FrontDeskReview · Data study · Updated September 2026

The Stale List Report

Most apparent price errors are not errors. Across 12 software categories we read every price that 45 top-ranking third-party buying guides state for a vendor we carry — 216 figures in all. 82 disagreed with our own recorded figure on a first read, but after independent re-reads only 12 — 5.6% of every price quoted — were genuinely out of date. The other 70 were the ordinary traps of reading a pricing page: a different plan, a monthly headline set against an annual-billed rate, a promotional figure, a line the page no longer carries. In the same quarter we corrected 271 figures of our own, found by auditing every plan whose recorded source was not the vendor's own page. Telling a stale price from a correctly-read one takes opening the vendor's page on the same billing basis — which is the whole finding.

12 of 216 quoted prices were out of date. Of the 25 rows a first pass flagged, 13 did not survive the independent re-reads that followed — 3 of them because the quotation stored for the row was not the text on the page. Those are control rows 12, 23 and 24: every control row filed under “The quoted sentence is not the sentence on the page”, plus any row whose own note calls the stored quotation a reconstruction.

Every row re-checked 2026-09-06 · 82 independent reads

5.6% of the 216 quoted prices were confirmed out of date
13 of 25 first-pass stale calls knocked out by further independent reads
271 of our own figures corrected in the same quarter

How 216 quoted prices became 12

Each step is a narrower claim than the one above it. The distance between step two and step four is the study: a disagreement is not an error, and a first look is not a finding.

Prices quoted for a vendor we carryacross 45 readable articles in 12 categories
216
Disagreed with our figure on a first reada disagreement, not yet an error
82
Looked out of date to an independent checkereach one re-read live against the vendor's own page
25
Still out of date, and the quotation found in the page a reader sees13 fell away — they are the control below
12

The 12 figures that were out of date

Ordered by the size of the gap between the article's figure and the vendor's figure today, as a share of the figure quoted. 9 publishers, 10 vendors. Every cell is checkable: the publisher links to the article, the vendor to our sourced pricing page, and the page's own stated date is reproduced verbatim — where a page shows no date, the table says so.

Confirmed out-of-date quotes, biggest gap first · re-checked 2026-09-06
# Publisher & the date the page states Vendor Article's figure Vendor's figure today Our figure Gap
1 learn.g2.com January 7, 2026 Respond.io Whatsapp Business $29 $79 Starter (cheapest paid plan; next tiers Growth $159/mo and Advanced $279/mo annual, then quote-only Enterprise) $79 Starter · read 2026-09-03 172.4% vendor now higher · 3 of 3 reads agreed
2 softwareadvice.com Updated September 3, 2026 Copper CRM CRM Software $12 $23 Basic $23 Basic · read 2026-09-02 91.7% vendor now higher · 3 of 3 reads agreed
3 prismic.io Published January 2, 2026 Unbounce Landing Page Builders $99 $29 Starter (lowest paid tier). The $99/month figure under test is today the second tier, "Build". Unbounce has no plan named "Launch" any more -- the catalog's plan name is stale even though its $29 figure is not. $29 Starter · read 2026-09-02 70.7% vendor now lower · 3 of 3 reads agreed
4 guideflow.com August 31, 2026 Zoko Whatsapp Business $34.99 $49.99 STARTER (rendered as "STARTER" on the pricing card; named "Starter" in the page's own JSON-LD). It is the cheapest of Zoko's four plans and therefore the entry plan. $49.99 Starter · read 2026-09-03 42.9% vendor now higher · 3 of 3 reads agreed
5 techradar.com 12 January 2026 Pipedrive CRM Software $10 $14 Lite $14 Lite · read 2026-09-02 40% vendor now higher · 4 of 4 reads agreed
6 guideflow.com August 31, 2026 WATI Whatsapp Business $49 $59 Growth (entry paid tier; tiers today are Growth / Pro / Business, with Growth the cheapest) $59 Growth · read 2026-09-03 20.4% vendor now higher · 3 of 3 reads agreed
7 learn.g2.com January 7, 2026 WATI Whatsapp Business $49 $59 Growth - Wati's entry paid plan (the tiers are Growth / Pro / Business; there is no cheaper tier) $59 Growth · read 2026-09-03 20.4% vendor now higher · 3 of 4 reads agreed
8 mailtrap.io On November 12, 2025 Google Workspace Email Hosting $6 $7 Business Starter (card labelled "Starter"; the page's own footnote calls it "Business Starter") $7 Business Starter · read 2026-09-02 16.7% vendor now higher · 3 of 5 reads agreed
9 serveravatar.com Published: 16 August 2025 Last Updated: 26 February 2026 Vultr Vps Hosting $6 $5 Cloud Compute - Regular Performance, 1 vCPU / 1 GB (plan id vc2-1c-1gb): 1 vCPU, 1 GB memory, 25 GB SSD, 1.00 TB bandwidth $5 Cloud Compute Regular — 1 GB · read 2026-09-03 16.7% vendor now lower · 4 of 5 reads agreed
10 mailtrap.io On November 12, 2025 Microsoft 365 for Business Email Hosting $6 $7 Microsoft 365 Business Basic $7 Business Basic · read 2026-09-02 16.7% vendor now higher · 3 of 3 reads agreed
11 techrepublic.com Aug 17, 2026 Microsoft 365 for Business Email Hosting $6 $7 Microsoft 365 Business Basic $7 Business Basic · read 2026-09-02 16.7% vendor now higher · 3 of 3 reads agreed
12 domaindetails.com no date shown to a reader Porkbun Domain Registrars $10.99 $11.08 .com domain - Porkbun standard published rate (registration, renewal and transfer are one and the same price; Porkbun has no tiers or plans for this) $11.08 .com (per year) · read 2026-09-02 0.8% vendor now higher · 3 of 3 reads agreed

