500 B2B SaaS Domains · Oct 2026 Study

One product. Conflicting facts across official pages.

In a selected case series, we verified 21 conflicts across 14 vendors. In the uniform screen, 3/500 vendors had a verified conflict. Explore one complete example: source evidence, owner decisions, and draft corrections.

See the evidence and draft fixes →

Open access · No email or login · Download the 1-fact worksheet (.md)

The 500-Domain Dataset at a Glance

52% of our 500-domain sample served /llms.txt. Only 8.2% published /pricing.md.

Every figure below comes from SHA-256-verified public HTTP captures and dated native buyer observations recorded in October 2026.

Sample Frame · n = 500

Screened B2B SaaS domains checked via bounded non-JS HTTP captures.

Uniform Screen · 326 Contexts

Repeated mentions of 1 selected commercial fact across 124 vendors (110 comparable).

AI Buyer Panel · 20 Scorable Runs

Recorded from 21 Google AI Mode & Perplexity checks on our 21 verified conflicts.

Optical prism diagram showing one canonical product truth splitting into diverging amber and red signals across secondary surfaces.
01 · Canonical Plan Table

Pricing pages state tier prices, units, and billing periods side by side.

02 · Secondary Surface Drift

Docs, FAQs, vs. pages, and /llms.txt omit billing scope or present differing figures across pages.

03 · Buyer & AI Tension

Buyers and AI engines encounter conflicting numbers across first- and third-party sources.

Public Files Across 500 SaaS Domains

n = 500 domains
Served /llms.txt Public root discovery file
261 / 500 · 52.2%
Numeric /pricing (non-JS) Paid plan numerals in bounded HTTP capture
253 / 500 · 50.6%
Served /llms-full.txt Extended machine-readable file
56 / 500 · 11.2%
Substantive /pricing.md 40 at /pricing.md · 1 at /.well-known/pricing.md
41 / 500 · 8.2%
Source: BrandAnchor v7.0 Archive (Oct 2026 bounded non-JS HTTP captures). HTTP 200 alone did not qualify—HTML fallbacks were excluded from /pricing.md.

Native AI Answers on Verified Conflicts

20 scorable · 21 total runs
20 scorable runs

  • 13 (65%) · Matched pricing reference

  • 3 (15%) · Repeated conflicting value

    No warning shown (Jungle Scout ×2, Segmently ×1)

  • 3 (15%) · Warned pages disagreed

    Flagged official conflict (Eraser ×2, Segmently ×1)

  • 1 (5%) · Qualified tier scope (Keragon)

Citation mix (all 21 runs): 14 of 21 (67%) included or relied on third-party sites alongside or instead of first-party pages.
Source: BrandAnchor Native UI Panel (Google AI Mode & Perplexity, Oct 2026; 1 of 21 lacked numerals). We separate repeated conflicting values (3) from conflict disclosure (3).

Key Takeaways

Three patterns that matter in practice.

Uniform Single-Fact Screen: 326 Reviewed Source Contexts

124 vendors · 110 comparable · 1 fact per vendor

258 · 79.1%

Aligned with /pricing

Matched numeric value, unit, tier, and billing scope.

63 · 19.3%

Missing scope qualifiers

Scope-ambiguous across 23 vendors (14 ambiguity-only, 9 with additional aligned contexts).

5 · 1.5%

Direct conflicts

Across 3 comparable vendors: Attio, Slite, and Taskade.

In total, 68 of 326 reviewed contexts (20.9% — roughly 1 in 5) introduced scope ambiguity or direct contradiction for one selected commercial fact per vendor.
01

326 Single-Fact Contexts Reviewed

Scope disappears when commercial facts leave the pricing table.

In 63 of 326 reviewed contexts for one selected commercial fact per vendor, secondary pages repeated a price, quota, or trial figure without the billing, unit, or tier conditions needed to interpret it.

Observed example

Slite’s saved /pricing page set its lowest paid annual rate at $10 per user/month, while its Slab comparison page stated “starts at $8/user/month.”

Your check

Attach the exact unit, plan tier, and billing period every time a commercial number appears outside /pricing.

02

21 Verified Conflicts Across 14 Brands

A website—or a single pricing page—can disagree with itself.

Across our multi-fact review, we verified 21 same-scope contradictions across 14 SaaS brands—spanning pricing tables, same-page FAQs, help articles, comparison pages, and /llms.txt.

Observed example

Jungle Scout’s saved Catalyst plan table listed 3 included seats for Brand Owner, while the FAQ on that exact same page listed 10 users (with 10 users also noted in page text).

Your check

Compare your plan table against its own FAQ, help docs, comparison pages, and /llms.txt whenever plans change.

03

20 Scorable Native AI Buyer Checks

Readable Markdown files and AI buyer answers reflect source tension.

