HubSpot rolled out seat-based pricing globally on March 5, 2024 (March 6 in Asia-Pacific), replacing seat minimums across its Starter, Professional, and Enterprise tiers with a new Core Seat and View-Only Seat structure (HubSpot). Within months, customers in HubSpot's own community forums were reporting that more than 30 percent of their account users had lost edit access, and that their bills had risen roughly 23 percent, because editing a contact or a deal now required a paid Core Seat (CX Today). No feature shipped. The pricing model itself produced the price increase.
That is the trap built into every seat-based system: seats charge for headcount, not for value delivered, and headcount is the easiest number for a customer's finance team to interrogate at renewal. Founders choosing between usage-based and per-seat pricing are really choosing what their bill measures, and that choice compounds for years.
What the retention numbers show
The case for usage-based pricing is not theoretical. OpenView Partners found that public SaaS companies pricing on usage grew revenue 38 percent faster than peers relying purely on subscription or seat-based pricing, with roughly 25 percent higher net dollar retention (OpenView Partners). In OpenView's "State of Usage-Based Pricing: 2nd Edition" report, built from a survey of private SaaS companies run in July and August 2022, usage-based companies posted 125 percent net revenue retention against 115 percent for subscription peers, and grew 33.7 percent against 23.2 percent for non-usage-based peers (OpenView Partners). Adoption moved with those results: OpenView's 2021 benchmarks report found usage-based pricing adoption rose 32 percent versus 2020, with close to 45 percent of SaaS companies using some form of it that year (PR Newswire).
Snowflake is the reference case for why the numbers move that way. The company built its business on consumption pricing instead of seat licenses, drawing about 90 percent of revenue from usage-based billing while growing 106 percent year-over-year before it crossed $1 billion in annual recurring revenue (OpenView Partners). Snowflake customers don't pay more because they hired someone. They pay more because they ran more queries, which points the vendor's growth and the customer's growth in the same direction instead of opposite ones.
Why seats still win in the room
None of this makes seat-based pricing wrong for every company. Seats are predictable for a finance team to budget against, and predictability is what a buyer wants when a purchase has to survive a renewal committee. The problem HubSpot ran into was not that seats are a bad model. It was that redefining what a seat includes, mid-relationship, functions as a price increase dressed up as a packaging update, and customers noticed immediately.
The hybrid middle: charge for engagement, not headcount
Slack found a version of this answer a decade earlier. Its 2014 "Fair Billing Policy" charged customers only for seats that were actively used, crediting them proportionally for provisioned accounts that sat idle, a model credited with helping Slack reach $100 million in annual recurring revenue without a traditional outbound sales team (Slack Story). Slack did not abandon the seat. It tied the seat to actual use, so the bill reflected what customers were getting rather than what they had provisioned and forgotten about.
How to actually decide
Three questions settle most of these debates before a pricing meeting turns into a negotiation with legal.
- Does the product have a usage unit that scales with the value the customer gets, and that the customer can see coming on their own dashboard? API calls, storage, compute cycles, and transactions all qualify. If the honest answer is "people log in," it is a seat product, and forcing a usage meter onto it adds billing complexity nobody asked for.
- Can product and finance instrument that usage unit cleanly before the sales team sells against it, not after? Snowflake and the companies in OpenView's benchmark set had metering built before they priced around it, not bolted on once customers complained.
- If seats already exist, does the next pricing change expand what a customer gets for the same seat, or does it fence off capability the customer already had? HubSpot's mistake was the second one, and it is the one that generates forum threads and churn risk instead of expansion revenue.
This week: pull the last twelve months of logo churn and expansion revenue, and sort each account by whether it grew through added seats or added usage. If expansion is coming almost entirely from usage and churn is concentrated in flat-seat accounts, the market has already told you which model to build toward. Price the next contract renewal on that evidence, not on which model is easier to explain on a sales call.