The re-test log

Most review sites quietly rewrite old pages and hope you never notice the tool they raved about last year is gone. We keep the receipts instead. Every addition, demotion, price correction, and removal lands in this log with a date and a reason. If this page goes quiet for months, hold us to it. The re-test dates printed on every stack page are promises, and this is where they're kept.

July 2026

• 2026-07-08: Site launch (Research Editions). Four stacks published: Freelance Writers, Solo Consultants, Real Estate Agents, Tradespeople. All labeled Research Edition, meaning picks based on feature analysis, real solo-tier pricing, and documented practitioner experience. First hands-on test cycle is underway; each page gets re-scored with logged results as its cycle completes. Comparisons published: Surfer vs Frase, Jasper vs Claude.
• 2026-07-09: Price corrections (Surfer, CompanyCam). Our price check caught two vendor changes: Surfer discontinued the $89 Essential plan (now Discovery ~$49 / Standard ~$99) and CompanyCam moved from ~$24/user to team plans from ~$63. Writer-stack math updated $141→$151; trades-stack math $85→$125. Original claims preserved here per our correction rule.
• 2026-07-09: Upheal pricing corrected same day it was published. Our therapists stack initially listed Upheal at "from ~$29/month"; the verified model is $1/session capped at ~$69/month. Corrected across the stack page, cheat sheet, and comparison within hours. Stack math updated to ~$70–135/month usage-based.
• 2026-07-09: Notion closed its affiliate program; our ranking didn't move. Notion stopped accepting new affiliates (terms had also changed to ~$50/signup + 20% of year one). Our Notion links now earn $0. Notion remains the pick in the Writers and Developers stacks and the Notion-vs-Coda verdict is unchanged. Commissions never touched the ranking on the way in, so losing one changes nothing on the way out. All labels and the Commission Table updated.
• 2026-07-21: Two comparisons published; Bonsai tier name corrected. Added eRank vs Marmalead and Bonsai vs HoneyBook, both Research Edition with vendor-page-verified pricing. While researching the latter we caught that Bonsai renamed its ~$25 tier from "Starter" to "Essentials" (price unchanged); the Solo Consultants and Freelance Developers pages and cheat sheets previously used the old name and now carry the current one.
• 2026-07-22: Surfer declined our affiliate application; the verdict doesn't move. Surfer's program (~25% recurring, the second-biggest payout on our list) declined us as a new site without traffic history. Result: our Surfer links now earn $0, exactly like our Claude links. The recommendation stays what it always was: in the writer stack, conditional, with the same skip rule ("skip unless SEO is ⅓+ of your income"). The Commission Table is updated. We'll reapply once traffic data exists; the ranking won't be waiting on it.
• 2026-08-06: Fathom declined our partner application; the free tier stays recommended. Fathom's Growth Partner program turns out to be built for agencies and consultants who implement it for clients, not for publications like ours, and our application was declined. Our Fathom links now earn $0, joining Claude, ChatGPT, Copilot, Notion, and Surfer in that column. The verdict is unchanged: Fathom's free tier is still the meeting-notes pick in the consultant stack and the fallback in the recruiter stack, which was always an odd thing for a commission-driven site to say and a very normal thing for us. The Commission Table is updated.
• 2026-08-13: correcting ourselves. Fathom did not stay declined, and we were three days late saying so. Fathom accepted us into its Growth Partner Program on 2026-08-10, four days after the entry above. We did not notice until reviewing partner accounts on 2026-08-13, so for three days this site claimed we had been declined by a programme that had in fact taken us on. The entry above is left standing because that is what our correction rule requires. Nothing about the verdict moves: Fathom’s free tier was the meeting-notes pick when we thought we would earn from it, when we thought we had been rejected, and now that we know we were accepted. The tracking link is not placed yet, so the links still earn \$0 as this is published, and the commission table now says so.
• 2026-08-13, same day: the Fathom link is live, making it our third paying link. Tracking link placed, so Fathom now earns us a commission where it earned nothing this morning. The on-page labels changed from “not yet an affiliate link” to “affiliate link, active” on the consultant stack, the real estate stack and the head-to-head. What did not change: Fathom’s free tier is still the pick we push, on a page where we now earn only if you upgrade to the paid one. We wrote that recommendation believing Fathom would never pay us anything, and we are not going to quietly upgrade the advice now that it does. eRank, Helium 10 and Fathom are the three links that pay us; the rest of the site earns nothing.
