How B2B SaaS proves it worked — the case-study index
What does a convincing case study look like? We collected 821 customer stories from 226 B2B software companies and measured how each one presents its results — the headline format, where the number lands, how long the story runs — so you can benchmark your own proof program against the cohort.
Executive summary
Customer proof is a scale program. 271 of 479 companies (56.6%) maintain a story hub, but the rate runs from 33.3% at 1–10 employees to 90.2% at 200+ — the steepest size ladder in this research program. When companies do publish stories, the formula is strikingly uniform: a percentage headline (42.0%, 60.5% of stories that carry any metric), placed a median 7.8% down the page text, in a 921-word story, with video 21.4% of the time and a stated time-to-result 11.3% of the time.
Two caveats shape how to read this. First, 251 of 821 stories (30.6%) carry no quantified result at all — nearly a third of "case studies" are testimonials with logos, not evidence. Second, these are self-reported marketing claims: the study describes presentation patterns, never ROI, and never ranks vendors by how inflated their numbers look.
Metric types: the percentage wins
| First result type | Stories | Share | 95% CI |
|---|---|---|---|
| Percentage claim | 345 | 42.0% | 38.7-45.4% |
| Multiplier (Nx) | 94 | 11.4% | 9.4-13.8% |
| Currency amount | 85 | 10.4% | 8.5-12.6% |
| Count / volume | 32 | 3.9% | 2.8-5.5% |
| Time saved | 14 | 1.7% | 1.0-2.8% |
| Qualitative only | 251 | 30.6% | 27.5-33.8% |
Result depth: front-loaded, with a long tail
The median first metric sits 7.8% down the main page text — hero-adjacent. But 168 of 570 metric-bearing stories (29.5%) place it past the first quarter of the text.
Shape of a story
Median length 921 words (P25 594, P75 1324). Video embedded on 21.4% of stories. Time-to-result stated on 11.3% — vendors show the win far more often than the wait. Named customers dominate slugs; anonymous framing ("a leading fintech", "anonymous") appears on 2.3% (slug/title lower bound).
Who runs a story hub
| Team size | Hub rate | 95% CI |
|---|---|---|
| 1-10 employees (n=129) | 33.3% | 25.8-41.8% |
| 11-50 employees (n=190) | 56.3% | 49.2-63.2% |
| 51-200 employees (n=111) | 73.0% | 64.0-80.4% |
| 200+ employees (n=41) | 90.2% | 77.5-96.1% |
By billing model
Matched against our pricing-transparency study of the same companies.
| Billing model | Hub rate | With stories |
|---|---|---|
| Usage-based (n=50) | 72.0% | 64.0% |
| Per-seat (n=63) | 68.3% | 60.3% |
| Flat / tiered (n=117) | 55.6% | 49.6% |
| Contact-only (n=115) | 67.0% | 53.9% |
| Page, no numbers (n=40) | 22.5% | 10.0% |
Usage-based vendors lead (72.0% hubs, 64.0% with stories fetched); pages with numbers but no tiers trail (22.5% / 10.0%). Contact-only enterprise vendors run hubs at 67.0% — proof matters where sales cycles are longest.
How big are the claimed wins
The median percentage claim is 70% — and the most common bucket is 51–100%. Read every number below as marketing self-report, never as ROI.
| Claimed improvement | Stories | Share of % claims |
|---|---|---|
| Over 100% | 22 | 6.4% |
| 51-100% | 180 | 52.2% |
| 31-50% | 73 | 21.2% |
| 11-30% | 54 | 15.7% |
| 10% or less | 16 | 4.6% |
Multipliers and money
Among 94 multiplier headlines the median is 6x, led by 10x (17 stories), 3x and 2x. Currency headlines cluster in the millions — and the single most repeated figure in the whole sample is $38M, appearing on 5 stories.
| Top multipliers | Value | Stories |
|---|---|---|
| 10x | 17 | |
| 3x | 14 | |
| 2x | 11 | |
| 5x | 8 | |
| 4x | 7 |
| Top currency figures | Figure | Stories |
|---|---|---|
| $38M | 5 | |
| $1 billion | 3 | |
| $750M | 3 | |
| $40M | 3 | |
| $11M | 3 | |
| $2M | 2 |
Marketing quantifies; analytics does not
Headline format by vendor sector — the irony table. Sectors are the vendors' own YC classifications, n≥40.
| Vendor sector | % headlined | Qualitative-only |
|---|---|---|
| Engineering, Product and Design (n=189) | 42.9% | 35.4% |
| Infrastructure (n=110) | 39.1% | 28.2% |
| B2B (n=91) | 47.3% | 26.4% |
| Productivity (n=53) | 41.5% | 35.8% |
| Sales (n=48) | 52.1% | 14.6% |
| Operations (n=44) | 34.1% | 20.5% |
Marketing software headlines percentages (58%) and almost never runs qualitative-only (9%). Analytics (25% / 50%) and Finance (18% / 40%) — the sectors that sell measurement — quantify themselves least. If you sell analytics, this gap is your opening: a quantified story stands out precisely because your competitors do not publish one.
