IG
The Vault
Innovative Group internal

Not quite. Try again.

Ask Chris if you need the passphrase.

Innovative Group
Internal · Three tracks · Rev A · Sep 2026

The Demand Engine

Four Tracks

The same machine, aimed at three different sellers. An individual selling online. A company selling to consumers. A company selling to a buying committee. And the businesses on your own high street. The five stages hold across all three. Almost nothing else does, and the places where they diverge are where most people get it wrong.

Crowd
Who, exactly, and where do they already gather?
Claim
What one thing will we be believed for, and what backs it?
Room
Where does the buyer decide? This one changes the most.
Ladder
What are the rungs, and what does each deliver?
Receipts
How do we know it worked, and can we prove what we said?

What changes between them

Read this table before picking a track. The row that surprises people most is the room. Track C has a clean conversion moment we can point at. Track A's room is an email sequence. Track B does not have one at all, which is exactly why its attribution is contested.

A · Individual onlineB · Company to consumerC · Company to committeeD · Local business
Who decidesOne person, alone, often at nightOne owner, spending their own moneyOne person, fast, on a phoneTen or more people over months
The roomA triggered email sequence. There is no page doing the selling.No single room. Conversion is spread across ad creative, product page and retention flows.A teaching session, then a conversation. The session qualifies; it does not close.A conversation in their shop, while they work
CycleDays to weeksMinutes to days10.1 months average BDays to weeks
Deadline worksYes, if the cart genuinely closesYes, if inventory or the season is genuinely finiteNo. They close on their budget cycle, not oursOnly your own capacity, stated honestly
Front-end economicsPaid acquisition barely returns cost: ROAS 1.19 in e-learning BROAS 1.88 platform-wide on Meta BFront end is not a revenue event at allThe free thing is the entire front end
What kills itNever reaching the audience size the arithmetic requiresNegative contribution margin at scale, hidden by growthReporting registrations instead of pipelineBuilding yourself a job rather than a business
Biggest regulatory exposureEarnings claims and testimonials under FTC §5Subscription cancellation, then reference pricingSubstantiation of performance claimsSubstantiation, plus local reputation, which travels faster
Evidence availablePoor. Most published numbers are platform marketingGood. Large platform datasets and public filingsModerate. Vendor benchmarks with disclosed samplesNone. Instrument your own from client one
An honest note on the evidence. Track B can be argued from SEC filings and datasets covering tens of thousands of brands. Track C has vendor benchmarks with disclosed samples. Track A cannot be brought to that standard, because the data does not exist. Almost every creator-economy benchmark in circulation is a platform surveying its own users, or an SEO page citing another SEO page. Where that is the case below, the page says so instead of filling the gap.
Pick one

The individual selling online

A solo creator, coach, consultant or course seller. No team. Their own audience is the whole asset, and the arithmetic of that audience decides everything else. This is the customer the operation we tore apart was selling to.

The ladder, with observed prices

  • Lead magnetFree
  • Low-ticket digital product$27–47
  • Course$95–200
  • Cohort or coaching$1,000–2,000
  • Done-for-you$2,000–35,000
  • Continuity$10/mo · $100/yr

Newsletter continuity at $10 monthly or $100 annual is the observed market standard. The rest are individually sourced and should be treated as observations, not a rate card.

The number that changes the plan

Median free-to-paid conversion is 0.62%. That is first-party platform transaction data across thousands of publications, and it sits roughly ten times below the 5–10% figure the platforms themselves market.

Work the arithmetic out loud, because nobody does. A 10,000-person list at 0.62% and $10 a month is about 62 paid subscribers, roughly $620 a month gross, before a 10–13% platform take and 5–17% monthly churn.

If one number from this track reaches a client, make it this one.

The room

A triggered email sequence, not a webinar. There is no product page doing the selling and no salesperson, so the sequence carries the entire argument.

Email converts at roughly 4.2% against 1.1% for paid social, and triggered flows outperform broadcast campaigns by a wide margin: in one large platform dataset, SMS flows were 7.6% of sends and 45.2% of SMS revenue.

There is no credible comparative evidence on webinar versus challenge versus DM funnel. Anyone who tells you otherwise is selling the method.

Platform take, actually

~14–17%
Effective Patreon take at $8–25 a month. The headline is 10%; the flat processing fee dominates at low tiers.
~13%
Substack, all in.
$179/mo
Kajabi, flat, no percentage.
~$990/mo
Where percentage platforms stop being cheaper than flat-fee ones.