• 2026-08-13: Home inspectors stack published (stack #14). First of a deliberate shift toward professions whose practitioners skew older: licensed, solo, second-career heavy, and almost entirely ignored by AI-tool coverage. The target here is unusually clear, because report writing costs a solo inspector three to four hours per job and none of it is billable. Research turned up something no comparison we found mentions: HomeGauge and Spectora are sold from the same storefront. HomeGauge’s own pricing page lists HomeGauge at $89/month and Spectora at $109/month side by side, each with its own trial button, so the rivalry everyone debates in inspector forums is a choice between two products under one roof. Prices verified August 2026 against US pricing. The page leads with the liability rule rather than the tools: AI writes sentences, the inspector supplies every finding, because the report carries a licence number and “the software wrote it” is not a defence. We also rejected CompanyCam here despite it being the right answer in our trades stack, since both report platforms already handle photos. None of the links on the page pay us anything.
• 2026-08-13: Financial advisors stack published (stack #15), and it inverts our usual conflict. Second of the older-skewing professions. The expensive decision is meeting notes, where advisor-specific tools cost about $100 per advisor a month (Jump) against $0 for a general notetaker’s free tier. Here is the awkward part, stated on the page itself: Fathom pays us and Jump does not. So the cheap pick is the one that earns us money, which is the reverse of the usual affiliate temptation, and we had to be careful in the opposite direction. The page argues both sides on merit: pre-CRM advisors should start free, and practices running a real book in a CRM should seriously consider paying, because the integration is the entire product. We also surfaced the option most comparisons skip, that Wealthbox sells its own notetaker as a $49 add-on, which undercuts Jump for anyone who just wants CRM sync. Prices verified August 2026 against US pricing. The compliance section is the real page: records retention, the Marketing Rule and AI washing, and the client-data vendor question, all flagged as compliance-officer decisions rather than ours.
• 2026-08-13: Insurance agents stack published (stack #16), and it has a hole in it on purpose. Third of the older-skewing professions, and the least price-transparent category we have covered in sixteen stacks. The agency management system is usually the largest software line in an independent agency, and it is the one we could not price. EZLynx, now part of Applied, publishes a page titled How Pricing Works that contains no prices. HawkSoft’s pricing URL returns a 404. We are not going to invent a monthly figure for either, so the page says so and instead lists the five questions to ask on the sales call, including the year-two renewal price and whether AI features are an add-on. AgencyZoom does publish: Essential $149, Growth $199, Pro $349 a month, verified August 2026, and it gets credit on the page for that. The honest headline is that the AI part of this stack costs $20: Claude plus Canva’s free tier. Everything expensive is agency infrastructure with AI features attached. The licence rule leads: the policy document is the contract, AI can help you explain coverage and must never state it. None of the links pay us anything.
• 2026-08-13: sellerboard declined us, hours after accepting. The pick does not move. We were accepted into sellerboard’s programme in the morning and declined in the afternoon, after what their marketing manager described as reviewing our profile again. The stated reason: they issue affiliate links only to partners who have their own channels, social media platforms or courses. We are a publisher rather than a creator with an owned audience, so we do not qualify. Our sellerboard links now earn $0, joining Claude, ChatGPT, Copilot, Notion, Surfer, Jungle Scout and the rest. sellerboard remains the profit-after-fees pick in our Amazon sellers stack, because it was chosen on merit before any commercial conversation and a decline is not a product review. Worth noting the pattern, since this is the third decline with the same shape: Surfer wanted traffic history, Fathom’s programme was built for agencies implementing for clients, and sellerboard wants creators with channels. Affiliate programmes in 2026 are largely built for audiences someone already owns, not for review sites. That is a real headwind for a site like this and we would rather say so than pretend every silence is a maybe. The Commission Table is updated.
• 2026-08-12: DeepSeek vs ChatGPT published; neither link pays us. Our first comparison covering a Chinese AI tool. The verdict turns on the client-data rule: DeepSeek's free chat is genuinely excellent for your own drafting and code, and its privacy policy states data is stored in the People's Republic of China with no BAA, so client-confidential material never goes in. Prices and policies verified August 2026. Both links on the page earn us $0, making it the cheapest honest opinion we publish.
• 2026-08-12: Amazon sellers stack published (stack #12). Our second e-commerce persona: Claude for listing copy, Helium 10 or Jungle Scout for research (pick one, never both), sellerboard for honest profit numbers, PhotoRoom for images. Research caught that Helium 10 retired its $39 Starter tier for new users, so the page states the real entry price instead of the one every old listicle still quotes. Includes the review-manipulation and AI-listing rules that actually suspend accounts. All prices verified August 2026; none of the new links pay us anything today.
• 2026-08-12, later the same day: Helium 10 approved us; the skip note stays. Our Helium 10 affiliate application was auto-approved hours after the Amazon sellers stack went live, making it the second link on this site that pays us (after eRank). The terms are strong: 100% of a referred subscriber’s first month plus recurring tiers. Which is exactly why we point out that the strictest price warning in that stack sits on Helium 10, written before we earned a cent from it. The Commission Table is updated.
