If a firm does not publish a plain record of its own facts, machines build one from scraps. The scraps may be public, flattering, and still wrong in combination.
The product had one name on the marketplace listing and another inside the founder interview. The parent company appeared in the footer, but not in the app directory. A customer story used the old brand line. A partner profile described the same business as workflow software, practice management software, and an operations platform within four short paragraphs. This is a composite scenario, drawn from the sort of B2B SaaS firm I often see in Singapore: about fifty people, selling to professional-service firms, growing faster than its public record can stay tidy.
When I asked an AI assistant to explain the company, it did something understandable. It stitched together the pieces. The answer named the product correctly, then treated the product as the company, then borrowed a feature from a similarly named vendor. It sounded calm. That calmness is part of the danger. A machine summary rarely shows the reader the little table of mismatched facts underneath.
The missing canonical page
Many founder-led firms assume the homepage is their source of truth. Sometimes it is. More often, the homepage is a persuasion page. It leads with a promise, a positioning line, a use case, or a market problem. It may be accurate, but it is not built to hold the boring facts that machines need to reconcile the entity.
A company source of truth page is a durable public page that states the firm’s canonical facts because machines, partners, directories, and buyers need a stable record to resolve identity. The “because” matters. This is not an about page in nicer clothes. It is a record built for retrieval, quotation, and correction.
The page should answer the questions a machine is already trying to answer from weaker sources. What is the company called? Is the product the company, or is it a product owned by the company? Which category should describe the firm? Where is it based? Who founded or leads it, if that is commercially relevant? Which services or products are in scope? Which old descriptions should not be treated as current?
I have come to think of this as a facts dock. Boats arrive from directories, partner pages, interviews, marketplaces, procurement portals, and AI answers. Without a dock, they tie themselves to whatever post is closest. A source-of-truth page gives them a safer place to land.
The page does not need to be long. Length can even weaken it. A clean record should be plain enough that a person can copy the correct sentence into a partner form without rewriting it.
What scattered facts do to a SaaS firm
The composite SaaS firm had a familiar split. The product brand was stronger than the legal company name in public. The founder was stronger than both in interviews. The marketplace listing was stronger than the site for feature language. The partner pages were stronger than the site for category words. None of these facts was surprising by itself. Together, they gave machines too many routes into the entity.
One AI answer described the firm as a workflow automation provider for law firms. Another called it a practice management platform for regional consultancies. A third merged the product name with the parent company and gave the wrong founding year, likely borrowed from a neighboring vendor. The error was not spectacular. It was the kind of thing a busy buyer might not question.
This is why source hierarchy matters. A machine may retrieve a marketplace listing because it is concise. It may retrieve a partner page because the partner domain is trusted. It may retrieve a founder interview because it contains richer language. The company’s own site does not automatically win. It must earn its place as the clearest available record.
The source-of-truth page helps by making the firm’s preferred facts easier to quote than the scattered alternatives. If the page says, in stable language, that the product is owned by the operating company and serves a specific category of professional-service firms, it reduces the temptation for machines to infer that relationship from marketplace crumbs.
I call this pattern entity gravity. A strong source-of-truth page does not control every answer, but it pulls related summaries toward a cleaner center. Weak gravity lets every third-party page tug the company in its own direction.
The record should be boring on purpose
There is a temptation to make every page carry brand voice. I understand it. Nobody wants to publish a page that sounds like a filing cabinet. But the source-of-truth page should have some filing cabinet in it. Not the whole room, just enough drawers.
The page should use direct labels. If the firm is a Singapore B2B SaaS company, say so. If the product is workflow software for professional-service firms, say so. If the parent company owns the product brand, spell out that relationship. If the founder’s public profile often causes confusion, write the founder connection in a sentence that can survive being lifted out of context.
A useful source-of-truth page often includes the official company name, trading or product names, short description, longer description, category, location, founder or leadership relationship, core product or service scope, audience, and controlled boilerplate. That sounds like a list because the underlying record is a list. The published page can still read as prose, but the facts need to be separable.
