Receipts
Retired sources
These are entries for claims this site no longer makes. Each one is kept word for word, with the date it was retired (three entries carry a dated note where one word was paraphrased), so the evidence stays on the record and the claim cannot quietly come back without it. The live register is on the sources page.
Retired entries
Claim retired
The leak section that carried this figure came off the home page on 29 Sep 2026, when the page was cut back to one offer. The figure is not used anywhere on the site now. The evidence stays on the record so the claim cannot quietly return without it.
Published source
Claim"The research is blunt: respond to an enquiry within the first hour and you are around seven times more likely to qualify it than if you wait even an hour more." Also shown as the stat card "~ 7 × more likely to qualify a lead when you make contact within the first hour, versus waiting even an hour longer."
What it saysFirms that made contact with an online sales lead inside the first hour were roughly seven times as likely to qualify that lead, meaning reach a meaningful conversation with a decision maker, as firms that waited even one hour longer.
Limitation It is 2011 data about online sales leads in general. It is not Australian real estate data, it is not my data, and it measures qualification, not sales, jobs, commissions or revenue. Response speed is one variable among many. Treat it as a reason to answer fast, not as a prediction of what will happen in your business.
Claim retired
The one-row salary comparison was removed with the maths section on 25 Aug 2026. The evidence stays on the record so the claim cannot quietly return without it.
Published source
Claim"An extra person to handle follow-up. $50,000+ /yr before super & leave" (the one-row comparison table).
What it saysThree independent salary sources put an Australian receptionist's annual salary between roughly $43,000 and $70,000, with averages clustering around $60,000.
Limitation I deliberately quote a number below every source, at $50,000 plus, so the comparison cannot be accused of inflation. Salary is also not the full cost of an employee, since superannuation, leave and on costs sit on top, which is why the site says "before super & leave". Most importantly, the row compares 24/7 call answering coverage only. An AI system is not a like for like replacement for a person, and the site says that in the footnote.
Claim retired
The Delivery Arm rung came off the public price ladder and out of /llms.txt on 29 Sep 2026. The site now sells one thing to the business owner who reads it. The price is unchanged in my own records; it is simply not offered on this site. The evidence stays on the record so the claim cannot quietly return without it. From the figure, as "$7,500 to $14,000 per white-label build, optional $900 a month", is listed again at S28 on the sources page only. It is still not offered on any sales page or in /llms.txt.
My own record
Claim"Delivery Arm. I build, your name goes on it, you keep the client. $7,500+ fixed, per build" (the ladder table), and the range "$7,500 to $14,000" given to AI assistants in llms.txt.
What it saysA fixed per-build price for white-label delivery, quoted to the agency rather than to their client. It starts at $7,500 and the upper end of the range reflects build scope, not an hourly rate.
Limitation This is a price I set and can change for future work, and the figure is a starting point rather than a quote. Nothing about it predicts what the build earns the agency or their client. It carried no entry on this page between 9 and 19 August 2026, which by the rule at the foot of this page means it should have been treated as wrong during that window. It is stated here now.
Claim retired
The live-proof section was removed on 25 Aug 2026. Its policy line now sits in the footer. The evidence stays on the record so the claim cannot quietly return without it.
My own record
Claim"This section doesn't run on testimonials I can't back. It runs on real system events and named client outcomes, which makes it honestly empty until the first install goes live." Plus "POLICY: No paid reviews. No invented numbers. Every figure carries its source."
Clarified . "No clients yet" above, kept word for word, meant no clients on the follow-up product. I have built websites for clients. I have no clients on this follow-up product yet, so I have no results of my own to show you.
What it saysThe proof section is empty on purpose. When the first result exists it publishes there with the client's name, their sign off and their own numbers, not my summary of them.
Limitation An empty proof section is honest, but it is still an empty proof section. It is a reason to be sceptical of me, and I would rather you were sceptical for the right reason than reassured by a fake one. That is also why the risk is bounded the other way: month to month, and you hear and approve your system before it ever speaks to a customer.
Claim retired
The interactive database illustration was removed on 26 Aug 2026 and replaced by a self-check that computes nothing. Both assumptions inside it were mine and neither was ever measured, which is why the section could not honestly produce a conclusion, and why it went. The evidence stays on the record so the claim cannot quietly return without it.
My own record
ClaimThe two assumptions inside the interactive database illustration: "62% of an aged list still has a working mobile" and "35% of conversations that become a booking". Also the default answer rate of 4%, which the visitor can change.
What it saysThey are the planning numbers I use when I size a database before quoting, and I have deliberately made the answer rate a control the visitor sets rather than a number I assert. The tool multiplies the visitor's own inputs by these two rates and shows the result. It describes arithmetic, not an expected outcome.
Limitation these are estimates I chose, and reasonable people would choose differently. A real database can be far worse than 62% reachable if it is old or poorly kept, which is exactly what the free audit checks before anything is sent. Nothing produced by the tool is a forecast, a target, or a promise, and no fee I charge depends on any of these figures being met.
Claim retired
The time-audit section was removed on 25 Aug 2026. The evidence stays on the record so the claim cannot quietly return without it.