• 2026-08-13: Resellers stack published (stack #13). One persona across four marketplaces: eBay, Poshmark, Mercari and Depop. Claude for descriptions and buyer replies, Vendoo or List Perfectly for crosslisting (run one, never two), PhotoRoom for phone photos, and free eBay sold-listings and Terapeak data for comps, which is the pick we tell you not to pay for. From ~$22/month. The compliance section covers the operational risk that actually costs resellers money: an item that sells on one platform and stays listed on another becomes an oversell, and oversells damage seller metrics on every marketplace involved. Prices verified August 2026; none of the new links pay us anything today.
• 2026-08-13, hours later: Vendoo free-tier claim corrected. The resellers stack said Vendoo offered a free plan of roughly 5 new items a month. Checking Vendoo's pricing page directly while setting up an account showed no free plan at all: the current tiers are Starter, Growth and Pro, with a 14-day free trial instead. Corrected in six places on the stack page, including the recommendation that told readers to start on the free plan. The original claim is preserved here per our correction rule. This is why we verify prices on vendor pages rather than trusting the last article that quoted them.
• 2026-08-13: Two head-to-head comparisons published. Helium 10 vs Jungle Scout and Vendoo vs List Perfectly, both expanding a question the stack pages previously answered in a single paragraph. Prices and verdicts carry over unchanged from the Amazon and reseller stacks, verified August 2026. The Helium 10 page is the first thing we have published around a link that pays us, and the verdict still sends every seller under roughly $5K a month to Jungle Scout or to Helium 10’s free plan: Helium 10 pays our largest commission of any program we are in, Jungle Scout pays us nothing, and the recommendation is unchanged from the day we wrote it with no commission at stake. Neither Vendoo nor List Perfectly pays us anything today.
• 2026-08-13: Metaview price corrected; recruiter stack math rebuilt. Our monthly price check flagged Metaview, and the vendor page confirmed a real change: Metaview has moved to per-agent pricing, and its Notetaker is now a free tier at 25 calls a month or Pro at about $60/month per user. We had published a solo tier at about $25/month, which no longer exists. Corrected on the recruiter stack page and its cheat sheet, and the stack math moves from ~$95/month to ~$72/month on the free tier or ~$132/month with Pro. The original claim is preserved here per our correction rule. The free tier is genuinely good news for solo recruiters and now leads the recommendation.
• 2026-08-13: Three flagged prices re-verified as correct. The same check flagged Surfer, Notion and Calendly. All three are unchanged: Surfer Standard $99/month and Discovery $49/month, Notion Plus $12/month billed monthly and $10 annually, Calendly Standard $12/month billed monthly and $10 annually. The flags were false positives, because those pages render prices with JavaScript and default to annual billing, which our fetch-based checker cannot see. Logged so the next check does not chase them again.
• 2026-08-13, same check: Recruit CRM price was badly wrong; no stack total until we confirm it. Following the currency warning above, we went back to Recruit CRM’s pricing page. It quotes CAD 125 per user per month for Pro and CAD 185 for Business, with no currency selector. We had published about $40/month, which the vendor page does not support at any plausible exchange rate, so that figure is gone. Because the CRM is the dominant line item and we have only seen it in CAD from Canada, we have stopped publishing a single recruiter-stack total until the US price is confirmed. The page now states the three smaller tools at about $32/month and says plainly that the ATS sits on top and is being re-verified. Preserving the original claim here per our correction rule: we said ~$40/month for Recruit CRM and a ~$95/month full stack, and both were wrong.
• 2026-08-13: Frase repriced; the Surfer comparison gets sharper, not softer. Frase has restructured. Starter is now $39/month billed yearly or $49 month to month, and its page states USD explicitly. We published a flat ~$45. Against Surfer Standard at $99/month billed yearly, the like-for-like gap on annual billing is $60/month rather than the ~$54 we claimed, and Frase is well under half of Surfer rather than “about half”. Corrected on the comparison page, the writer stack and its cheat sheet. This one happens to strengthen our verdict, which is exactly why it still gets logged: corrections that flatter us are not a different category from corrections that do not.
• 2026-08-13: Jobber retired the Core plan; the trades stack was understated by about $78 a month. Jobber’s current lineup is Connect at $199/month, Grow at $299 and Plus at $499, all including five users, and the page states all prices are USD. There is no Core tier any more. We published “Jobber (Core) from ~$39/month” and a ~$125/month full stack, both of which assumed a plan that no longer exists. The entry point is now $120/month for the first 12 months billed annually, then $149. Stack math corrected to ~$203/month on the promotional year, ~$232 after it reverts, and ~$282 at list. Fixed across the stack page, its FAQ, the meta descriptions and the cheat sheet. This is the largest single correction we have had to make, and it sat on the most expensive stack we publish.
