Method
Editorial log
Every published page on this site, who reviewed it, when, and what the review changed.
Scaled content is penalised when it is published without review. Rather than assert that review happens, the record is kept here and the build will not publish a page whose review block is missing or dated before the page was written.
One collection is reviewed differently and the table says so rather than hiding it. Market pages carry no third-party figure — every publishable fact on one is a named authority, a section number, an institution or a link — so each is checked by retrieving the source and searching it, and the URL, the status, the hash and the matched passage are recorded. A field that cannot be established that way is dropped; a page whose required fields cannot be is not published. Everything that carries a figure somebody else measured is still read by a person before it ships.
| Page | Type | Reviewed by | Reviewed | What the review changed |
|---|---|---|---|---|
| How fast should a practice reply to a patient enquiry? | Answer | Siddharth Sharma | 2026-08-06 | Rewrote the automation answer twice. The first version implied the systems described here satisfy a practice's privacy obligations, which is not something this business supplies and not something a marketing page should imply — it now states the boundary as a design constraint and sends the agreements to whoever advises the practice. Added the full-practice question, which is the most common real objection and the one where the honest answer is counterintuitive. |
| How much should a small business spend on marketing? | Answer | Siddharth Sharma | 2026-08-06 | Deleted a paragraph recommending a specific split between media and fees. No sourced figure supports one and it would have read as advice dressed up as a benchmark. Added the gross-versus-net question, which is the single most common way this band gets misapplied and produces a budget that cannot fund the work. |
| Is it cheaper to do marketing in-house or hire an agency? | Answer | Siddharth Sharma | 2026-08-06 | Added the fourth paragraph making the case for in-house. Without it the page was an argument for hiring me wearing a question's clothes, which is exactly the register this page type exists to avoid — and the conditions are specific enough that a reader can check whether they meet them. |
| Should you measure average response time? | Answer | Siddharth Sharma | 2026-08-06 | This page states the standard the research layer holds itself to, so it was reviewed against that page rather than on its own. Removed a worked example with invented response times — illustrating a percentile with made-up numbers on a site that requires sourced ones is the wrong trade even when nothing is claimed. The low-volume question was added after realising the whole argument assumes a sample size most readers do not have. |
| What does a legal lead actually cost? | Answer | Siddharth Sharma | 2026-08-06 | Cut a paragraph estimating how much of the lead cost is recoverable — there is no sourced figure for it and the estimate was doing persuasive work rather than informative work. Added the practice-area caveat to the FAQ after re-reading the opener as though it were a quote for a specific firm. |
| Which leads should you call first? | Answer | Siddharth Sharma | 2026-08-06 | The first draft read as an argument against lead scoring altogether, which is not the position and is not what the sourced figures support. Rewrote it as an argument about where qualification sits in the sequence, and added the structural exceptions — they are the part a reader can act on immediately. |
| Why did so many programmatic sites lose rankings? | Answer | Siddharth Sharma | 2026-08-06 | This page describes the rule this site's own build is gated on, so the first draft was closer to a description of our architecture than an answer to the question. Removed the internal references and kept only the parts a reader can apply to their own site. Added the self-test question, which is the most useful thing on the page and was originally buried in the third paragraph. |
| Why do home services clicks cost more than average? | Answer | Siddharth Sharma | 2026-08-06 | Removed a claim that missed calls in this industry are never recovered — some are, and the honest version is that they are recovered rarely enough not to plan around. Added the marketplace question, which is how a majority of smaller contractors actually buy leads and was missing from a page about click cost. |
| Why do manufacturing enquiries take so long to close? | Answer | Siddharth Sharma | 2026-08-06 | Cut a paragraph asserting a typical industrial sales cycle length. No figure in the sourced appendix supports one, and an invented range would have been the most quotable and least defensible line on the page. Replaced it with the mechanism that actually sets the timing — capital cycles and shutdown windows — which is checkable against any plant's own calendar. |
| Where Austin real estate leads are actually lost | Geography | Siddharth Sharma | 2026-08-06 | Every claim on this page was checked against a primary or association source before it was written, and one was wrong on the first pass: the MLS is Unlock MLS, not Unlocked MLS, and getting a market's own institution's name wrong is the single fastest way to lose a local reader. Dropped the association's own website from the resources list because it could not be confirmed to resolve — the association is named in prose instead, and only links that were verified are printed. Also cut a paragraph on median price and days on market: those figures are real and published monthly, but not by us and not in the sourced appendix, and a geo page that opens with a borrowed statistic is the pattern this layer exists to avoid. |
| Where home services enquiries are lost before the van is booked | Industry | Siddharth Sharma | 2026-08-06 | Cut a paragraph selling faster response as an unqualified good and replaced it with the capacity argument, which is the thing an operator in this trade would otherwise have corrected me on. Added the unaccepted-estimates section after realising the page described only the missed-call half of the loss. Removed a specific sales-cycle range for replacement work that could not be traced to a primary source. |
| Where patient enquiries are lost between the search and the chair | Industry | Siddharth Sharma | 2026-08-06 | Named HIPAA once, in the verifiableBy handle for the regulatory evidence point, where a statute is a stronger handle than a paraphrase of it. The framing is unchanged and deliberate: the regulation is the practice's operating constraint, never a credential or a service of ours, and the page still says outright that this is not a compliance service. Rewrote the privacy section after a first draft that read as an implied compliance claim — it now states plainly that this is not a compliance service and that vendor agreements are a separate matter. Moved insurance eligibility from a bottleneck mentioned in passing to the structural difference, because it sits upstream of everything else on the page. Dropped a patient-acquisition cost figure that could not be traced past a secondary synthesis. |
