When I set up sales at Purcell Ventures, HubSpot was the obvious choice, and it has a free tier. I spent two days evaluating it, then decided to build my own CRM, which took a week.
HubSpot pricing
The free tier excludes the features below. These are the plan prices HubSpot listed in May 2026, and their current prices are on HubSpot's pricing page.
- Custom properties beyond 5, at $50/mo (Starter)
- Workflow automation, at $890/mo (Marketing Hub Pro)
- Pipeline forecasting, at $720/mo (Sales Hub Pro)
- Custom reports, at $50/mo+
- API access beyond rate limits, at $720/mo
I needed multi-rep visibility, AI-assisted next steps, custom contact stages, an activity timeline and email templates with tokens. That would have cost $200 to $400 a month, or $2,400 to $4,800 a year.
What I built instead
One week of focused work produced a CRM with 6 views (List, Kanban, Tasks, Stats, Coach, Reps), per-rep auth, AI features specific to my workflow, and an activity log built around the sales conversations my reps and I have.
The cost so far is $15 in Anthropic API usage for the AI features. There is no monthly subscription.
Upkeep compared with HubSpot
The usual objection is that a custom CRM takes more upkeep than HubSpot. For me it has taken less, for four reasons.
- I control feature scope. When HubSpot adds features I don't need, its interface gets more complex. When I add a feature, the interface grows by that one feature.
- I push fixes myself. If my CRM has a bug, I fix it the same day. If HubSpot has a bug in my workflow, I file a ticket and wait.
- Integrations are small jobs. The /sales-rep form creates a CRM contact. The Steady intake form opens an email to me. With HubSpot, that is either a paid feature at $50 a month or a Zapier connection at $30 a month plus API setup time.
- AI features cost extra on third-party CRMs. HubSpot's AI features (Breeze) are tied to a $720 a month plan. Mine run on Claude Haiku at $0.001 per call.
What did I give up?
- Polish. HubSpot looks more sophisticated. Mine looks like a small-team CRM, because that's what it is. It would matter if I needed to impress investors.
- Mobile app. HubSpot has a native mobile app. Mine is a mobile-responsive web app, which works and is less smooth.
- Ecosystem integrations. HubSpot integrates with hundreds of tools. Mine connects to the tools I choose to wire up. So far that has been enough.
When should you not do this?
- If you don't code: use HubSpot.
- If your team has more than 10 people: you need the permissions, audit logs and compliance features that commercial CRMs ship with, and building them costs far more.
- If you need SOC 2 or HIPAA compliance: pay HubSpot or a similar vendor, because meeting those standards yourself is a large project.
- If you have more than 1,000 contacts and a complex pipeline: the database work grows.
- If you don't enjoy maintenance work: it will need maintenance now and then. If you hate that, pay someone else to do it.
When should you do this?
- You work alone or on a team of two or three
- You code (or have access to someone who does)
- Your workflow is non-standard and requires custom logic
- You care about owning your data
- You will run this business for a year or more, so the build cost spreads out
- You want to use AI in custom ways the SaaS CRMs don't support
Learning my own sales process
Building my own CRM forced me to learn my sales process in detail, because I had to design every stage and field.
The starter kit and custom builds
I packaged the system prompts, the reasoning behind the database schema and the operating handbook into the $19 PV AI Starter Kit. It includes the prompts behind my CRM's AI features, and the kit runs on any AI tool you choose.
If you want me to build a custom CRM for your business, see Custom Software. I would start from the patterns I used for my own.
Related reading
Part of the guide AI integration for small businesses.