• 2026-08-13: Virtual Staging AI tiers clarified; a per-photo figure did not match its price. Verified tiers: Basic is $16/month billed yearly or $25 monthly for 6 photos, and Standard is $19/month billed yearly or $35 monthly for 20 photos. Our $16 was right but it is the annual rate for the 6-photo tier, which we did not say. Worse, the agent cheat sheet paired that $16 with “~$1/photo”, and those two do not go together: $16 buys 6 photos at $2.67 each, while ~$0.95 a photo is the Standard tier. Both corrected, and the stack math now states which tier and billing mode the ~$67 total assumes.
• 2026-08-13: Three vendors we could not verify, and why. Closing out the August price sweep honestly. gamma.app sits behind a Cloudflare bot challenge that we cannot pass. canva.com and recruitcrm.io both quote us in Canadian dollars with no currency selector, and Canva sets regional prices independently rather than converting, so the CAD figure cannot be turned back into a reliable USD one. Their published figures are unchanged rather than guessed at. Related near-miss worth recording: claude.com also served us CAD with no label, showing Pro at $28 and Max from $140 against the real $20 and $100. Both are exactly 1.4x. Had we “corrected” that, we would have put a wrong price on roughly thirty pages while looking diligent.
• 2026-08-13, later: the two CAD-blocked prices resolved over a US VPN. Recruit CRM Pro is $95 per user per month in USD, with Business at $135 and up to 20% off on annual billing. We had published ~$40, so the earlier correction was right to remove it. With the ATS figure known, the recruiter stack carries a real total again: ~$127/month on Metaview’s free notetaker tier, or ~$187 with Metaview Pro, against the ~$95 we published for the whole stack before today. Canva came back clean: Pro is US$180 a year for one person, exactly the $15/month we already publish, so nothing changed. Worth recording why the Canadian reading was useless for checking it: Canva shows CAD 142/year there, roughly 42% below a straight conversion, because it prices per region rather than converting. gamma.app remains unverified behind a bot challenge.
• 2026-08-21: Virtual staging page published, and it is mostly about the law. Search data showed people asking us some version of “estate agents ai to stage listings” four different ways, landing on a stack page that answered it in one paragraph. The new AI virtual staging page answers it properly, and the research turned up something our own agent stack was understating. Virtual staging disclosure is no longer only an ethics and MLS-rules question: California AB 723 was signed on 2025-10-10 as Chapter 497, adding section 10140.8 to the Business and Professions Code, and it requires a disclosure on or next to any digitally altered listing image plus a link, URL or QR code to the original unaltered version. “Available on request” does not satisfy it. We also pinned the ethics citation to NAR Standard of Practice 12-10 after finding published guides citing 12-1 instead, which is a different standard. The page states plainly that we have not read individual MLS rulebooks and will not summarise anyone’s from memory. On price, the four verified tiers are laid out with the per-photo arithmetic done, which exposes that Basic is the worst value on the list at $2.67 a photo against Standard’s $0.95, for the sake of $3 a month. And we deliberately did not write the roundup this query wants: we cover one staging tool, because it is the only one already in a stack, and a ranked list of products we have not evaluated is exactly the article we exist to not publish.
• 2026-08-21: a live undercount corrected. Three links pay us, not two. Fourteen pages carried the line “exactly two links on this site pay us anything”. That stopped being true on 2026-08-13 when the Fathom tracking link went live, which the changelog recorded at the time while the disclosure line on every stack page did not get swept with it. Corrected everywhere to three: eRank, Helium 10 and Fathom. Preserving the error here per our correction rule, and noting the shape of it, because this is the failure mode that matters most on a site whose claim is that it tells you who pays it. Understating our own commercial interest is still getting it wrong.
• 2026-08-26: Spectora vs HomeGauge published, and the rivalry does not exist. Our home inspectors stack already called this an either-or choice, but the new Spectora vs HomeGauge comparison settles a question every inspector forum still argues about: they are not two companies. Spectora announced its acquisition of HomeGauge on 2025-04-01, HomeGauge’s own pricing page now carries a logo reading “HomeGauge, a Spectora Company”, sells Spectora right beside it, and points both free-trial buttons at spectora.com. The arithmetic is the other finding. Everyone leads with $89 against $109 and stops. Carried out to a year, HomeGauge is $1,068 and Spectora’s annual plan is $1,090, so the real gap is about $22 a year, not $20 a month, unless you pay Spectora monthly and hand back $240. That moves the decision off price entirely and onto the thing nobody prices: HomeGauge’s classic writer is Windows desktop software and its own FAQ sends Mac owners to Parallels, while Spectora runs in a browser. Also logged: Spectora publishes a second-inspector price ($99/month or $999/year) and HomeGauge’s pricing page publishes none. Both prices verified on the vendor pricing pages on 2026-08-26. Neither link pays us anything and we have applied to neither.