| Where legal enquiries are lost before the consultation | Industry | Siddharth Sharma | 2026-08-06 | Cut a paragraph implying this is a compliance service and replaced it with the narrower true statement about approved language per jurisdiction. Moved the conflict check from a bottleneck mentioned in passing to the structural constraint it actually is, and rewrote the response-window section around it. Removed a claim that legal is the slowest industry to respond — the sourced figure covers cost, not speed. |
| Where manufacturing enquiries die inside somebody else's process | Industry | Siddharth Sharma | 2026-08-06 | Removed a stated sales-cycle range that traced only to a secondary synthesis, and rewrote the section around what is observable instead — the cycle outliving the reporting period. Inverted the response-speed argument after the first draft imported the five-minute framing from the other hubs, which is wrong here and would have been the first thing an industrial seller dismissed the page for. |
| What running it yourself already costs you each month | Calculator | Siddharth Sharma | 2026-08-06 | Replaced a pre-filled example stack with zeroed inputs after checking whether any of the prices could be sourced — none could. Swapped the comparison anchor from my own entry tier to the published market floor so the tool is usable by someone who is not buying. Rewrote the first assumption once the real-estate hub began opening the software box at a sourced figure: the clause claimed the tool always opens at zero, which had stopped being true on one of the two surfaces that render it. |
| What a business your size typically spends on marketing | Calculator | Siddharth Sharma | 2026-08-06 | Moved the upper bound out of the headline output and into the arithmetic after the first draft led with it. Added the paragraph on cost per click being a national average, because the click figure is the part a reader in a dense metro will find wrong first. |
| What the enquiries nobody answered are costing you | Calculator | Siddharth Sharma | 2026-08-06 | Swapped the multiplier from an untraceable thirty-eight percent to the HBR-traced 23% and rewrote the deck, the body and the source answer around it. Added a fourth assumption clause: the figure was measured across firms, not across leads, and applying it to one business's enquiry count is an approximation that now says so on the page. Earlier in the day: removed an annualised output, and added the paragraph naming the answered-too-late population the tool does not count. |
| Are AI Overviews reducing clicks to websites? | Answer | Siddharth Sharma | 2026-08-05 | Reordered so the citation-versus-ranking finding leads the substantiation instead of the click statistics. Added the measurement question. |
| Does publishing your prices actually matter? | Answer | Siddharth Sharma | 2026-08-05 | Added the competitor objection, which is the one that actually stops people publishing. Cut an opening line that editorialised about agencies. |
| How fast should you respond to an online lead? | Answer | Siddharth Sharma | 2026-08-05 | Cut a paragraph that restated the close-rate figures a second time. Added the out-of-hours question, which is the objection that actually gets raised. |
| How long do real estate agents take to reply to a lead? | Answer | Siddharth Sharma | 2026-08-05 | Reframed around persistence after the speed figure alone read as a repeat of the general lead-response answer. Added the automated-sequence caveat. |
| Why do so many online leads never get a reply? | Answer | Siddharth Sharma | 2026-08-06 | Swapped the thirty-eight-percent multi-study figure for the traced 23%, which forced the question itself to change: at 23% of firms, "most" was an overstatement and the page was named after it. Added the paragraph explaining what the study actually counted, and an FAQ answering the thirty-eight-percent version rather than pretending it is not the one people have read. Originally published 2026-08-05 after removing an unsourced CRM adoption claim. |
| Where real estate leads are actually lost | Industry | Siddharth Sharma | 2026-08-05 | Split the buyer clock from the seller clock after the first draft applied one response target to both. Added the fair-housing and licensing answer, which was missing entirely and is the reason automated sequences get switched off here. Cut the closing paragraph's software recommendation and replaced it with the measurement-first argument. |
| Most sales need five contacts. Most businesses make one. | Solution | Siddharth Sharma | 2026-08-05 | Added the consent answer after realising the text-message path was described without it. Rewrote the closing paragraph to state the client's own time cost plainly instead of implying the sequences write themselves. |
| The average first reply takes 42 hours because the work is queued | Solution | Siddharth Sharma | 2026-08-06 | Corrected “median” to “average” in six places on 2026-08-06: the 42-hour figure is a mean, and calling it a median asserts a statistic nobody measured — which survives review, because a reader who checks finds 42 and stops. Earlier: rewrote the opening to stop describing the 42-hour figure as a failure of care. Added the virtual assistant comparison, which is the alternative this is actually bought against, and made the judgement boundary explicit rather than implied. |
What the build refuses to publish.
Four checks run before any page in this system is generated, and each one stops the build rather than producing a warning nobody reads.
A page must name, in writing, what is on it that could not appear unchanged on a sibling page — and no two pages may give the same answer. Independently of that, every page is reduced to a fingerprint with its own subject and place names masked out; if two fingerprints match, or come close, both pages are named and the build stops. That is the check that catches a page whose only distinguishing feature was the city in the heading.
A page must also carry at least two datapoints that are specific to its own subject rather than to its category, and if the same figure is claimed as specific on two different pages, it was specific to neither. Pages ship in cohorts with a size cap so indexation can be observed between them, and every page carries the review block that produces the table above.
All four are properties of the schema, not of anyone remembering. Reviewed by Siddharth Sharma, except the 1 market page(s) above, which are checked against their sources and carry the record of it.
GATES
A gate enforced by a person reading pages stops being enforced somewhere around page 60.