A pricing page is a marketing document. The real price lives in the footnotes: the seat that wasn't included, the credits that don't roll over, the feature that turns out to be an add-on. We hand-researched the pricing of 487 B2B and AI tools, reading the fine print, and mapped where the money actually leaks. Every number below is recalculated from that corpus whenever it changes, and every example links to the dossier that proves it.
The most common pattern in our corpus, found in half of the tools: a reasonable-looking subscription with a meter running underneath — credits, overage minutes, per-form fees, or a percentage of the money flowing through the tool. Metering isn't dishonest; it's misjudged. The sticker assumes usage limits that an actively used tool routinely exceeds, which means the price rises precisely when the tool is working.
The defense is arithmetic, not suspicion: calculate the price at your real volume — this month's and the volume you're hoping for. A fee of 10% on managed ad spend is ten times a $49 minimum once that spend reaches $5,000 a month. That's not hidden; it's just not on the button.
A quarter of the corpus scales its price with every user you add. Per-seat is the most honest-looking model — the number is right there — but it hides a compounding assumption: that everyone who touches the tool needs a paid login. Field crews, occasional viewers, the bookkeeper who checks once a month.
The question that saves the money: who needs a seat versus who needs a report? Many tools offer free guest, viewer or client roles; the difference between a 5-seat and a 12-seat bill is often just role hygiene.
In 28% of researched tools (137 of 487), the feature you actually came for is a paid add-on on top of the advertised plan. This is the trap that most reliably breaks budgets, because the gap between sticker and real can be multiples:
Rule: price the features you need, never the plan. Make a three-line list of must-haves before opening the pricing page, then find which tier — plus which add-ons — actually contains all three.
23% of the tools (114 of 487) show their headline price only on annual billing, or ask for a yearly commitment before a plan is available at all. The monthly rate is higher, and a year paid up front is money you cannot get back if the tool doesn't fit.
Start monthly when that option exists, even at the higher rate, and switch to annual only once the tool has proven itself in your real workflow.
1 in 70 tools in our corpus (7 of 487) publishes no usable price at all — a demo call stands between you and a number. Quote-based pricing isn't a scam; it usually means pricing genuinely varies by deployment. But it moves the negotiation to their home field.
Tools that publish an entry price and ask for a quote only above it are not counted here: the Quote-Only Club lists them separately.
The counter-move: demand the all-in figure in writing — base, seats, add-ons, implementation, and what happens at renewal. A vendor who won't write it down is telling you something.
The classic intro-price-then-triple trap appears in only 1 in 49 of our corpus (10 of 487), and it clusters in IT & Productivity and Financial Operations rather than mainstream B2B SaaS.
The defense takes two minutes: our free Before You Renew check finds your notice deadline and puts a calendar reminder in before the cancel window closes. Check the renewal rate when buying hosting or antivirus, but don't let renewal-paranoia distract from the everyday leaks above.
159 of the 487 researched tools offer a genuinely free plan or version — not a trial, a tier. That's an evaluation strategy: for many jobs you can assemble a zero-cost stack first, prove the workflow, and pay only where the free tier's ceiling actually constrains you. Our own 79 free tools require no signup at all.
The numbers come from our research dossiers on all 487 tools in the corpus: for each, we read the vendor's own pricing page, dated the reading, and wrote an assessment that includes who should skip the tool. Cost-trap flags are derived rule-based from that research text, and every flag carries the sentence it rests on. This page is recalculated from the same data as the Atlas and the open dataset whenever a dossier changes; the date above is the most recent price check in the corpus, not the date every tool was checked.
Price checks across the corpus: 409 of the 487 price notes were re-read in the last 30 days, 78 between 30 and 90 days ago.
What the corpus is, and is not. Tools enter it because readers ask about them, because they sit in the categories we cover, or because we joined their partner programme. It is not a random sample of the SaaS market, so every percentage on this page describes our 487 researched tools, not software pricing in general. A flag is counted when at least one sentence in the dossier supports it; the definitions are on the Atlas, and each flag in the open dataset carries its sentence. Corrections are dated on the Manifesto.
Open the Software Cost Atlas →
Usage metering: 52% of the 487 tools we researched (255 of 487). Calculate the price at your real volume and headcount before trusting a sticker price.
1 in 70 tools (7 of the 487) publishes no usable price. Ask for the all-in figure in writing: base, seats, add-ons, implementation, renewal.
Less common than people fear: 1 in 49 of our corpus (10 of 487), mostly in IT & Productivity and Financial Operations. The everyday leaks are meters, seats and add-ons.
New reviews, price changes we recorded, and the cost conditions we found. No sponsored placements. Unsubscribe anytime.
This report contains no affiliate links in its claims; example links go to our own dossiers, which disclose affiliate relationships individually. Our advice never changes based on commissions. How we review →