• 2026-09-01: Monthly price pass, and it is the largest set of corrections we have published. Sixteen figures moved, one of them because we got it wrong rather than because a vendor did. In order of how much they change a decision:
  • Jobber: we were wrong, and the trades stack was overstated by about $90 a month. In August we published that Jobber had retired its Core tier and that the entry plan was Connect at $199 a month. Core is on Jobber’s pricing page today at $29/month billed annually, $39 on a one-year commitment and $49 month to month. The $199 we quoted is the month-to-month price of Grow, two tiers above Core. The tradespeople stack goes from ~$203/month to ~$112/month. If you priced that stack here in August and walked away, price it again.
  • SimplePractice: from ~$29 to $49/month for the Starter plan (Essential $79, Plus $99). This is the largest single increase we have recorded. The therapists stack moves from ~$70–135 to ~$90–153/month.
  • Topaz Video AI: ~$299 one-time is now $299 a year. Topaz stopped selling perpetual licences on 2025-10-03. Existing owners keep what they bought; a new purchase is a subscription, Personal $299/year or Pro $699/year. We recommended this tool partly because you bought it once and owned it, and the video editors page now says so plainly.
  • Descript: Creator from ~$16 to ~$24/month billed annually ($35 month to month) after a tier restructure; the ~$16 rate now buys Hobbyist. Video editors goes ~$55 to ~$61, real estate agents ~$67 to ~$75.
  • Jasper: the ~$49 Creator tier is gone. Jasper Pro is $69/month, $59 billed annually. Against Claude Pro at $20 the gap is now more than three to one, and the verdict on that page was already Claude.
  • Fathom Premium: $15 to $20/month ($16 billed annually). Every stack that uses Fathom uses its free tier, so no stack total changed except the solo consultants’ optional upgrade line.
  • TaxDome: ~$60 to ~$67/month, billed as $800 per seat per year on a one-year commitment ($750 on two years, $700 on three). Accountants goes ~$110 to ~$112, and to ~$45 without TaxDome.
  • Notion Plus: $12 to $10/month. A price cut, which is rarer here than it should be. Freelance writers goes ~$53 to ~$50, freelance developers ~$67 to ~$65, the freelancer hub spine ~$57 to ~$55.
  • HoneyBook Starter: ~$39 to $29/month, and it is $29 whether you pay monthly or yearly. Bonsai is still cheaper on annual billing, but by about $10 a month rather than $20.
  • Calendly Standard: $12 to $10/month ($8.30 billed annually). Recruiters goes ~$127 to ~$125.
  • Reclaim Starter: ~$8 to $10/month. Gamma Plus: ~$10 to $9/month billed annually ($108/year). Solo consultants lands at ~$64, essentially where it started.
  • Dext: the tier we named does not exist. There is no Solo or Starter plan for businesses any more, only a Business plan that starts at five users and 250 documents: ~$25/month billed annually ($302.50/year), about $31.50 month to month. We had it at ~$30.
  • MagicSchool Plus: ~$8 to $8.33/month billed annually ($99.96/year), $12.99 month to month. The teachers stack total is unchanged at ~$25.
  • Vendoo has a free plan, and in August we told you it did not. Their pricing page lists a $0 tier capped at 5 items and 3 background removals, alongside the 14-day trial. Five items will not run a business, but it beats a trial clock as a first look. Corrected on the resellers stack and the Vendoo vs List Perfectly page.
  • Blueprint’s free Core tier is gone. We described it as a real evaluation rather than a teaser. Blueprint now charges $0.99 per session alongside your existing EHR or $1.49 for its all-in-one, after 5 free sessions or a 30-day trial, with no monthly cap. That makes Upheal’s ~$69 cap the whole difference between them.
  • Metaview no longer publishes a notetaker price. Its pricing page now lists sourcing-agent tiers only ($0 for the first 100 profiles, $100/user/month for 200, $300 unlimited), and the notetaker sits inside a quoted bundle. We have removed the “free tier: 25 calls a month” claim because we can no longer confirm it, and marked the ~$60 figure as last-confirmed rather than current.

  Two things we could not verify, said out loud rather than quietly restamped. Artlist’s pricing page now leads with its AI plans and gives the stock Music and SFX tiers no reachable URL of their own, so we could not read the current figure on the vendor’s own page. The ~$17 on the video editors page is July’s number carried forward, and it is flagged there as such. ChatGPT’s Pro tier gave us conflicting readings, so we removed the $200 figure from the DeepSeek comparison rather than guess. Free, Go at $8 and Plus at $20 are confirmed.

  One name change worth knowing: coda.io/pricing now redirects to superhuman.com/plans/docs, and Coda is sold as Superhuman Docs alongside Grammarly and Superhuman Mail. Doc Maker billing and the tiers are unchanged at $12 per maker billed annually, $15 month to month. Nothing about the recommendation moved.

