Self-Built · SaaS Pricing & Monetization
How 191 SaaS Companies Really Price in 2026
Self-Built · Open Dataset · Tableau
- LocationBoston, MA
- Year2026
- TypeSelf-Built Project
Every SaaS company in 2026 is asking the same two questions: is per-seat pricing dying, and how should we charge for AI? The industry answers with confident statistics — for example, an ICONIQ Capital survey published by BCG found 68% of vendors charge separately for AI or restrict it to premium tiers.
The trouble is the credible estimates measure different things and don’t agree — a named 230-company survey by Kyle Poyar puts it near 29%, while the ICONIQ/BCG figure is 68% — and the highest numbers come from aggregator pages with no stated method. There was no open dataset to settle it.
So I built one: I collected pricing data from 198 leading SaaS companies directly from their live pricing pages, and used it to answer the questions with evidence instead of citation.
Sampling. 198 recognisable B2B SaaS companies, 18 in each of 11 categories — dev tools, data & analytics, CRM, martech, collaboration, design, security, HR, fintech-ops, support and AI-native — balanced so no segment skews the result.
Collection. Each live pricing page was read and converted into 23 structured fields: pricing model, tier names, entry price, how AI is charged, where SSO sits. Prices normalised to USD per month. No third-party datasets, no synthetic data.
Recovery. The most valuable pages fight back — Salesforce, OpenAI, Adobe and Canva block scrapers or render prices in JavaScript. Those were recovered with a headless browser and manual verification, lifting coverage from 80% to 96%.
Validation. Every headline number was recomputed through three independent paths — a direct recount from the raw files, the analysis pipeline, and the exported dashboard dataset — and all three agree exactly. Field-level extraction accuracy is roughly 90%, and every row carries its source URL and a quoted snippet so any value can be audited.
From 191 companies with usable pricing data (96% of the sample), across 11 categories.
I built the analysis into a Tableau dashboard, published live on Tableau Public. It is fully interactive — filter the whole dataset by category and explore every finding below.
The headline finding is that the most-cited statistic in SaaS pricing overstates what a customer actually pays. I am not claiming the reports are dishonest — I am claiming three specific things: credible sources disagree (a 230-company survey says ~29%, ICONIQ/BCG says 68%), the higher figures are not reproducible from public pricing pages, and the gap is largely definitional. Self-reported “we monetise AI” is a much easier bar than “the pricing page actually charges you more.”
| Source | What it measures | Figure |
|---|---|---|
| Kyle Poyar, Growth Unhinged (survey, n=230) | Companies that actually meter AI via credits | ~29% |
| This study (n=191) | Pricing page actually charges the customer extra for AI | 34% |
| ICONIQ Capital (via BCG, 2025) | Charge separately for AI, or restrict it to premium tiers | 68% |
My observed 34% lands right next to the one rigorous independent survey — Kyle Poyar’s 230-company survey finds ~29% actually meter AI — and at roughly half the self-reported 68%. The number you can check tracks the independent survey; the higher figure relies on self-report and a looser definition.