Published source
Claim"40+ hours a week the average Australian business spends on work that could be automated" (the time audit).
What it saysThe figure covers data entry, follow-up emails, invoice processing, report generation and appointment booking across Australia businesses generally, not real estate specifically.
One word changed . A word this site no longer uses on any public page was replaced with a plain paraphrase in the line above. The meaning is unchanged; nothing else in this entry was edited.
Limitation it is a customer publication, not peer-reviewed research or an official statistic, and "the average Australian business" hides enormous variation. I use it to describe the shape of the problem, never to size a saving for a particular business. Nothing I sell is priced on it.
Claim retired
The time-audit section was removed on 25 Aug 2026. The evidence stays on the record so the claim cannot quietly return without it.
Published source
Claim"5 to 15 hours a week a working agent loses to repetitive admin" and "30 to 200 enquiries a week through a typical business" (the time audit).
What it saysThe admin figure covers writing property descriptions, sending contract documents, chasing tenant references, booking open homes and drafting agent introduction emails. The enquiry range covers jobs, appointments, rentals and general agent contact across a typical business week.
One word changed . A word this site no longer uses on any public page was replaced with a plain paraphrase in the line above. The meaning is unchanged; nothing else in this entry was edited.
Limitation both are industry commentary from an AI customer rather than an audited survey, and the ranges are wide enough that your business could sit outside them. They describe the category, not your business, and I do not use them to project a result.
Claim retired
The time-audit section was removed on 25 Aug 2026. The evidence stays on the record so the claim cannot quietly return without it.
Second recurrence, . The 25 August pass removed this from every page and missed /llms.txt, which is a page: it is served at a public URL and quoted verbatim by assistants. Both figures, 44% and 5%, were still published there in the answer to "Is AI actually worth it for a small Australian business". Removed 27 August. This is the same miss as the GST sentence, also taken off every page on 25 August and also left standing in llms.txt. The source below is sound and unchanged. What was wrong was the register saying the site had stopped citing it while one page still did.
Published source
Claim"5% of Australian small and medium businesses already using AI are fully enabled to realise its benefits" (the time audit).
What it saysThe same body reports SME AI adoption at 44% as at February 2026. The gap between those two numbers is the entire reason this business exists: adoption is common, competent implementation is not.
Limitation "fully enabled" is the source's own definition and survey self-reporting is unreliable in both directions. It says nothing about what any particular business will achieve.
Claim retired
These service commitments were removed from the site. A same-day callback is a response-time claim and this business does not make them. The evidence stays on the record so the claim cannot quietly return without it.
Narrowed . This entry retired four claims at once and gave a reason that applies to one of them. Only the same-day callback was removed. The under-two-years contact window is still on the site and is still true, and the two appointment lengths were re-entered as S19 on 27 August. So the register was contradicting sentences that are live and correct, which is the same failure as a claim outliving its entry, arriving from the other direction. The contact window is now entered on its own at S21. What stays retired here is the same-day callback and the ten-minute export, which was cut from the page on 27 August rather than re-entered: an estimate of how long somebody else's software takes, never timed, is a representation about a future matter with nothing behind it, and this entry was right about it the first time. Describing these four claims rather than quoting them is also why nothing could test this entry in either direction, and why the export sat on the page for two days after it was retired here, under a green tick.
Corrected . The note above says the under-two-years contact window "is still on the site and is still true". The second half never held: the ranking in the product scores contacts by age and drops nobody for being old, so two years is a ranking band, not a filter. The last sentence on the page that put it as a filter, the answer to "Is it legal?" under the price ladder, came off on 11 October 2026, and S21 now says what the code does. The note above stays exactly as it was written, because this register marks a correction rather than removing the thing it corrects.
My own record
ClaimTwo service commitments, both retired, quoted as the site used to write them: "a same-day callback before the day closes", and "One export, usually ten minutes of your admin's time". Neither goes back. The other two claims this entry used to bundle are live and are entered separately at S21 and S19.
What it saysThese are promises I keep, not measurements of anything. The callback rule is written so that it can never require a call outside the permitted windows.
Limitation they bind me, not the market, and they say nothing about what any campaign will produce. If I ever miss one, tell me and I will own it.
Claim retired
The interactive database illustration was removed on 26 Aug 2026 and replaced by a self-check that computes nothing. Both assumptions inside it were mine and neither was ever measured, which is why the section could not honestly produce a conclusion, and why it went. The evidence stays on the record so the claim cannot quietly return without it.
My own record
ClaimThe day timeline inside the interactive illustration: the slider maps onto a working day of calls from 8:30am to 6:00pm, the morning block ramps first, the arvo block lands late, and bookings lag conversations. Plus the phase captions attached to each position.
One word changed . A word this site no longer uses on any public page was replaced with a plain paraphrase in the line above. The meaning is unchanged; nothing else in this entry was edited.
What it saysThe shape is the day I actually run, not a measurement of anything. Dragging the control redistributes the same total across the day; it does not add to it. At knock-off the figures are identical to the untimed version, because the timeline describes WHEN the arithmetic lands, never how much of it there is.