How it actually fails

Not a bad funnel. Arithmetic. Creator earnings follow a power law with an exponent near two, which one peer-reviewed analysis describes as behaving closer to concentrated capital income than to labour income. The median is near zero and the average is meaningless.

More than half of creators earn under $15,000 a year, and that figure comes from a survey distributed through a platform, so the real distribution is likely worse.

The end often arrives as a payment processor shutdown rather than a slow decline: education and training chargebacks run 1.02%, above the level at which card networks start monitoring a merchant.

Urgency that survives scrutiny

Legitimate: a cart that genuinely closes and stays closed. A cohort that genuinely starts on a date. A price that genuinely rises and stays risen.

Not: a cart that reopens next week. A recorded session presented as live. A countdown that resets per visitor.

Note where the enforcement actually lands. One 2026 order against a course seller turned as much on burying refund terms and delaying refund requests as on the income claim itself.

The exposure on this track is the sharpest of the three The Business Opportunity Rule does not clearly cover coaching and course sellers today, and the 2025 proposal to extend it drew dissents from two commissioners, one of whom now chairs the Commission. None of that matters, because Section 5 already applies. Recent orders: a publishing course seller paid $1.5M in April 2026 over claims that buyers could copy a system making $1,000 to $3,000 a month, and over testimonials from employees and relatives of principals without disclosure. Related actions have run to $15.9M. Separately, Meta's own ad policy names "Be your own boss" and "Make $1,000 every month" as prohibited examples, so most copy in this category is non-compliant on its face and the account ban arrives faster than any regulator would.

The company selling to consumers

A DTC brand, physical product or subscription. One buyer deciding fast, usually on a phone. This is the track with the best evidence available and the least forgiving unit economics.

The ladder

  • Welcome offer, popup opt-in3.5% of displays
  • First purchase, median AOV$61.22
  • By category$43–131
  • Subscription, annual discount15–25%

Median AOV across paid channels comes from a dataset of more than 53,000 brands. For what a mature subscription base can look like: Chewy's Autoship is 83.3% of net sales, per their own filing.

The acquisition picture, and what it is telling us

$38.99
Meta CPA, platform-wide
$15.06
Meta CPM, up 13.2% year over year
2.39%
CTR, up 16.0%
1.53%
Conversion rate, down 4.7%
1.88
ROAS, flat

Read those five together. Cost up, clicks up, conversion down, return flat. Meta is delivering more clicks at a higher price that convert worse. Rising click-through alongside falling conversion is the signature of creative fatigue being masked by algorithmic click optimisation.

What the money actually has to do

Payback is the metric, not LTV to CAC. Pure DTC wants three to six months, tolerates six to twelve, and past twelve months is cash-intensive and fragile. Subscription should pay back in one to three.

Marketing runs at a median 13.3% of revenue across ten public DTC brands. Private brands under $50M typically need 20–35% because they have no brand pull. A useful cap: a third of gross profit until awareness exists.

How LTV gets inflated, in order of damage

30–65%
Using revenue LTV instead of contribution-margin LTV
40–80%
Counting only paid media as CAC
10–40%
Optimistic retention assumptions
5–15%
Ignoring returns and discounts
unbounded
Projected rather than realised cohorts

Returns are not a rounding error: 15.8% of retail sales and 19.3% of online sales, with 9% of returns fraudulent.

How it actually fails

Spending gross-margin dollars you do not have to buy customers who do not come back. The clearest case in the public record is Allbirds: a $4B peak valuation sold for $39M in March 2026, revenue down 48.8% from 2022, gross margin down twelve points, marketing peaking near 33% of revenue as sales were falling, never profitable as a public company.

The signature to teach: marketing as a share of revenue rising while revenue falls. It is visible in the filings years before the end.

Churn, and the cheapest fix in DTC

Monthly subscription churn runs roughly 4–7% for replenishment CPG, 8–15% for meal kits, 10–15% for apparel boxes. Annual plans hold far better than monthly: for supplements, 0.5–1.5% against 5–8%.

Involuntary churn, meaning failed cards, is 12–42% of all losses. It is the cheapest retention win available and it is consistently nobody's job.

The one audited churn comparison worth knowing: Peloton discloses 1.6% monthly on connected fitness against 7.0% on the app. Hardware lock-in retains four times better than content at a similar price.