The evidence behind each row

For every row above: the article's own sentence carrying the figure, as that sentence stands in the text a reader sees on the page, and when and where the vendor's figure was verified — with the billing basis the comparison was made on. Numbered to match the table.

Each quotation was re-read against the live article on 2026-09-06 and matched against one canonical reading of that page — the text a reader sees, set out in full under Method — and the count printed beside each row is the count that reading produces. A quotation that reading could not find was not rewritten to fit: 3 rows were moved to the control below on the quotation — control rows 12, 23 and 24: every control row filed under “The quoted sentence is not the sentence on the page”, plus any row whose own note calls the stored quotation a reconstruction. Each quotation is published whole, its length stated beside it. Nothing on this page is cut.

  1. #1 Respond.io · Whatsapp Business · $29 quoted vs $79 today

    The article says, verbatim (217 characters, in full) respond.io: Best shared inbox for WhatsApp and multi-channel support Centralizes customer conversations from WhatsApp and other channels with workflow automation, routing, and team collaboration. (Pricing from $29/mo)

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://respond.io/pricing.

    Compared onUSD per month in the page's default view, billed yearly at $948/yr. Billed monthly the same Starter plan is $99/month. Seat- and contact-metered on top: additional users from $12/month and Monthly Active Contact tiers priced separately; WhatsApp fees charged by Meta are not included.

    Baseline exceptionThe vendor figure is the rate respond.io's pricing page opens on, $79 a month billed yearly; our catalogue holds the month-to-month rate, $99, read on 2026-09-27. The publisher's $29 is below both.

  2. #2 Copper CRM · CRM Software · $12 quoted vs $23 today

    The article says, verbatim (301 characters, in full) Copper4.4 (622 reviews)Learn MoreCompareLearn MoreCompareAverage score:80/100Average Score:80/100Pros: seamless gmail and google workspace integration intuitive and user-friendly interfaceCons: poor customer support responsiveness insufficient customization optionsPricing:Starting at $12.00 per month

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://www.copper.com/pricing.

    Compared onUSD, per seat, per month, paid annually (Copper's default-displayed Annual tab; $29 per seat, per month on month-to-month billing, shown struck through on the annual card). Excludes taxes and fees. Copper's entry tier is named "Basic"; there is no plan called "Starter".

    Baseline exceptionThe vendor figure is the rate Copper's pricing page opens on, $23 per seat per month paid annually; our catalogue holds the month-to-month rate, $29. The publisher's $12 is below both.

  3. #3 Unbounce · Landing Page Builders · $99 quoted vs $29 today

    The article says, verbatim (68 characters, in full) Pricing starts at $99/month, which is higher than some alternatives.

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://unbounce.com/pricing/.

    Compared onLowest paid tier (Starter), month-to-month billing, USD, list price, per month, excluding tax. The equivalent annual-billing rate is $22/month ("Save 25% with yearly billing"). Like-for-like on month-to-month billing, Unbounce's entry price is $29/month; $99/month is the month-to-month price of the Build tier, one step up.