  Re-verified and unchanged, so that a missing correction is not mistaken for a missing check: Claude Pro $20 ($17 annual), Surfer Discovery $49 and Standard $99, Frase $39 annual and $49 monthly, Grammarly Pro $12, Otter Pro $8.33 annual, Bonsai Essentials $25 ($19 annual), Canva Pro US$180/year, CompanyCam Core $63, Upheal $1/session capped ~$69, Recruit CRM $95 / $135 / $215, GitHub Copilot Pro $10, eRank Basic $5.99, PhotoRoom Pro $12.99 ($7.50 annual), Helium 10 Platinum $99 promotional and $129 list, Jungle Scout $29 and $49, sellerboard $19 ($15 annual), List Perfectly $29 / $49 / $69 / $99+, Virtual Staging AI $16 / $19 / $39 / $79, Spectora $109 and $1,090/year, HomeGauge $89, Jump $100 per advisor, Wealthbox $59 / $75 / $99, AgencyZoom $149 / $199 / $349, Diffit $14.99, Frame.io Pro $15, Marmalead $19 with the $300 lifetime option, DigitalOcean from ~$6. Most months, that list is the whole report.

• 2026-09-02: Jobber pricing gets its own page, built around the mistake we made reading it. Yesterday’s price pass corrected a $90-a-month error on the trades stack: we had published that Jobber retired its Core tier and that the entry plan was Connect at $199, when Core is still on their page at $29 billed annually and $199 is the month-to-month price of Grow, two tiers up. The new Jobber pricing page exists because that error was easy to make and the reason is structural. Jobber’s pricing page shows one plan at a time against a team-size selector and a billing selector, so a number read off it can be attached to the wrong plan without anything looking wrong. The page puts the whole grid on one screen instead: four plans against four team-size brackets against three billing options, all read off getjobber.com/pricing on 2026-09-02. Three findings came out of laying it flat. Core is a one-user price, and Jobber stops offering Core entirely once the team-size selector goes above one person, so the published step from one person to two is $29 to $149, not $29 plus a seat; their Core card still carries the same add-a-user-for-$29 note as every other plan, and we have left that contradiction visible rather than guessing which reading is right, because the answer is worth about $1,092 a year to a two-person shop. The 14-day trial runs on Grow, the third tier, so a fortnight of automated follow-ups, two-way SMS, job costing and QuickBooks sync disappears on day fifteen for anyone who then subscribes to Core. And payment processing is identical on all four plans at 2.9% plus 30 cents, so upgrading never reduces it: at $20,000 a month across 100 invoices that is about $610, more than every plan except Plus. The add-ons are priced too (Marketing Suite $99, Pipeline $49, AI Receptionist $29), which means all three attached to Core cost six times the plan itself. Our August error is written up on the page in full, with the dollar figure and the 19 days it stood. Jobber pays us nothing: we applied on 2026-08-13 and have had no reply.
• 2026-09-07: Kilo Code vs GitHub Copilot, and a correction to the premise we started with. Kilo Code’s native JetBrains agent was the number one launch on Product Hunt for the week of 31 August, and the tagline (“fully native, open-source coding agent built for JetBrains”) invites you to assume JetBrains support is what Copilot lacks. We assumed it too, and it is wrong: GitHub’s own plans page lists the JetBrains suite of IDEs among supported environments, next to VS Code, Visual Studio, Vim, Neovim, Xcode, Eclipse and Zed. We caught that before publishing rather than after, and the new comparison now opens by dismantling the framing instead of repeating it. What is left once the IDE question goes away is a genuine difference in who prices your inference, and one number nobody seems to mention: Copilot Pro is $10 a month and includes $15 of monthly GitHub AI Credits (Pro+ $39 includes $70, Max $100 includes $200), so part of that subscription is a credit purchase at better than face value. Kilo’s extensions are free and open source with “usage is billed separately” printed underneath, no markup on inference, and 500+ models across 60+ providers, or your own key, or a local model. The page refuses to say which is cheaper, because that answer needs a logged month of real inference spend against the same work and we have not run it: what we will say is that Copilot’s bill is knowable in advance and Kilo’s is uncapped in both directions. Both plan pages read 2026-09-07. Two disclosures. Copilot pays us $0 and has no programme to join, unchanged. Kilo does run a direct partner programme, and we deliberately did not read its terms until the verdict above was written; we are not enrolled, and if that ever changes it appears in the commission table and here before it changes a word of the comparison. Also logged as unverified rather than repeated as fact: several review sites report Anaconda acquired Kilo Code in July 2026, which we could not confirm against a first-party statement.
• 2026-09-09: Jump vs Fathom for advisors, and the first decision chart. The advisor stack had no page of its own beneath it, so the meeting-notes decision gets one. It is also the first page to open with a decision flowchart: two questions, three outcomes, prices on the chart, and the same logic repeated in words underneath for anyone the chart does not suit. The verdict pays us less than the alternative: Fathom’s free tier for advisors without a CRM (pays us), Wealthbox’s own $49 notetaker add-on when sync is all you need (pays us nothing), and Jump at $100 for prep plus sync (pays us nothing).
