database, not a per-seat SaaS subscription.
Here's what it actually looked like.
What we were replacing
HubSpot Marketing Hub Professional. The team was paying for it because they needed CRM basics — contacts, deals, a pipeline view — and marketing attribution. The attribution was the whole point. Which trade shows produce deals? Which LinkedIn campaigns convert? Which content drives demos?
HubSpot's answer at the Professional tier was mediocre. Their attribution reporting required manual configuration that nobody maintained, and the data was spread across HubSpot, Airtable, and spreadsheets. The team rated it 1.8 out of 5 internally.
The cost: $3,600 per month. $43,200 per year for a tool the team actively disliked.
What we built
A custom CRM called GROPE (it's an internal tool, the name stuck). The feature set:
Authentication and security. Custom magic link auth — no passwords to manage, no OAuth dependency. TOTP-based multi-factor authentication. Session management, CSRF protection, rate limiting. Not "we'll add security later." Security from day one.
Contact management. Full contact records with company associations. Apollo CSV import with fuzzy matching so existing prospect lists transferred cleanly. Segmentation by tags, company, engagement history.
Deal pipeline. Drag-and-drop kanban board. Custom stages matching the team's actual sales process, not a generic template. Deal values, expected close dates, associated contacts.
Event management. Trade shows, conferences, webinars — all tracked with attendee lists, costs, and post-event follow-up status. This was in Airtable before. Now it's in the same system as the contacts and deals.
Funnel attribution. This is the part HubSpot couldn't do. Every form submission, every page visit, every UTM parameter captured and associated with the contact record. When a deal closes, the team can trace it back to the original source: which ad, which event, which blog post. First-touch and multi-touch models, not just "last click."
The stack
Next.js 16 on Vercel. Neon Postgres with Drizzle ORM. shadcn/ui components with a custom design system (we called it Warm Void — dark theme, specific color palette, consistent spacing). Solar icon set.
No external auth service. No third-party CRM framework. No "CRM starter kit" that locks you into someone else's data model. Every table, every API route, every component written for this specific team's workflow.
The real cost breakdown
Here's where the internet's "$50K-$150K" estimates break down. Those numbers assume:
- A team of 3-5 developers
- A project manager
- A designer producing mockups in Figma
- Sprint planning, standups, retrospectives
- QA as a separate phase
- Deployment as a separate phase
- "Discovery" as a billable phase
Our process was different. One developer. Tests written before features (456 of them). Design decisions made in code, not in mockups. Deployment continuous from day one — every commit that passed tests was live within minutes.
What we spent money on:
| Item | Cost | |------|------| | Vercel Pro (hosting) | $20/month | | Neon database | $19/month | | Domain (grope.app) | $14/year | | Total monthly | $39/month |
No designer fees. No project management overhead. No QA team. No "discovery phase" invoice.
What we spent time on:
Five working days. Not five sprints. Not five weeks. Five days of focused building with continuous testing and deployment.
Day one: auth system, database schema, core API routes. Day two: contact management, company associations, import pipeline. Day three: deal pipeline with kanban. Day four: event management, attribution tracking, form submission capture. Day five: design system application, MFA, security hardening, seeded data.
What the agency quotes miss
The honest version: our situation had advantages that most custom CRM projects don't.
We knew exactly what to build. The team had been using HubSpot for years. They knew what they needed and what was bloat. There was no "discovery" because the requirements were the daily frustrations of people who'd been living with the wrong tool. When someone says "I need to see which trade show produced this deal," that's a spec.
The user was in the room. Not "we'll schedule a review meeting next sprint." The person who'd use the CRM every day was available for questions in real time. This eliminates the telephone game that adds weeks to agency projects.
Testing replaced QA. 456 automated tests meant we didn't need a separate QA phase. Every feature was verified before it shipped. Every edge case was documented in code, not in a spreadsheet that someone checks manually.
No handoff tax. The person who wrote the auth system also wrote the deal pipeline also wrote the attribution tracker. No context lost between teams. No "the backend team didn't know the frontend expected this format" bugs.
But here's the thing: these aren't unusual advantages. They're what happens when you skip the process theater and focus on the actual work. A solo developer who understands the domain, talks to the user, and writes tests can move faster than a team that spends half its time coordinating.
The math that matters
HubSpot Professional: $43,200 per year.
Custom CRM hosting: $468 per year.
Break-even on even a $50,000 custom build: 14 months. After that, it's $42,732 per year in savings.
Our build cost far less than $50,000. But even at agency rates, the math works within two years for any team paying for HubSpot Professional or above.
And the custom build does something HubSpot never did: show exactly which marketing dollar produced which revenue. The attribution that was supposed to be HubSpot's value proposition is now actually working, because it was built for this specific team's workflow instead of being a generic module that nobody configured correctly.
When this doesn't apply
Custom isn't always the answer. If you need marketing automation — email sequences, lead scoring, workflow triggers across thousands of contacts — HubSpot or ActiveCampaign or similar tools have years of infrastructure you'd be foolish to rebuild. If your team changes CRM requirements monthly, a configurable SaaS tool absorbs that churn better than custom code.
The sweet spot for custom: your team has stable, specific workflows that generic tools handle poorly, and you're paying enterprise prices for features you don't use. That described this team exactly.
What happened after
The CRM is deployed and in use. The team can see attribution data they never had before. The $3,600 monthly invoice is gone.
The building-grope series on this blog covers the technical details — the gate that changed behavior, the integration tax, the POST that never arrived. Real bugs, real architectural decisions, real lessons from shipping a production application under pressure.
If your team is paying for a CRM that doesn't do what you need, the question isn't whether custom is affordable. It's whether you can find someone who'll build the right thing instead of billing you for the wrong process.
I build things like this. Here's the portfolio.