  4. #4 Zoko · Whatsapp Business · $34.99 quoted vs $49.99 today

    The article says, verbatim (32 characters, in full) Pricing: Starts at $34.99/month.

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://www.zoko.io/pricing.

    Compared onStanding monthly list price of the entry (Starter) plan, in USD, quoted by the vendor as "$49.99/mo". Per-conversation charges ($0.015/conversation on Starter) and platform fees (e.g. Instagram $9.99/month) are billed on top and are not part of this figure. No annual/monthly billing toggle exists on the page, so this is the only published price for the plan. Same basis as the article's "Starts at $34.99/month": entry-plan monthly subscription fee.

  5. #5 Pipedrive · CRM Software · $10 quoted vs $14 today

    The article says, verbatim (177 characters, in full) Besides, Pipedrive is one of the most affordable CRM tools, starting at just $10 per user per month, compared to a rival like Salesforce which starts at $25 for the same period.

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://web.archive.org/web/20260827062707/https://www.pipedrive.com/en/pricing.

    Compared onUSD per seat per month on the Lite plan billed annually (code lite_annual_v2; CA$ 19 / one payment of CA$ 228 per seat/year as rendered to the Canada-geolocated crawler). Month-to-month Lite is USD 24 per seat per month (lite_monthly_v2). Figure as of the 2026-08-27 capture, the most recent available; the live page is Cloudflare-blocked.

  6. #6 WATI · Whatsapp Business · $49 quoted vs $59 today

    The article says, verbatim (49 characters, in full) Pricing: Starts at $49/month for the Growth plan.

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://live-mt-server.wati.io/wati/api/v1/pricing/getPlansByGeo?countryCode=US.

    Compared onUSD, Growth plan, per month billed annually ($708.00/year, i.e. $59.00/mo) - the figure the pricing page shows by default, since the annual toggle is the default view. Month-to-month for the same plan is $69.00/mo, and quarterly is $207.00 per 3 months ($69.00/mo). List price, excluding the per-message WhatsApp conversation charges Wati bills separately on top of the plan for every tier.

    Baseline exceptionMeasured against Growth's annual rate, $59.00/month, because wati.io/pricing opens on its Annual tab and that is the figure a reader arriving from the article sees. Our catalogue records the month-to-month rate, $69.00, which is the basis monthlyUsd holds. WATI's own geo endpoint returns both, in cents: 70800 for a year, 6900 for a month. 14% apart.

  7. #7 WATI · Whatsapp Business · $49 quoted vs $59 today

    The article says, verbatim (215 characters, in full) Wati: Best WhatsApp marketing tool for SMB automation Offers easy WhatsApp Business API access, automated replies, broadcasts, and team support tools designed for small and midsize businesses. (Pricing from ~$49/mo)

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://live-mt-server.wati.io/wati/api/v1/pricing/getPlansByGeo?countryCode=US.

    Compared onUSD per month, entry plan, billed annually: $59/month ($708/year, Growth-Plan-USD-Yearly-WEST). Billed monthly the same plan is $69/month (Growth-Plan-USD-Monthly-WEST, 6900 minor units); quarterly is $207 per quarter, also $69/month. The pricing page opens on the Annual tab, so $59 is the figure a visitor sees first. Excludes WhatsApp per-message charges, which Wati bills on top ('Additional charges apply for messages').

    Baseline exceptionMeasured against Growth's annual rate, $59.00/month, because wati.io/pricing opens on its Annual tab and that is the figure a reader arriving from the article sees. Our catalogue records the month-to-month rate, $69.00, which is the basis monthlyUsd holds. WATI's own geo endpoint returns both, in cents: 70800 for a year, 6900 for a month. 14% apart.

  8. #8 Google Workspace · Email Hosting · $6 quoted vs $7 today

    The article says, verbatim (321 characters, in full) Google Workspace offers a 14-day trial for new users, allowing up to 10 team members. The paid plan starts at $6 per person and includes 30 GB of cloud storage per person and up to 30 alias addresses. The Enterprise plan, which provides 5 TB of storage per user with the option to request more, is available upon request.

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://web.archive.org/web/20260903022406/https://workspace.google.com/pricing.

