Software Pricing Findings, 2026 Q3 — CostBench
Software Pricing Findings, 2026 Q3 — CostBench
Shortlist
Findings · 2026 Q3

Five things the software pricing corpus says right now

Five findings from CostBench's own pricing corpus — 3,297 software products across 255 categories, read from the vendors' own pricing pages.

Products
3,297
Categories
255
Publishing a price
1,925
As of

In short

  1. The median software product starts at $25 a month.
  2. 31.4% of software products publish no price at all.
  3. Only 38% of software products offer a free tier.
  4. Annual billing discounts pile up on two numbers: 20% and 17%.
  5. 75% of apparent SaaS price changes are measurement artifacts.
Finding 1

The median software product starts at $25 a month.

$25/month

Across 1,925 products that publish a price, the cheapest paid plan has a median list price of $25 per month. Half of the market sits between $10 and $60; the interquartile spread is 6.0x.

Scope
1,925 of 3,297 tracked products — those publishing at least one priced plan
Method
Entry price = the lowest positive monthly plan price on the vendor's own public pricing page. Annual-only prices are normalised to a monthly figure by the extractor.
Limitations
List prices, not negotiated prices. Enterprise buyers routinely pay less per seat than the published entry plan implies, and quote-only products are excluded entirely.
As of
Finding 2

31.4% of software products publish no price at all.

31.4% quote-only

1,036 of 3,297 tracked products list plans but no number on any of them — every tier is "contact us". Opacity is not evenly spread: it runs to 81% in SIEM while 11 of the 96 categories we can measure publish a price for every single product.

Most quote-only categories
CategoryProductsShare
SIEM 13 of 16 81%
Email Security 12 of 16 75%
Procurement Software 12 of 16 75%
AI Legal Tools 14 of 19 74%
Learning Management System 12 of 17 71%
Scope
all 3,297 tracked products; the category breakdown covers the 96 categories with at least 15 products
Method
A product counts as quote-only when it has plans, none of them carries a public price, and every plan is flagged custom-priced. Read directly from the vendor's pricing page — this is an observation, not an estimate.
Limitations
Measures what the vendor publishes, not whether a price exists. A quote-only vendor still has a rate card; it is simply not public.
As of
Finding 3

Only 38% of software products offer a free tier.

38.0% with a free tier

1,254 of 3,297 tracked products offer a free plan or a free trial. The spread between categories is the story: 96% in Developer Tools against 0% in Nonprofit CRM. Free-to-start is a norm where a single practitioner can adopt the tool alone, and close to absent where the buyer is a committee.

Widest free-tier gap, top three against bottom three
CategoryProductsShare
Developer Tools 27 of 28 96%
AI Image Generators 18 of 21 86%
Project Management 24 of 28 86%
Nonprofit CRM 0 of 18 0%
Email Security 0 of 16 0%
Legal Practice Management 1 of 21 5%
Scope
all 3,297 tracked products; the category breakdown covers the 96 categories with at least 15 products
Method
Counts any product whose pricing page advertises a free plan or a free trial, as recorded in the free-tier field of its record.
Limitations
A free plan and a time-limited trial are different products commercially; this figure combines both. Where the distinction matters, the per-product records separate them.
As of
Finding 4

Annual billing discounts pile up on two numbers: 20% and 17%.

20% or 17% off for paying annually

Of 585 products that state an annual-billing discount, 125 land on exactly 20% and 104 on exactly 17% — 39% of them on just two values. They encode two different conventions: 20% is a round number, while 17% is what "two months free" works out to (paying for ten months of twelve is a 17% discount). The median is 20%.

Stated annual discount, products per value (values with at least 3 products)
DiscountProductsn
5% 6
8% 8
10% 33
11% 9
14% 12
15% 21
16% 17
17% 104
18% 8
19% 9
20% 125
21% 10
22% 9
25% 38
26% 8
29% 11
30% 13
33% 24
34% 10
40% 11
45% 5
50% 15
Scope
585 of 3,297 tracked products — those that publish an annual-billing discount
Method
Percentage parsed from the vendor's own stated annual-billing discount. Products that offer no annual discount, or state it only as a currency amount, are excluded from the denominator.
Limitations
Self-reported by the vendor. A stated discount is measured against the vendor's own monthly list price, which the vendor also sets.
As of
Finding 5

75% of apparent SaaS price changes are measurement artifacts.

12 of 16 candidate price changes were false

We diffed same-named plans across quarterly snapshots, took every candidate move that survived our automated guards, and hand-checked all 16 against the vendor's own record. 12 turned out to be artifacts — promotional price read as list, annual rate compared against monthly, a plan renamed or restructured, regional price drift, per-seat versus per-account. 2 could not be resolved either way. Only 2 were real repricings. This is why CostBench publishes no price-trend series.

What the hand-check found (sample)
Product / planVerdictWhy
coderabbit — pro Real per user, annual billing Evidence
aws — enterprise support Real minimum monthly fee; 2026 support-plan consolidation Evidence
flyio — support: premium Artifact SKU mislabel: $99 is the HIPAA/SOC2 compliance add-on, not Premium support ($199). No same-plan move. Evidence
barracuda-email-protection — advanced Artifact Billing-term/reseller SKU: $2.66/user/mo annual remains current list; $5 is a 1-month-term SKU. Evidence
fresha — team Artifact Region drift: $9.95-$14.95 varies by geography, not a same-region temporal increase. Evidence
tana — pro Artifact Annual-vs-monthly: $18-20 is the annual rate, $30 the monthly rate. Billing-term normalization. Evidence
Scope
all 16 candidate same-plan list-price changes surfaced in the 2026 H1 window and sent to hand verification
Method
Each candidate was checked against the vendor's live pricing page, changelog, or official documentation by a human, with the verdict and evidence URL recorded in a permanent ledger so it is never re-litigated.
Limitations
A hand-checked sample, not a census of the whole market. It bounds the error rate of snapshot-diff price tracking; it does not measure how often prices genuinely move.
As of

Use these numbers

All five findings are computed from data you can download and re-run yourself. The corpus is published as CSV with the vendor source URL and verification date on every row.

Licence: CC BY-NC 4.0. Attribution required — credit CostBench and link to https://costbench.com/. Non-commercial reuse of the dataset; commercial redistribution rights: https://costbench.com/api-docs/.
Quoting or citing individual figures from this dataset — in articles, reports, decks, or AI assistant answers, commercial or not — is always permitted with attribution to CostBench.

Download the dataset How the data is collected The full Pricing Index

Published quarterly. Current-state figures are recomputed each edition; movement claims are published only after hand verification against the vendor's own record, which is why you will not find a price-trend chart here.

Pricing Alerts

Get the weekly SaaS Price Watch

SaaS prices change constantly — we verified 126 changes last month. Get the weekly Price Watch digest of verified changes and deals. One-click unsubscribe.