• 2026-09-09: The accountant stack gets a workflow figure. One image, five boxes, between the short version and the stack list: TaxDome, Dext, Claude, Calendly, and a small red box for the accountant who files. It shows how the tools fit together as the vendors document them and as the page recommends using them. The hand-offs are documented features or manual steps, not a workflow we have timed, and the caption says so on the page. The Claude box reads “drafts, anonymised only” because that is the §7216 rule, not a stylistic choice. More stacks get the same treatment as the figures are made.
• 2026-09-09: Seven more stacks get workflow figures, and the accountant one is redrawn. Financial advisors, insurance agents, recruiters, teachers, video editors, freelance writers and solo consultants each get the same kind of figure the accountant stack got this morning: the input on the left, the tools in the order the page recommends, one green box for the pick the page names first, and a small red box for the person who signs off. The accountant figure was redrawn to start at the client and put Calendly where it belongs, at the intake. Every figure shows documented features and manual hand-offs, not a workflow we have timed, and each caption says so.
• 2026-09-15: Fergus vs Jobber, a challenger that relaunched three days before we wrote about it. Fergus, a job-management platform already used by more than 26,000 tradespeople across Australia, New Zealand and the UK, announced an AI-first relaunch on 2026-09-13: Fergus Assistant, Fergus Go, and a new Fergus AI layer. The new comparison against our pick, Jobber, found the interesting gap one tier up from the entry plan: Fergus Professional is $44/month against Jobber Connect's $99 to $139 for what reads as a comparable feature set on both vendors' pages. Fergus bills month to month only, so like for like the entry tier is $31 against Jobber Core’s $49, or $29 if you commit to Jobber for a year. Both trials fail in opposite directions: Fergus trials its bottom tier, Jobber its third. We caught a currency trap on the way: fergus.com/pricing with no region path showed different, higher figures than fergus.com/us/pricing, so only the US-region page was used, and even that page never prints an explicit USD label, which the page says plainly rather than asserting as settled fact. We are not crowning a winner. Fergus's AI layer is three days old with one customer quote in the press release and nothing we have tested ourselves; Jobber has the longer US track record and the price grid we already mapped in detail on 2026-09-02. Neither link pays us anything today: we applied to Jobber on 2026-08-13 with no reply, and Fergus runs a one-time $250 to $500 per-sale referral programme through PartnerStack that we have deliberately not applied to yet, the same verdict-first order as Kilo Code and Frase.
• 2026-09-16: Four more workflow figures, and a loop we had left open. Etsy sellers, real estate agents, resellers and tradespeople get the same kind of figure the other eight stacks carry. While checking them we measured every figure on the site for a complete return loop and found the solo consultants one had never drawn the riser under the human box; it is completed now. Same rules as before: documented features and manual hand-offs, not a workflow we have timed, and each caption says so.
• 2026-09-23: Topaz Video pricing, laid out in full. Our September price pass recorded that Topaz ended perpetual licences, but our video editors page only ever carried two of Topaz's prices. The new pricing page puts the whole grid on one screen, read off topazlabs.com/pricing on 2026-09-23: the same standard licence costs $299 a year prepaid, $468 on a 12-month commitment billed monthly, or $708 month to month. The Studio bundle is $399 a year for every Topaz app, which is cheaper than Topaz Video plus Photo bought separately. Topaz's own wording puts the Pro line at $1M in organisation revenue, so most freelancers do not need Pro. One conflict on Topaz's page is flagged rather than resolved: Studio Pro reads $79 a month on the billing tabs and $75 in a lower block. The same edit fixed two sentences on the video editors page that still called Topaz a one-time purchase, one of them directly below the note saying it had become a subscription. Both had been live since the September price pass. Topaz pays us nothing.
• 2026-09-30: Underpriced AI vs Terapeak, a challenge to our $0 pick. Our reseller stack tells you to pay nothing for comps research. Underpriced AI, a photo-to-sold-comps app from $5 a month, is the strongest argument against that advice we have seen, and people searching its name were already landing on our reseller page. The new comparison reads its pricing and its own published accuracy grade on 2026-09-30. The grade is the finding: a 40% typical gap on 214 sales where sellers set their own price, and on the 79 sales ChatGPT answered from the photo alone, both missed by about $33 on a $100 item. So the $5 buys identification from a photo and the sold comps attached, not a better number, and it is worth paying for only if you source items you cannot name. Our $0 pick stands for everyone else. The page also notes that the app’s delist is a reminder rather than automatic, and that SaleFreaks, another name in the same searches, sells Amazon-to-eBay dropshipping that eBay’s own policy says is not allowed. Neither link pays us anything.