    Compared onUSD, per user / month, Annual plan with a 1 year commitment (page toggle: "Annual (Save 16% with 1 year commitment)"). This is the cheapest standing entry rate; Flexible month-to-month billing is higher. A 30%-off-for-3-months promotion ($4.90, Sep 16 to Dec 16, 2026) was running at time of check and is excluded from the standing figure.

    Baseline exceptionMeasured against Business Starter's annual rate, $7.00/user/month, because workspace.google.com/pricing opens on its Annual toggle and that is the figure a reader arriving from the article sees. Our catalogue records the Flexible month-to-month rate, $8.40, which is the basis monthlyUsd holds. Same plan, the two terms Google offers, 17% apart.

  9. #9 Vultr · Vps Hosting · $6 quoted vs $5 today

    The article says, verbatim (711 characters, in full) 3. Vultr Vultr is known for having a simple pricing system and servers all around the world. It has more than 30 data centers throughout the world, so it’s great for projects that demand low-latency performance for a global audience. Benefits: Hourly or monthly billing options High-frequency compute instances for faster performance One-click app deployment (WordPress, Docker, etc.) Automatic backups and snapshots Why It’s Popular:Developers like Vultr because they can pay for it in a lot of different ways and choose from a lot of server locations. It works well for businesses because it is reliable and easy to grow. Pricing (2025): Starts at $6/month for 1 vCPU, 1 GB RAM, 25 GB SSD, and 1 TB bandwidth.

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://api.vultr.com/v2/plans?type=vc2&per_page=100.

    Compared onMonthly on-demand list price in USD, per Vultr's public pricing table and public plans API (monthly_cost 5, hourly_cost 0.007, invoice_type "monthly", deploy_ondemand true). Standard rate, not promotional; identical preemptible price. Regional variation exists per Vultr's own FAQ - $5.00/mo in 30 of 31 listed regions, Sao Paulo $7.50/mo.

  10. #10 Microsoft 365 for Business · Email Hosting · $6 quoted vs $7 today

    The article says, verbatim (134 characters, in full) The starter plan costs $6 per user/month, offering 50 GB of email storage, 1 TB of OneDrive storage, and access to the suite of tools.

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://www.microsoft.com/en-us/microsoft-365/business/compare-all-microsoft-365-business-products.

    Compared onUSD, en-us storefront; $7.00 per user/month, paid yearly, on an annual subscription that auto-renews. Rendered from class 'oc-displayListPrice' (list price) with unit 'user/month, paid yearly'; no MSRP strikethrough and no discount block on this SKU. The page prints no monthly-billing alternative — all 13 unit spans read 'user/month, paid yearly'. Same per-user-per-month annual-commitment basis as the article's $6.

  11. #11 Microsoft 365 for Business · Email Hosting · $6 quoted vs $7 today

    The article says, verbatim (106 characters, in full) Microsoft 365 Business Basic: $7.20 per user per month or $6 per user per month with a one-year commitment

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://www.microsoft.com/en-us/microsoft-365/business/microsoft-365-plans-and-pricing.

    Compared onUSD per user/month on an annual commitment. The card states '$7.00 / user/month, paid yearly / (Annual subscription–auto renews)'. Every plan card on the page is quoted on this same paid-yearly basis; no month-to-month figure is displayed. This is the same basis the article uses for its $6 ('with a one-year commitment' / 'billed annually'), so the comparison is like-for-like.

  12. #12 Porkbun · Domain Registrars · $10.99 quoted vs $11.08 today

    The article says, verbatim (65 characters, in full) The $10.99 price is consistent for both registration and renewal.

    Found once in the page a reader sees, as fetched on 2026-09-06 — the reading is defined under Method, and it is the only reading this study counts on.

    The vendor's figure Verified on 2026-09-06 from https://porkbun.com/products/domains.

    Compared onUS dollars per domain-year for .com, list price before any coupon. Registration $11.08, renewal $11.08, transfer $11.08 - the same basis the article uses in its Year 1 / Renewals columns, so the comparison is like for like.

The control: 13 rows that a first look called stale and further reads did not

These are not the study's failures — they are its control, and the reason the headline number is 12 rather than 25. Every one was flagged by an independent checker reading the article live, and every one was knocked out by the reads that followed — most when a later reader opened the vendor's page on the same billing basis, and 3 when the quotation stored for the row turned out not to be the text on the page — control rows 12, 23 and 24: every control row filed under “The quoted sentence is not the sentence on the page”, plus any row whose own note calls the stored quotation a reconstruction. None of them was carelessness by a checker; each is one of the ordinary traps of reading a pricing page, which is exactly why a single read is not enough. No publisher is named here: these rows record how a price can be misread, not anything about a page.