Two things commonly reported wrong, and one narrower than people think Click-to-cancel did not go away. The Eighth Circuit vacated the 2024 rule in July 2025 on procedural grounds, not on substance. The FTC restarted rulemaking in 2026 and, more importantly, kept enforcing under ROSCA and Section 5 regardless: Amazon paid $2.5 billion in September 2025, after the vacatur. California AB 2863 has been in force since July 2025 and is stricter, requiring cancellation through the same channel used to enrol. Telling a client this rule disappeared would be malpractice.

Reference pricing is being enforced, just not by the FTC. It is private class actions: Boohoo $197M, RugsUSA $14.2M, Hot Topic $12.5M. California requires a reference price to reflect the prevailing market price within the previous 90 days. A perpetual sale is the exposure, and the numbers are larger than most agency fees.

The junk fees rule is narrower than assumed. In force since May 2025, but it covers live-event tickets and short-term lodging only. It does not reach general ecommerce. Correct anyone who says otherwise.

The company selling to a committee

Ten or more people, over about ten months, most of whom we never meet. This is the track the rest of the Demand Engine was built for, and the one where nearly everything we already have applies unchanged.

The ladder

  • Teaching sessionFree
  • Paid diagnostic$5–15k
  • Scoped project$45–90k
  • Retainer$7–30k/mo
  • Advisory$4–8k/mo

Price the diagnostic at 5 to 15% of the work it leads to, ten percent as the default. Forty-four named formats with published prices sit in the Rung Two catalogue.

The room

Forty minutes, monthly, replay open permanently. Retention runs 73% at 35 to 45 minutes against 51% at 60. The session's job is a qualified conversation, never a card transaction.

Give away the method completely. What we sell is execution, accountability and speed.

The numbers that matter

41.6%
Registration to attendance, B2B median
2.4×
Replay unique viewers against live
58%
Opportunities whose first touch was the replay
11.2%
Blended attended to pipeline
73%
Event leads going cold inside 72 hours

How it actually fails

Reporting registrations. This is the single most common failure when an agency imports consumer funnel mechanics into B2B: the funnel produces impressive counts and no pipeline, and everyone can see the dashboard is green while the pipeline is empty.

The second failure is delivering an audit without priced options, which is how a diagnostic becomes a shelved deck.

What does not port from A or B

Cart deadlines, countdowns and last-day mechanics. A B2B deal closes on the buyer's budget cycle, and an artificial deadline gets you excluded rather than accelerated.

The 90-minute pitch webinar. A working buyer will not give you two hours, and the long close block reads as a hostage situation.

Closing high ticket in the room. Anything above roughly $10k involves procurement and legal.

What ports the other way

From A: the discipline of a real deadline when a cohort genuinely starts, and the email sequence as a sales asset rather than a newsletter.

From B: contribution-margin thinking. Most agencies measure a B2B programme on revenue and never on what it cost to deliver, which is the same error that killed Allbirds in a different currency.

The local business near you

One owner, spending their own money, deciding in a conversation whether they trust you. The smallest version of a committee sale, and the one place where proximity beats budget.

The ladder

  • Something useful, free, once$0
  • A paid look at one thing$500–1,500
  • The fix, scoped$2,500–8,000
  • Keeping it working$500–2,000/mo

Four to six retainers is a living. That is the whole target, and it is closer than it sounds.

The room

A conversation, usually in their shop, usually while they are doing something else. There is no landing page in this business for a long time.

What converts is one specific thing about their business they did not know. Not a template audit: they have had three and all three were a pitch wearing a costume.

What is different from Track C

No committee, no procurement, no ten-month cycle. The decision happens in one conversation with the person whose money it is.

That cuts both ways. You can close in a week and you can be dropped in a week, and almost nothing about the relationship is contractual the way enterprise work is.

How it actually fails

You build yourself a job. Six clients who all need you personally every week is employment with worse benefits.

Then automation takes the rest. A productized audit at $750 delivered in 24 hours sets the ceiling on anything a machine can do alone. What survives needs a person in the room.

Urgency that is real

Your own capacity, stated honestly. "I can take one more build before December" is true, checkable and effective.

Nothing else. There is no cart here, and inventing one in a town where people talk to each other is worse than useless.

The evidence gap

No credible published benchmark set exists for local service businesses at this size. The bands here are observed ranges from published rate cards, not a dataset.

So instrument your own from the first client, because nobody is going to hand you the numbers.

Do not print these

Numbers that circulate widely in all three markets with no traceable origin. Several appear in competitors' decks. If one of these turns up in our work, it came from a content farm.