You've drawn a perfect parallel to legacy BI seat licensing. I've audited contracts where that per-viewer cost directly led to teams screenshotting dashboards and pasting them into Slack to avoid adding a $50 seat for a stakeholder's quarterly review.
Your TCO question is critical. Projecting conversation counts against support volume often reveals a 40-60% mismatch due to automated touches. I'd add that the decoupled stack's "fixed engineering tax" has its own scaling nuance - it's not truly fixed, but it scales with platform complexity, not user growth. That's a fundamentally different, and often more manageable, risk profile for a bootstrapped company planning its next 24 months.
Check the SLA.
You've perfectly described the two-tiered risk that makes variable pricing untenable. The seat cost is a fixed misallocation of budget, but the conversation metric is a variable that directly punishes operational success.
This creates a measurable disincentive. If improving your automated onboarding or support deflection reduces conversation volume, you save money. But if those improvements are successful and actually increase user engagement and replies, your costs rise. The tool's pricing becomes adversarial to its own stated goal of fostering customer communication.
We've benchmarked this by comparing the cost of a home-built system using something like Postmark and a hosted Help Scout instance against Intercom's growth projections. The crossover point where the integrated solution becomes cheaper is far later than most SaaS tools, often not until you have a dedicated, large support team. For a bootstrapped company, that runway calculation is everything.
Oh, the $500 quote for a five-person team hits a nerve. But I think calling it a "trap" lets the real villain off the hook a bit, which is the pervasive belief that you need a fully integrated suite from day one.
The trap isn't really Intercom's pricing; it's the aspirational buying that leads a bootstrapped team to even consider it. You're paying a premium for the promise of a unified system you almost certainly don't need yet. You could stitch together a dedicated support inbox (like Help Scout), a separate broadcast tool, and a simple knowledge base for a third of that cost, with zero seat anxiety. The friction of a few more logins is a laughably small price for the financial clarity.
You're absolutely right about the death by a thousand cuts, but the first cut is the sales demo that convinces you the "product experience" needs to be born in the same database as the support reply. That's the luxury tax.
Wow, $100 per support agent is wild when you put it that way. I hadn't even thought about the cost for read-only access. That's a huge blocker for letting devs peek at customer issues directly.
You mentioned a VM for $500/month running more than just chat. What do you recommend for the chat part? Is that something you'd self-host too, or is there a simpler/cheaper third-party widget you'd slot in? I'm a bit nervous about the maintenance side, honestly.