Each row carries the checker's own note as recorded. 3 of the 13 named the site they had read, so every publisher name in these notes is replaced with [publisher redacted], and 11 of the 13 are stored in the record at exactly 260 characters and so stop where the record stops rather than where a sentence ends — each note's own length is printed beside it, and the other 2 are published whole. Nothing else about them is altered, and nothing is trimmed to look complete.

The quoted figure is no longer on the page 3

The figure the first pass read is not there on a later read. An article can be corrected at any time, so a quote that has moved on is not evidence of anything.

  • row 5 MXroute · figure read as $4.08 · 4 independent reads · note stored at 260 characters and stops where the record stops · 269 characters published KNOCKED OUT on the quotation, not on the economics. The string "$4.08/mo" does not appear on the [publisher redacted] article. I fetched the raw HTML (436,132 bytes) and grepped: "$4.08" occurs 0 times, and the entity form "$4.08" 0 times. The bare token "4.08" occ
  • row 13 Bitdefender · figure read as $2.92 · 2 independent reads · note stored at 260 characters and stops where the record stops · 237 characters published The article at [publisher redacted] (Last updated Sep 4, 2026) no longer contains the string "2.92" anywhere. Two independent fetches of the page with differently-worded extraction prompts confirm it. What the page states for Bitdefender
  • row 14 Bitdefender · figure read as $2.92 · 2 independent reads · note stored at 260 characters and stops where the record stops · 260 characters published Article fetched today (HTTP 200, 2.49 MB; JSON-LD dateModified "2026-07-17T15:29:02+00:00", matching the stated "Last updated 17 July 2026"). The string "2.92" occurs 9 times in the raw HTML and every single occurrence is inside an SVG path coordinate or minif

The quoted sentence is not the sentence on the page 2

The figure may be on the page, but not in the words the first pass recorded — the sentence belongs to another vendor, or the number appears only inside a comparison table. A quotation we cannot find in the one reading this study counts on — the text a reader sees — is not evidence about a page, whatever the arithmetic beside it says, so the row is knocked out however large its gap looked.

  • row 12 Leadpages · figure read as $49 · 4 independent reads · note stored at 260 characters and stops where the record stops · 260 characters published Knocked out on the quotation. Two readers fetched the article by separate routes and neither found the sentence the record stored anywhere on it. The page carries that wording under a different vendor's heading and at a different figure; the section on this ve
  • row 23 Instapage · figure read as $199 · 3 independent reads · note in full · 1457 characters published ROW KNOCKED OUT ON THE RE-READ AGAINST THE CANONICAL READING OF THE PAGE - THE TEXT A READER SEES. The stored quotation was a reconstruction, not page text: the sentence "Pricing starts at $199/mo, which is higher than some alternatives." occurs zero times in the article, in raw bytes or in normalised text. What the article carries is that same sentence about a DIFFERENT vendor, at $99/month, under the heading immediately above this one; the stored row took that sentence, changed the number, and attributed it to this vendor. The vendor's own write-up on the page states no figure at all. The $199 is genuinely on the page, but only inside comparison tables, never in prose - recorded here as the table row in document order with cell boundaries rendered as single spaces, nothing reflowed into prose and no fragments joined. On the substance the row would have failed too: $199/month is the vendor's live, standing Optimize price today, so no figure has gone stale. The article's defect is narrower and different - it files the Optimize price under a column headed "Starting Price" when the entry plan is Create at $99/month ($79/month billed annually). That is a plan-labelling error by the publisher, not price drift, and it is not what this study measures. Recorded for our own catalogue: the pricing URL we held, instapage.com/en/pricing, is dead - it 302s to a marketing page carrying no prices - and the live pricing page is instapage.com/plans.

The figure names a different plan, not the entry price 3

The article's figure is right for the plan the article names; our own figure is the vendor's entry plan. Two correct numbers about two different plans.