• 2026-09-30: Every stack now opens with a clickable workflow. The generated images are gone. Each of the sixteen stacks now draws its workflow in the page itself: the input on the left, one tile per tool in the order the page recommends, our pick tagged, and a person at the end who signs off. Each tool tile is a link, and it goes through the same link registry as every other link on the site, so the ones that pay us are marked the same way. The tiles use each vendor’s official logo, taken from its own brand page or website and left unmodified; where a vendor’s terms reserve its logo, the tile shows a plain icon instead. Four stacks that had no figure (Amazon sellers, freelance developers, home inspectors, therapists) have one now. The therapist figure deliberately shows only the clinical half of the practice, because our pick there never touches client data. As before, the figures show documented features and manual hand-offs, not a workflow we have timed.
• 2026-09-30: Every comparison now opens with its decision in one chart. Fifteen head-to-heads get a decision chart in the same style as the stack workflows: the question that decides it, a branch for each answer, and the tool that answer points to. Each tool is a link, and the small line under it saying whether it pays us is read straight from our link registry when the chart is built, so it cannot say anything the commission table does not. The same goes for the sentence under each chart. Logos are the vendors’ own; where a vendor’s terms reserve its logo (OpenAI, DeepSeek, eBay, HoneyBook and Frase among them), the chart shows a plain icon instead.
• 2026-10-01: monthly price check, two corrections. We read every vendor pricing page from a US location (or, where a page hides its prices from scripts, from the page data itself) and compared it with what we publish. Upheal stopped selling the plan our therapists stack was built around: $1 per session, capped at about $69 a month. Its pricing page now offers a Free plan (notes from dictation, typed summaries and uploads only, individual therapy only), an AI Notes plan at $39 a month flat for unlimited sessions, an EHR plan at $1 per session with no cap we can find, and EHR Pro at $99 a month. The old Starter and Premium plans sit on a legacy page, still honored for existing customers. We moved the therapists stack from ~$90–153 a month to ~$108–123 (SimplePractice $49, Upheal $39, Claude $20, Canva Pro optional), rewrote the Upheal vs Blueprint comparison around a break-even of roughly 39 sessions a month at Blueprint’s $0.99, and removed our own line that Upheal and Blueprint were both usage-based. We also fixed the freelancer hub, which still carried a stale “~$70–135” for this stack. Canva Pro is US$144 a year on Canva’s US pricing page, about $12 a month on annual billing. Two pages said US$180 a year; both are corrected. We could not read Canva’s month-to-month price, so the other pages keep their ~$15 figure as a monthly-billing ceiling rather than a confirmed number. Re-verified and unchanged: Claude Pro $20 ($17 annual), Surfer Discovery $49 and Standard $99, Frase Starter $39 annual and $49 monthly, Grammarly $12, Otter $8.33 annual, Notion Plus $10, Fathom Premium $20 ($16 annual), Reclaim $10, Gamma Plus $9 annual, Bonsai Essentials $25, Descript Creator $24 annual and $35 monthly, Jobber Core $29 annual and Connect $99, CompanyCam $63, Virtual Staging AI $16/$19/$39/$79, SimplePractice $49, Recruit CRM $95/$135/$215, Calendly $10, Topaz Video $299 a year, TaxDome Essentials $800 a year, Dext Business $25.21, MagicSchool $8.33, Diffit $14.99, HoneyBook $29, Jasper Pro $69, Blueprint $0.99 a session, Marmalead $19, Vendoo $14.99, Copilot $10, sellerboard $19 monthly ($15 annual), Helium 10, Jungle Scout, List Perfectly, Spectora, HomeGauge, Wealthbox and AgencyZoom. Not re-verified this month: Metaview still publishes no notetaker price, and Artlist’s own plans article lists Music and SFX Pro at $16.58 a month on annual billing, in line with our ~$17, but its pricing page was blocked to our fetch, so the earlier caveat stays on the page. Every “prices verified” stamp now reads October 2026.
• 2026-10-08: Gamma vs Canva for client decks. Our consultants stack picks Gamma for decks while eight other stacks pick Canva, and the two had never been set side by side. The new page reads both vendors' pricing pages from a US location: Gamma Plus is $9 a month billed annually ($108 a year) and Pro, which adds custom branding, is $18 a month ($216 a year); Canva Pro is US$144 a year. The finding is that the cheaper plan stops being cheaper once you need your own branding. Research Edition, no hands-on test, and neither tool pays us.

What counts as a loggable change

• Pick added or removed from any stack, with the head-to-head reasoning.
• Demotion: a pick that got worse (pricing, quality, support) and lost its slot or gained a warning.
• Research Edition → Tested upgrades, when a stack's hands-on cycle lands and editorial ratings become logged results.
• Price corrections: our monthly check catches vendor changes; corrections are noted here, not silently patched.
• Mistakes. When we get one wrong, the correction is logged with the original claim intact. Trust is built in exactly these moments.

Want these as they happen? The weekly email carries every log entry. Most weeks it says "no changes," and that line is the product working.