Video is a scale good; timelines are nobody's habit
| Team size | Video | Time-to-result |
|---|---|---|
| 1-10 employees (n=111) | 6.3% | 9.0% |
| 11-50 employees (n=319) | 22.6% | 9.7% |
| 51-200 employees (n=294) | 22.1% | 13.3% |
| 200+ employees (n=94) | 34.0% | 13.8% |
Result depth does not depend on format: median first-metric position is 7.3% for percentages, 6.5% for multipliers and 5.1% for currency. Whatever the number, it sits in the same place.
Anonymous stories quantify anyway
19 stories hide the customer name ("a leading fintech", anonymous slugs). They quantify at 78.9% against 69.2% for named stories — anonymity costs nothing in numbers (small sample, descriptive only).
If a customer will not lend its name, take the anonymous story: the cohort's anonymous pages carry metrics at least as often as named ones. What anonymity does cost is video (11% vs 22%) — faceless stories stay text.
Proof and attack travel together (weakly)
Matched against our comparison-page study: 77 story-publishing companies also publish comparison pages, averaging 4.8 stories each against 4.3 for non-publishers.
Companies that argue against rivals also document their own wins — slightly more of them, slightly more often. Proof libraries and comparison libraries are the twin surfaces of scaling competitive content, but the gap is small: story volume is driven far more by size than by competitive posture.
Who this is for
Copy the shape, not the numbers
The cohort's winning shape: a percentage headline in the first screen, ~900 words, video optional. If your stories run long before the first number, you are in the 29.5% that bury it — move one quantified result above the first scroll.
Five stories beat fifty ambitions
We capped this study at five stories per company because five maintained stories is what the median publisher actually sustains. Start with five customers and one number each; a hub grows by maintenance, not by launch-day volume.
Read past the headline percentage
30.6% of stories carry no quantified result at all, and only 11.3% state how long the win took. Ask for the denominator and the timeline before you let a logo do the arguing.
Presentation data, not ROI data
These are self-reported marketing claims coded for format and placement. Do not compare values across customers as returns — cite this study for the shape of proof, never its substance.
Methodology, deviations & reproducibility
What we read: for each company we found its customer-story hub (usually /customers or /case-studies) and read up to five stories. 226 companies yielded 821 stories from 2,423 discovered links. Pages that blocked our crawler were left out, never worked around.
How results were measured: the first number on each page, its format (percentage, multiplier, currency, volume, time saved), and how far down the page it sits. Headcounts ("5,000 employees") were excluded — they describe the customer, not the result. A 20-story spot-check caught and fixed two measurement errors before publication.
What changed from the plan: we wrote the rules before collecting and logged five deviations. (1) "Below the fold" is measured as depth in the page text, not a rendered screenshot. (2) No page-history comparison — archived pages mislead too often. (3) The enterprise-customer comparison was dropped: customer sizes could not be read reliably at scale. (4) No customer logos republished; no vendor ranked by how big its claims look. (5) One reviewer so far; no independent second check yet.
Downloadable datasets: metrics.csv (metric-type counts) and companies-case-studies.csv (per-domain hub and story counts).
python3 scripts/research/s4_discover.py # hub + story discovery (resumable)
python3 scripts/research/s4_fetch_extract.py # fetch up to 5 stories/domain
python3 scripts/research/build_s4_page.py # rebuild this page
Disclosure and corrections
DataDab sells AI-visibility work and paid inclusion in its research pages. A research mention costs $400–$1,200. Payment does not change a figure in this study. DataDab is not part of the measured set. Corrections are published on the page; send factual corrections with the URL to amit@datadab.com.
Covering this research? Please link the primary source so readers can verify the data and methodology:
Markdown: [The B2B SaaS Case-Study Index 2026 by DataDab](https://www.datadab.com/research/b2b-saas-case-study-index/)
Disclosure: DataDab sells AI-visibility work and paid inclusion in its research pages. A research mention costs $400-$1,200. Payment does not change a figure in this study. DataDab is not part of the measured set. Corrections are published on the page (amit@datadab.com).
What this study does not show
- ROI. Self-reported marketing claims describe presentation, never return on investment.
- Vendor rankings by claim size. Deliberately never computed.
- Unlinked and gated stories. PDF-only, login-gated, and JS-only story sets are missed; 253 domains yielded no fetchable story.
- Single reviewer. Coded with a 20-story evidence check; no independent second check yet.