Limitation real campaigns do not follow a clean curve. Replies cluster unpredictably, some people answer in week four, and plenty never answer at all. Nothing in the timeline is a forecast, a target or a commitment, and no fee depends on any figure it displays.
Claim retired
The seven-question check came off the home page on 29 Sep 2026, when the page was cut back to one offer: the demo you can run and the approval queue now do the job it did. Its script, js/check.js, is no longer loaded by any page. The evidence stays on the record so the claim cannot quietly return without it.
My own record
ClaimThe seven-question self-check that sits above the price, and the result screen it produces. The two questions in it that carry a figure, word for word: "When did you last ring somebody you quoted months ago, just to ask where it landed?", whose options are "This month", "In the last three months", "Longer ago than that" and "Never. Once it is quoted it is done"; and "Going back two years, roughly how many past customers and old enquiries are sitting in there?"
What it saysThe check performs no arithmetic on anything you tap. It multiplies nothing, forecasts nothing, and produces no dollar figure, no percentage, no score and no rating. It counts how many of five questions you answered with an option that names a gap, and hands your own answers back to you in the words you tapped. Two of the seven questions are about size only and cannot register a gap at all, because a check where every answer condemns you is a sales trick. It can and does return "on this evidence, you do not need me", and on that result the booking buttons are removed from the page rather than hidden. Nothing you tap is stored, sent anywhere, or written to your browser: no upload, no cookie, no local storage. Every word of the result is written in advance and ships in the page source. The only things that vary are which of the three results appears and which of your own answers is echoed back under it. It replaced an interactive illustration that multiplied a visitor's guess by two assumptions of mine, which is why S08 and S14 are retired above.
Limitation which answers count as a gap is my judgement, not a measurement, and I have put it in the page source rather than hiding it: the flagged options carry data-gap in the HTML and you can read them without asking me. The wording of the options is mine too, so the check can only reflect back a gap I thought to ask about. It is a prompt for a conversation, not a diagnosis of your business, and nothing it displays is a forecast, a target or a promise.
Claim retired
It was on the hero strip, in the same position and the same weight as two facts that are true and checkable ("your accounts", "a person sends"). That is what made it dangerous: it read as the third member of a set of verified facts. It is replaced by a term of the deal, "month to month, 30 days notice either side", which is in the agreement and is mine to keep.
My own record
Claim"10 days" / "signature to live", the third tile in the strip under the hero, AND "10 DAYS · FIXED PRICE", the length label on the build step of the how-it-works list. It was in two places in two different casings. The first removal found only the lower-case one, because the check that found it was a case-sensitive grep; the sweep found the second.
Claim, third recurrenceRemoved again on , this time spelled in WORDS rather than digits, from seven further places: "Live in ten business days" and "Ten business days." on the home page, "live in ten days" in the price ladder, the heading "How the ten days go" with its "Day 1 / Days 2 to 7 / Day 8 / Days 9 and 10" step labels underneath it, and three instances in /llms.txt including "What is fixed is the delivery: The Desk is live in ten business days", which had escalated it from a description to a commitment. Both truth gates reported the site clean throughout, because both matched digits: this entry quoted "10 days", the page said "ten business days", the similarity fell to 50% and the gate declared the live claim correctly absent. The 26 August lesson was to sweep in every casing. The real lesson is to sweep in every SPELLING, and the gates were changed the same day to normalise number words to digits before matching.
What it saysIt said an install went from a signed agreement to a working Desk in ten days. Nothing in the repository, and neither of the two builds done so far, records a signature-to-live time at all, so there was no basis to average and no basis to publish. It was found by a sweep that checks every figure on the site against this register, which is the opposite direction to the check that had been running.
Limitation a delivery time is a representation about a future matter under section 4 of the Australian Consumer Law, which reverses the burden of proof: the person who made it has to show reasonable grounds for it, not the other way round. No delivery time goes back on this site until real installs have been timed and the times appear here.
Claim retired
The wording went further than the handover list I give a client at the end of an install. It said you could run the system without me and that you were not tied to me. What the list actually covers is your accounts, your credentials, your data, your call sheet, your message templates and a written handover runbook, so that is all the site says now. The old lines came off the home page, /why.html and /llms.txt on 29 September 2026. The evidence stays on the record so the claim cannot quietly return without it.
My own record
Claim"Sack me and you keep the lot, and that runbook is how your own people run it without me." (the list under the price ladder on the home page, which also opened with the heading You own it outright), "You own the lot, and you are not locked to me." (the line under the hero), "You own it and you are not locked to me." (the share description), and "If they stop working with Craig they keep all of it, and it works exactly the same way without him." (/llms.txt). The price table's closing words, you own it, and the footer's closing words, yours to keep, said about the same thing in fewer words, and went with them.
What it saysIt said a client could run the system without me and was not tied to me. Nothing in my handover list promises that. The list covers accounts, credentials, data, call sheet, message templates and a runbook, and the site now says exactly that, in the same place.
Limitation this entry records a wording change and adds no promise. A claim about what happens after a client and I stop working together is a representation about a future matter, so the site makes only the one my handover list supports.