Clean /llms.txt or /pricing.md formatting did not prevent cross-page disagreement—and in 20 scorable native AI checks on our verified conflicts, 3 answers repeated a conflicting value and 3 told the buyer that official pages disagreed.

Observed example

For Eraser (3 free AI diagrams on /pricing vs. 5 in docs.eraser.io/plans), both Google AI Mode and Perplexity reported both numbers and flagged the page inconsistency.

Your check

Test real buyer questions in Google AI Mode and Perplexity rather than relying on HTTP 200 or syntax checks.

Start with one commercial fact

Check one fact on your own site.

Download Markdown worksheet →
  1. STEP 01 Pick one buyer fact

    Starting price, free allowance, trial days, included seats, or SSO/RBAC tier.

  2. STEP 02 Compare official URLs

    Copy exact wording from /pricing, FAQs, docs., vs. pages, and /llms.txt.

  3. STEP 03 Classify & confirm owner

    Mark Aligned, Scope unclear, Conflicts, or Not checked / Insufficient evidence.

Filled example (fictional AcmeCloud): Team plan included seats — /pricing lists “5 workspace seats ($49/mo annual)” while Billing FAQ lists “Up to 10 workspace users” → Conflicts (confirm with Product Marketing before updating FAQ & /llms.txt).
Preview all 6 worksheet fields +
Field What to record Illustrative entry (fictional AcmeCloud)
Fact Single commercial claim buyers verify Included seats on Team tier
Product & plan Exact product line and tier name AcmeCloud Core — Team
Units & conditions Seat/workspace unit, currency, annual vs. monthly Base workspace seats; USD annual vs. monthly
Official source URLs Every first-party URL checked (or marked not checked) /pricing table vs. billing FAQ & /llms.txt
Scope status Aligned · Scope unclear · Conflicts · Not checked / Insufficient evidence Conflicts (table lists 5 seats; FAQ states 10)
Authorized fact owner Person who approves canonical replacement copy Product Marketing / Revenue Operations

Six company examples from the 14-vendor case series

Where saved official pages disagreed.

Selected from the 21 verified same-scope conflict pairs across 14 vendors in v7.0-final (cards for Attio and Plain summarize two verified facts each). Every quote matches its SHA-256-hashed October 2026 capture verbatim.

Jungle Scout

junglescout.com
Brand Owner included seats

Saved Pricing Capture

3 included seats in Catalyst plan table (/pricing/catalyst-plans/)

Conflicting Official Surface

“Brand Owner plan includes 10 users” in same-page FAQ, page text, & Help Center

Recorded AI Buyer Check Repeated conflicting value

Google AI Mode repeated 10 included seats across two runs.

Keragon

keragon.com
RBAC plan availability

Saved Pricing Capture

“RBAC, and invoice billing are Enterprise only” (/.well-known/pricing.md)

Conflicting Official Surface

RBAC listed under “every plan with no tier restrictions” (/hipaa-zapier-alternative)

Recorded AI Buyer Check Qualified tier scope

Separated standard role permissions on paid plans from Enterprise granular RBAC.

Open full 6-surface proof →

Eraser

eraser.io
Free AI diagram allowance

Saved Pricing Capture

“3 AI diagrams” on Free tier (/pricing)

Conflicting Official Surface

“5 AI diagrams” on Free tier (docs.eraser.io/plans)

Recorded AI Buyer Check Flagged page conflict

Both Google AI Mode and Perplexity reported 3 vs. 5 and noted page inconsistency.

Attio

attio.com
Plus seat price & Free credits

Saved Pricing Capture

$35 annual / $44 monthly per user; 100 Free credits/mo (/pricing)

Conflicting Official Surface

$29 annual / $36 monthly per seat; 500 credits/mo (/llms.txt & /llms-full.txt)

Recorded AI Buyer Check Matched pricing + 3P mix

Matched /pricing values while citing attio.com alongside third-party blogs.

Plain

plain.com
Trial duration & starting price

Saved Pricing Capture

“7-Day Free Trial”; $45/seat/mo annual floor (/pricing)

Conflicting Official Surface

“14-day free trial” (/llms-full.txt); “$35/mo” first seat (/llms.txt)

Recorded AI Buyer Check Matched pricing reference

Matched 7-day trial and $45 annual / $55 monthly pricing reference.

Taskade

taskade.com
Pro plan monthly price

Saved Pricing Capture

$10/mo equivalent; $120 billed yearly (/pricing)

Conflicting Official Surface

“$20/mo ($16/mo annual)” in official billing FAQ (help.taskade.com)

Recorded AI Buyer Check Matched pricing + 3P mix

Matched $10 annual / $20 monthly reference while citing a third-party review.

Named companies are independent research examples from saved public captures (22–23 October 2026 UTC), not BrandAnchor customers. Live websites and single-session AI answers change over time.