  • row 1 TaxJar · figure read as $99 · 4 independent reads · note stored at 260 characters and stops where the record stops · 274 characters published ROW KNOCKED OUT. The premise is a misattribution by the first pass. Fetched the [publisher redacted] article today (2026-09-05; JSON-LD datePublished 2025-12-26, dateModified 2026-03-23 -- matches the stated date line). Its TaxJar pricing table is a single run of text listi
  • row 9 Landingi · figure read as $49 · 2 independent reads · note stored at 260 characters and stops where the record stops · 260 characters published The article's $49/month is explicitly attributed to a named upper tier, the "Automate plan", and the sentence itself frames it as the more expensive of "Landingi's two packages" — the one you pay up for to get A/B testing, automation and integrations. The Cons
  • row 10 Leadpages · figure read as $37 · 6 independent reads · note stored at 260 characters and stops where the record stops · 260 characters published Knocked out on plan identity. Two readers found the quoted sentence live and correctly attributed, but the row measured it against Grow at $99/mo, the entry tier of the CRO line rather than the cheapest plan on the page. The vendor's own headline reads "Plans

A monthly headline read against an annual-billed rate 3

Both figures are on the vendor's page today. They differ because they are quoted on different billing bases — or on different regional rate cards under one plan name — which is the single easiest way to read a price wrongly.

  • row 2 ManageEngine Mobile Device Manager Plus · figure read as $2.50 · 2 independent reads · note stored at 260 characters and stops where the record stops · 260 characters published Knocked out on basis. ManageEngine's pricing page publishes no per-device-per-month rate anywhere; it prices MDM Plus in device-count tiers as a bundle total (Standard Cloud 50 devices = $64/month or $645/year; Professional Cloud 50 devices = $119/month or $1,
  • row 7 Leadpages · figure read as $37 · 2 independent reads · note stored at 260 characters and stops where the record stops · 260 characters published The row dies on the article's own wording. The premise records the quote as "$37/month" read as basis monthly, but every one of the three places the figure appears on the live page carries an explicit annual-billing qualifier, and the main sentence supplies th
  • row 24 WATI · figure read as $49 · 3 independent reads · note in full · 1530 characters published ROW KNOCKED OUT ON THE RE-READ AGAINST THE CANONICAL READING OF THE PAGE - THE TEXT A READER SEES - on two independent grounds. First, the stored quotation was a reconstruction: the string "$49/month" occurs zero times in the article, and the bare substring "49/month" does not appear anywhere in it either. The article states the figure three ways and every one of them is explicitly approximate - body prose "Plans start around $49 a month...", a table cell "From ~$49 / mo" with the page's own tilde, and the Sources line recorded here, which is the smallest self-contained run naming both the vendor and the figure. Second, the price is not stale. $49.00/month is a live, standing Wati Growth rate today - "Growth Plan USD Monthly EAST", isActive true, 4900 minor units - with no promotional, introductory or time-limited marker, and the same $49/mo recurs in the quarterly plan, which an introductory rate would not do. Wati runs separate WEST and EAST USD rate cards under one plan name: our recorded $59 is Growth USD Yearly WEST ($708/yr), and the article's $49 is Growth USD Monthly EAST. Both are standing prices for the plan Wati currently calls Growth, so the difference is billing term and region, not the passage of time. A reader in the US or EU who follows the article's own link sees $59/mo billed annually or $69/mo month-to-month and will not see $49 from that address - a fair criticism of the article's regional precision, but a geo-scoping criticism rather than staleness, and not the claim this study makes.

A promotional rate rather than the standing price 2

The vendor was showing a promotional figure. Quoting it is accurate on the day; it simply is not the standing price we record.

  • row 4 Hostinger VPS · figure read as $2.99 · 4 independent reads · note stored at 260 characters and stops where the record stops · 260 characters published Knocked out on the vendor side. Two readers re-fetched the vendor's page by separate routes and both found KVM 1 at $6.49/mo under a 67%-off badge, struck through from $19.49, with "Renews at $11.99/mo for 2 years" printed on the same card. A first-term promot
  • row 15 Hostinger VPS · figure read as $5.99 · 2 independent reads · note stored at 260 characters and stops where the record stops · 260 characters published Article quote is still live verbatim and is presented as Hostinger's starting price, so the quote-gone and different-plan exits do not apply. But the vendor's $6.49/mo is not a list monthly price: hostinger.com/vps-hosting shows it as a 67%-off promotional rat

What we searched, category by category

All 12 categories are listed, including the 6 where nothing was confirmed out of date — a category with no stale rows is a result, not an omission. Of 48 articles found, 3 could not be read (the page did not return readable text to our fetcher) and are excluded from every figure on this page.