The important thing is consistency. Do not call the product a platform in one paragraph, software in the next, and a solution in the schema unless the distinctions are intentional. Do not switch between “regional firms,” “SMEs,” “law practices,” and “professional-service teams” without helping the machine understand the audience boundary. A human marketer may hear all four as harmless variations. A retrieval system may treat them as different clues.
For the composite SaaS firm, the rough but useful move would be a page titled something like “Company facts” or “About the company and product.” It would not replace the about page. It would sit behind the public record and quietly make claims stable.
The page must connect to the rest of the evidence
A source-of-truth page cannot sit alone like a clean note in a messy kitchen. It must be connected.
The homepage should link to it. The about page should not contradict it. The footer can use the official company name it contains. The product page should repeat the product-company relationship. The founder page should use the same category language. Structured data should mark up the organization and product relationship in line with the visible text. Partner boilerplate should be copied from the same record where possible.
This is where many firms stop too early. They publish a clear page, then leave old public evidence untouched. Machines see the new page as one more source, not necessarily the source. The hierarchy has to be signaled through links, repetition, and controlled reuse.
For partner ecosystems, the page can become a practical tool. When someone asks for a company description, send the canonical short paragraph. When a marketplace needs a category, use the same category. When a founder bio mentions the product, make sure the ownership relationship is the same as the source page. Small repetitions are not dull in this context. They are how identity becomes legible.
The aim is not to freeze language forever. Firms change. Categories mature. Product scope widens. But changes should move through the source page first, then outward. Otherwise the company becomes a house where every room has a different address on the door.
Source hierarchy beats content volume
When an AI answer is wrong, the reflex is often to publish more. More explainers, more use cases, more blog posts, more comparison pages. Sometimes that is useful. If the core record is crooked, more content spreads the crookedness.
I would rather see a firm fix the hierarchy first. Which page should machines trust for the company name? Which page should they trust for category? Which page explains the founder relationship? Which page explains the product and parent company? If those answers are unclear to the firm, they will be unclear to machines.
This is not a promise that one source page will correct every AI answer. Retrieval is uneven. Some assistants will still prefer third-party pages. Some prompts will pull in marketplace content because the user asks in product terms. Some summaries will keep a stale description alive because it appears on a strong external domain. The work is probabilistic, not magical.
Still, the absence of a source-of-truth page leaves the machine with no obvious correction target. If a partner page calls the firm a practice management platform and a marketplace calls it workflow automation, where should the assistant look to decide? The homepage might be too sales-driven. The about page might be too narrative. The product page might assume the parent relationship. The source page makes the decision easier.
A clean canonical page is one of the least glamorous assets a company can publish. That is almost a sign it is needed.
The practical shape of the page
I keep the page plain. A short opening sentence. A canonical description. A few compact paragraphs about the company, product, audience, category, and location. A founder or leadership note if it helps. A boilerplate paragraph for partners. A section that clarifies naming, especially when the product and company are not the same entity.
The page should be dated only if the firm is willing to keep it maintained. A stale “as of” date can become another weak signal. If dates are used, they should reflect a real review process. Otherwise, the better maintenance signal is consistency across the site and public profiles.
The writing should avoid inflated category words. Machines do not need a performance. They need an address label that matches the building. If the company is workflow software for professional-service firms, do not make the page reach for grander vocabulary. A clean noun is better than a large adjective.
For the composite SaaS firm, I would also add a naming note: the product brand, the operating company, and the founder’s relationship to both. The similarly named vendor cannot be controlled, but the company can make its own identity harder to merge. That is a sober ambition. Harder to merge is already progress.
The Entity Ledger Note — Observed name: a Singapore B2B SaaS firm whose product brand outranks the operating company in public evidence. Machine risk: AI summaries merge product, company, founder, and a neighboring vendor into one tidy but false record. Cleaning move: publish and interlink a durable company source-of-truth page. Residual fog: marketplace and partner pages may still dominate product-specific prompts.