Method & Reference Archive

Methodology, denominators, and study limits (v7.0 archive) +

500-vendor frame & bounded HTTP capture: The study covers 500 unique eligible B2B SaaS domains. Bounded public HTTP discovery (up to 36 requests per vendor, without browser JavaScript execution) classified pricing capture availability as: 253 (50.6%) public numeric paid tiers, 45 (9.0%) public Free/trial terms with opaque paid prices, 43 (8.6%) sales-led or custom pricing only, and 159 (31.8%) with no qualifying pricing surface in the non-JS HTTP archive (which does not mean a live site lacks a JS-rendered pricing page). Across machine-readable paths, 261 of 500 (52.2%) had a saved /llms.txt capture (56 had /llms-full.txt), and 41 of 500 (8.2%) served a content-validated Markdown pricing file (40 at /pricing.md and 1 at /.well-known/pricing.md).

Separate denominators: The uniform single-anchor screen selects one commercial fact per vendor (entry paid base offer; if none is public in capture, core Free allowance; then trial duration). It contains 326 reviewed source contexts for one selected commercial fact per vendor across 124 vendors (110 comparable vendors; 258 aligned, 63 scope ambiguities across 23 vendors: 14 ambiguity-only vendors and 9 with additional aligned contexts, and 5 verified conflicts across 3 vendors: Attio, Slite, and Taskade—3/110 comparable; 3/500 overall). That uniform screen is strictly separate from the selected multi-fact case series (21 verified conflict pairs across 14 vendors). In small reviewed subsets of the uniform screen, Help / Docs showed 1 conflict vendor in 19 comparable (5.3%), Comparison pages 1 in 39 (2.6%), and Machine-readable files 1 in 37 (2.7%)—sample sizes too small to rank one page family across all SaaS sites.

Native AI panel breakdown & non-causality: Across 21 native Google AI Mode and Perplexity observations covering the 21 verified conflict pairs (20 scorable; 1 unscorable missing numerals), 3 repeated a conflicting value, 3 reported multiple values / noted official-page disagreement, 1 separated standard vs. Enterprise role scope, and 13 matched the selected pricing reference. Citation mix across all 21 observations: 7 first-party only, 12 mixed first- and third-party, and 2 third-party only (14 of 21 included third-party citations). Single-session UI observations do not establish model stability or causal proof that a specific page generated an answer.

Archive integrity vs. live product truth: All 663 referenced capture files and all 7 benchmark tables in publication-v7.0-final passed SHA-256 hash and literal-substring checks with 0 errors. Hash verification proves the saved excerpt matches the archived HTTP capture; it does not establish chronological edit history or certify current internal billing behavior without brand-owner confirmation. The 500-domain cohort is a screened convenience frame, not a random sample of all B2B SaaS companies.

Six public /pricing.md implementations worth studying (41 of 500 subset) +

Only 41 of 500 vendors (8.2%) served a substantive Markdown pricing file (40 at /pricing.md and 1 at /.well-known/pricing.md). These six public files use clear units and tier tables—though as Keragon shows, clean file formatting still requires cross-page consistency checks.

Beehiiv beehiiv.com/pricing.md ↗

YAML frontmatter, Markdown tier tables, explicit billing cadences, and free/trial bounds.

DailyBot dailybot.com/pricing.md ↗

Explicit updated/synced date, annual vs. monthly qualifiers, and alignment with /pricing.

LiveChat livechat.com/pricing.md ↗

Dated pricing snapshot, per-agent billing cadences, trial terms, and tier limits.

PostHog posthog.com/pricing.md ↗

Usage-based tiers and monthly free allowances expressed with explicit unit thresholds.

Supabase supabase.com/pricing.md ↗

Base plan rates paired with exact compute, storage, MAU inclusions, and overage units.

Clerk clerk.com/pricing.md ↗

Clear plan headings, MAU quotas, add-on pricing, and alignment with its main pricing page.

Pricing & Plan Consistency Check · $99 One-Time

Skip the manual hunt across your pages.

Built for GEO agencies onboarding a paying SaaS client and PMMs after a pricing or plan update. Receive a dated 5-URL source comparison, 2 clean US buyer observations on Google AI Mode & Perplexity, and ready-to-paste corrections within two business days.

$99 one-time · 1 English B2B SaaS product · Up to 5 official URLs (including docs.) · 3 commercial facts · 100% refund if out-of-scope or technically blocked.

Order the $99 check → Or inspect the open example first →

Need a wider multi-surface audit (up to 12 official URLs & 6 core facts)? See the $390 extended option →

Optical lens aligning scattered amber rays into one unified electric citron and emerald beam.

Included in your 2-business-day report

3-Fact Official URL Matrix (up to 5 URLs) ✓ Included
2 Clean US AI Checks (Google AI Mode & Perplexity) ✓ Included
Affirmative Facts Block + Draft Fixes ✓ Included

Research by Yair Zilberman, Founder at BrandAnchor.io · Questions: hello@brandanchor.io