Search coverage per category · as of 2026-09-06
Category Articles found Articles read Confirmed out of date
Whatsapp Business 4 4 4
Email Hosting 4 4 3
CRM Software 4 3 2
Domain Registrars 4 4 1
Landing Page Builders 4 4 1
Vps Hosting 4 4 1
Antivirus Software 4 4 0
Consent Management Platforms 4 4 0
Mobile Device Management 4 4 0
Sales Tax Automation 4 4 0
VoIP Business Phone 4 4 0
Visitor Management 4 2 0

Method, limits & data

For twelve software categories we searched for the highest-ranking third-party 'best <category>' articles, read every price each one states for a vendor in our catalog, and compared it with the vendor's own pricing page. Every disagreement was then re-read live by an independent checker, and every checker that reported a stale price was itself double-checked by two more. A row counts as stale only when a majority of independent checks agree that the article still presents that figure as the vendor's starting price and the vendor's own page states a different figure on the same billing basis today. Before publication every surviving row was read once more against the live page, and its quotation was matched against one canonical reading of that page - the same reading for every row, stated in full below. A row is published only where the recorded sentence is found in that reading, and the count the row carries is the count that reading produces. The sentence is a run of the page's text rather than of its bytes: on several of these pages markup falls between the words of the sentence, so the quotation exists as one string only once the tags inside it are gone. A quotation that reading could not find was never rewritten to fit - the row was moved to the control below, whatever its arithmetic said.

The one reading of a page this study counts on. Every “found N times” above is the count this recipe produces, and no other. Parse the served HTML into a DOM and take the document's text content, skipping the contents of script, style, template and noscript elements — a reader sees none of them. Then decode HTML entities, map every no-break space to an ordinary space, collapse each run of whitespace to a single space, and trim. Nothing is put in where a tag was, so a sentence broken across tags reads as one string — which is why several of these quotations exist as one string only in this reading and not in the page's raw bytes. A quotation is then counted once per non-overlapping occurrence in that text, and that count is the number the row prints.

  • The denominator is every price, not every disagreement. 12 of 216 quoted prices were out of date — 5.6%. Reading it against the 82 disagreements instead (14.6%) would flatter the finding, and against the 25 first-pass calls (48%) would flatter it more.
  • We are in the same position. We corrected 271 of our own figures in the same quarter — 22.6 for every one confirmed here — found by auditing every plan whose recorded source was not the vendor's own page. A price index is a perishable thing to keep, and this study measures the perishing, not anyone's care.
  • A row is one figure on one day. It is not a judgement of an article, a publisher, or the rest of that page — most prices in these articles matched. An article can be corrected at any time, and a row here may already be out of date in the other direction.
  • Only vendors we carry. A price is checkable only where we hold our own dated, sourced figure, so the 216 prices are the subset each article states for a vendor in our catalogue — never its whole table.
  • The baseline is the plan we record. A row can only test a published figure against the plan we ourselves hold, and a row measured against any other figure for that plan has to say why on the row. In 7 of the 12 confirmed rows above, the vendor's figure today is the price our catalogue records for that plan. The other 5, #1, #2, #6, #7 and #8, were measured against another figure for that plan, and each says why on the row, printed with the evidence as its baseline exception.
  • Every quotation is a run of the page's own text, not a reconstruction. Before publication each surviving row was read once more against the live article and its quotation matched against the single reading described above. A row is published only where the sentence is found in that reading, and the count beside it is the count that reading produces. Several of these sentences are not a run of the page's raw bytes at all — markup falls between their words — which is exactly why the study counts on the text a reader sees rather than on the bytes. A quotation that reading could not find was never rewritten to fit: 3 rows were moved to the control on the quotation, whatever their arithmetic said — control rows 12, 23 and 24: every control row filed under “The quoted sentence is not the sentence on the page”, plus any row whose own note calls the stored quotation a reconstruction. Every quotation is published whole, with its length beside it; nothing on this page is silently cut.
  • Same-basis or nothing. A figure counts as out of date only where the vendor's own page states a different figure on the same billing basis. Every row records that basis, and the rows where the basis explained the difference are published in the control above.

Download: JSON · CSV · Markdown (CC-BY, see open data) · all research · our accuracy standard. Re-checked 2026-09-06; re-run quarterly.

To cite or embed this finding, copy and paste this: