Why this page is a checklist and not a ranking
An earlier version of this page ranked six platforms and concluded that one of them was the industry benchmark worth acquiring. It carried market-size figures, growth rates, technology tags and a monthly price band for each platform.
We removed all of it, and the reason is worth stating precisely: the claims could not be verified, and in one important case we checked and found them false.
Specifically: the previous version attributed computer-vision posture analysis and AI-driven interactive feedback to taichiwuji.com. We fetched that site’s own homepage and searched it. Whole-word AI returns zero matches. computer vision, camera and pose return zero matches. The 102 apparent hits for the substring “ai” are ordinary words — available, email, maintain. Two uses of “posture” are ordinary prose in lesson descriptions. The technology claim was false, and it was the load-bearing reason for the ranking.
That is not a reason to stop evaluating these platforms. It is a reason to evaluate them differently.
Ownership disclosure: FitnessNav operates taichiwuji.com. We appear here. Nothing on this page is derived from our own operating data — no traffic, no revenue, no retention — because we do not publish those and would not if we did.
The governing principle: in this category the number of unverifiable claims is the investment risk. A platform that publishes its pricing, its refund terms and its named instructors has already told you the three things that decide whether the business works.
What is publicly verifiable in this category
Four things, and only four, can be established from outside a company:
1. The published price, and the published change to it. If the pricing page is public, you have the price. What most investors never collect is the change in that price over the life of the business. A platform that has never raised its price is telling you its acquisition cost is low. A platform that has raised it twice is telling you something about its retention. Neither fact is in a pitch deck; both are in an archive.
2. The refund and cancellation terms. This is the single most informative public document in a subscription business, and it is almost always a page nobody reads. Read three things: the notice period, whether cancellation refunds the current period, and whether the trial converts automatically. A trial that silently converts to an annual plan is a different business from one that asks.
3. Named instructors, with provenance a third party can check. “Ten generations of lineage” is not a provenance claim; it is a marketing sentence. A provenance claim has a name, a lineage that terminates in a documented person, and a body or association that a stranger can look up. If the instructors cannot be named, the content is not credentialled, whatever the page says.
4. The published curriculum structure. Order of progression, duration, and how a beginner is told what to do next. This is verifiable by reading, and it is the thing that determines whether a subscriber stays past week three.
What is not publicly verifiable, in any of these businesses: retention, churn, lifetime value, completion rates, revenue per subscriber, instructor productivity, and outcomes of any kind. These are the numbers that determine value, and they exist only inside the company. This is not a limitation of our research; it is the defining structure of the category.
The three claims that are almost never verifiable
These three recur in every deck we have seen a version of, and they deserve individual scepticism.
”AI-powered form correction”
The claim: the platform uses computer vision to analyse the user’s movement and correct it.
Why it is nearly always unverifiable: it requires the vendor to describe what the camera sees, what the model infers, and what threshold triggers feedback — and to say so on a public page. Vendors rarely do, because the honest description invites the obvious follow-up question about false positives, and about what happens to the camera feed.
How to test it. Ask for a demonstration with a deliberately wrong movement in frame, and ask what the user sees. A system that gives confident feedback on a wrong movement is a rule-based timer, not perception. Then ask for the model’s false-positive rate. A vendor who cannot state one has not measured one.
Status at our own site: we cannot make this claim. taichiwuji.com publishes no AI or computer-vision features, and our own site record says so. Any page of ours that said otherwise was wrong.
”Clinically validated outcomes”
The claim: users see measured health improvements.
Why it is nearly always unverifiable: a health-outcome claim requires a study. A study requires a protocol, an outcome measure, an ethics review where applicable, and — critically — a definition of the comparator. “Members report improved balance” is a testimonial aggregate. “Participants showed a statistically significant change in a validated balance measure against a control group” is a finding. Only one of those is evidence.
How to test it. Ask which outcome measure, against what comparator, with how many participants, published where. If the answer involves the word “users report”, you have your answer.
Note on what is real. Tai Chi’s association with fall prevention and balance in older adults is supported in peer-reviewed literature. That is a statement about the practice. It is not a statement about any particular platform, and a platform that cites the literature is borrowing credibility it has not earned. We have not verified the specific clinical-outcome claims made about any platform in this category, because the platforms do not publish the underlying work.
”Market growth of X%”
The claim: the category is growing at a stated annual rate, usually sourced to a consultancy.
Why it is nearly always unverifiable: the “category” is usually defined by whoever is quoting it. Definitions vary enough that two analysts can quote different CAGRs for the same year without either being fraudulent. A growth rate for a category you defined yourself is not a fact about the world.
How to test it. Ask for the category definition and the denominator. If the answer is a press-release number with no methodology, treat it as marketing. If it is a paid report, the definition is in the first chapter and is usually narrower than the slide that quoted it.
What to demand before price is discussed
Eight requests. None of them require the company to reveal a trade secret.
| # | Request | Why it matters | What a refusal tells you |
|---|---|---|---|
| 1 | Monthly cohort retention, twelve months, as a chart | Retention is the business. Everything else is commentary | The company does not measure it, or measures it and does not trust its own number |
| 2 | Gross margin by pricing tier, last four quarters | Revenue is a purchase; margin is a business | Pricing may be below cost in the tier you are buying |
| 3 | Instructor payment structure and the top-ten concentration | Content cost is usually the largest variable cost, and instructor concentration is usually the largest single risk | The content is owned by individuals, not the company |
| 4 | The full refund and cancellation policy as customers see it | Retention claims mean nothing against the actual refund rate | Ask for the refund rate, not just the policy |
| 5 | Named instructors with lineage that terminates in a documented person | Distinguishes a credentialled platform from a video library | The content cannot be credentialled |
| 6 | Paid acquisition cost by channel, last two quarters | Growth bought above lifetime value is a liability wearing a growth chart | Growth is being bought and is not profitable |
| 7 | Number of trial users who convert without a sales call, and the trial-to-paid rate | Separates self-serve product from assisted sales | There is a sales team, and the acquisition cost reflects it |
| 8 | What the subscriber can export, and at what fidelity | Data portability is the exit path for the subscriber and the acquirer | Subscribers are held, which is a strategy — but it is a strategy an acquirer pays for |
On the platforms discussed here
We examined the publicly visible material on six platforms in this category. We are not publishing a ranking, for the reason given at the top of this page: our earlier ranking was wrong, and the axis it was wrong on is the one that mattered.
What we can say:
taichiwuji.comis operated by FitnessNav. Its public site describes a programme structure, a knowledge map, a glossary, Zhan Zhuang standing practice, walking work including Cat Step, and dated articles on fundamentals such as Dantian with a self-check. It describes its own lineage as 15th-generation Wudang, taught by Master Gu — a lineage claim we cannot verify from outside, because it is our own. Consumers who want to know how to evaluate a tai chi platform should read our consumer-facing checklist, which sets out seven checks anyone can run.- The remaining platforms we could establish only the most basic public facts about: founding lineage, and whether the programme is instructor-led or library-based. Several sit behind bot protection that prevents us from reading them at all. Per our standard, “we could not read it” is recorded as unread, not as absence of merit — we make no claim that these platforms lack the features the previous version of this page attributed to our own.
What we could not verify
Stated plainly, because a page about verification that hides its own gaps is worth nothing.
- Retention, revenue, margin and lifetime value for every platform in this category, including our own.
- Any health outcome attributable to any platform.
- Any technology claim except where a platform describes it on its own public page.
- Pricing for five of the six platforms. Several do not publish a price publicly at all, which is itself a commercial decision worth noting.
- Whether any platform in this category is a good investment. That question requires financial statements we do not have and, in most cases, are not published.
We also do not provide transaction advisory, brokerage or valuation services, and nothing here should be read as a recommendation to buy or sell anything.
Related Reading
- Evaluating a Tai Chi or walking platform: seven consumer checks — the consumer-side counterpart to this page, and the one a potential subscriber should read first
- How we verify claims, and what we do not claim — the standard applied throughout this site, and the reasons the earlier version of this page failed it
- FitnessNav brand and category records — for manufacturers rather than instruction platforms
Correction notice: this page was rewritten 2026-10-06
The previous version was an investment-facing ranking of six platforms. It contained:
- A technology claim attributed to
taichiwuji.comthat its own website does not make. We checked. Whole-word “AI”, “computer vision”, “camera” and “pose” return zero matches on the site. This was the stated basis for the ranking. - A monthly price band for
taichiwuji.comthat we have no source for, and which the site’s own pricing page contradicts or does not support. - Market-size, growth-rate, regional-share and penetration figures for the wellness and virtual-fitness categories, sourced to consultancy forecasts without definitions. We do not hold the underlying reports.
- A projection that a technology layer increases ARPU by a stated range, presented as an established relationship.
- Technology tags — IoT, wearable sensor platforms, machine learning — applied to platforms whose public pages do not describe them.
- The superlative “industry benchmark” and the “#1 asset” framing for our own platform, with no ownership disclosure anywhere on the page.
- A framing addressed to investors and business leaders that provided no financial information capable of supporting an investment decision.
The honest version of an investor-facing page in this category is a due-diligence checklist, because the numbers that determine value are structurally inside the company. That is what this page now is.
If you saved the earlier version, please disregard it. If you can show us a claim we have removed that you believe is true, we want that — contact FitnessNav with the source document.
Methodology note: every claim on this page is either from a platform’s own published material, attributed to a named third party, or explicitly marked as unverifiable. We operate one of the platforms discussed and have disclosed it at the top of this page and in the footer. We hold no operating data for any platform in this category and have not ranked them, because our previous attempt at ranking them was wrong in a way we can document.
FitnessNav operates taichiwuji.com. No other platform discussed here has any commercial relationship with us.
