Investor update · September 2026
Summer's over. The engine is running. Now we tune it.
Eight months in, WikiSleep has gone from a rebuilt product with no attribution to a paid-acquisition machine that pays for itself, at record volume, with the next problem clearly diagnosed. Here's the full picture — the good, the leak, and what we're doing about it.
~1.2×
LTV:CAC in the US and Canada, up from 0.3× in May
830
US installs in August from Meta alone, up from 22 at relaunch
44–47%
Trial-to-paid conversion, holding steady all summer
+34%
Growth in paid subscribers over the last 90 days
TL;DR
- Paid acquisition pays for itself again. Over July and August, every dollar spent acquiring a subscriber in the US returned $1.18 in expected lifetime value; in Canada, $1.19. In May, right after the V2 relaunch, both markets were at roughly 0.3×. From here on, every acquisition dollar goes to these two markets.
- Volume is at all-time highs. 1,518 installs in the last 30 days at $2.97 CAD each. The cost per install actually fell while we raised spend 25% week over week — the opposite of what usually happens.
- The funnel is diagnosed. Installs work. Trial-to-paid works. The leak is getting people from install to free trial, and the fixes are already in TestFlight.
- We're not chasing 100,000 users right now. We're chasing the next few hundred, and the model below shows what that does to the business.
- B2B is starting to land. Sooke Harbour House is live, Hotel Zed/Accent Inns is in conversation, and 258 people are currently in a free WikiSleep trial through Therabox.
01 — Where we are
The year so far, without the spin
In May we launched V2 — a full rebuild with proper attribution, so every dollar and every tap is now measurable. The trade-off was starting the ad account from scratch, and the efficiency numbers cratered. Since then the trend has run one direction in the two markets that matter. Canada has been above breakeven for three straight months and the US for two — a dollar of ad spend now brings back more than a dollar of expected subscriber value in both. The UK never got close, and we've stopped spending there. From here, it's the US and Canada, full stop.
Lifetime value divided by cost to acquire a paying subscriber, all channels, by market. Above 1.0× means paid acquisition pays back. Canada had no paid spend in January. Source: WikiSleep August data, LTV:CAC tab.
Two things worth naming so you're not surprised later. First, US Meta — our biggest channel — slipped from 1.62× in July to 1.15× in August — not because the ads got worse (installs went up), but because the cost of getting an install to start a trial rose from about $50 to $65. That's the leak we talk about in the next section. Second, Apple Search Ads in the US is running well below 1× and we're restructuring it. The UK is off. Every dollar is going to the channels that work.
02 — The funnel
Two of three stages are working. We know exactly which one isn't.
For the first time, we can see the whole funnel with real attribution instead of guessing. It breaks into three stages, and the picture is unusually clean:
Install
22 → 830
Working, and scalingMonthly US installs from Meta, April to August. More installs per dollar the more we spend — which is rare, and a sign the creative and targeting have found their groove.
Start a trial
$50 → $65
The leakCost per trial start, July to August. People install and don't hit the paywall with intent. This is a product and messaging problem, and it's the one we're on.
Trial → paid
44–47%
Working, and steadyOnce someone starts a free trial, close to half convert to paid. That's the product doing its job: people fall asleep, and they come back.
Why I'm not worried about this: fixing a leak in the middle of a funnel is a known kind of problem with known tools — onboarding, notifications, paywall tests, email. Fixing a product nobody wants isn't. We have the first problem, not the second.
03 — The levers
A subscription business comes down to three numbers
Our team built a forecast model this month, and the thing that clicked for me is how simple the levers become once the product works and the data is real: how much we spend, what it costs to acquire a paying subscriber, and how many we keep. Everything we're working on maps to one of those three.
What surprised me is how little has to change. Shaving $40 off the cost of a subscriber or holding onto one more person in a hundred each month doesn't sound like much, but those gains apply to every dollar we spend afterward, and they compound month over month. A few small tweaks turn a business that breaks even into one that funds its own growth. And these aren't mysteries: we can see exactly where the leak is, we know which tools close it — onboarding, notifications, paywall tests, email — and every one of them is already in motion. Tightening this up is work we know how to do.
This is a live model. Drag the sliders and watch what happens. It starts from where we actually are today.
Twenty-four month forecast
US and Canada only. Starts from ~200 paid subscribers and today's unit economics, with each subscriber paying a blended $10 a month. The base case uses July–August averages on Meta, the channel we're concentrating on — not the best or worst single month.
Where the budget starts. We're around $4,500 a month today.
How fast we ramp. Capped at $50K a month. We've been growing spend ~25% a week recently, so 10% a month is conservative.
US cost to acquire a paying subscriber, in CAD. On Meta, the channel we're concentrating on, it ran about $100 in July and $120 across July–August; August's trial leak and Apple Search Ads pushed the all-channel figure to $150. Closing the leak takes it back toward $100.
Share of subscribers who leave each month. Our blended rate this year sits around 6–8%. Every point matters.
Hotels, subscription boxes, employer perks, at near-zero acquisition cost. One 28-room hotel is worth 15–20 a month; with the hotel group and box partners live, 50 a month is the floor we're building to.
What we'd have in month 24
Paid subscribers
Monthly revenue
Ad spend that month
Over the full 24 months
Total revenue earned
Total spent on ads
Net value created: revenue earned plus the future revenue in the subscriber base, minus ad spend
At these settings a subscriber is worth over their lifetime against to acquire — an LTV:CAC of , paying back in . Because a subscriber's revenue arrives over many months, a ramping ad budget runs behind on cash until spend flattens — that's the shape of every subscription business, not a warning sign. What matters is the asset being built: revenue earned plus the future revenue held in the base comes to against spent on ads.
Try this: leave spend alone and drag cost per subscriber to $100 and churn to 5%. $100 isn't a stretch — that's what Meta delivered in July before the trial leak opened up — and 5% churn is what notifications and win-back email are for. Then put it back and double the ad budget instead. The funnel fix wins — which is why we're doing it first. Then drag B2B to 100 to see what the hotel group is worth.
Assumptions: US and Canada only; revenue per subscriber is held at $10 a month blended across monthly and annual plans; spend is capped at $50,000 a month; B2B subscribers churn at the same rate as paid ones. "Net value created" is before app store fees and team costs — it shows whether the acquisition engine is building an asset worth more than it costs, not company profit. Future revenue in the base is subscribers multiplied by lifetime value at the churn rate you've set. This is a directional tool, not a promise. The real model, with channel-level inputs and bear/base/bull scenarios, is available to any of you on a call with me and the team.
04 — What's shipping
Every item on this list moves one of the three levers
Our first ever. Permission prompt on the third open, then nightly nudges back into the app. The single biggest tool we didn't have for turning an install into a habit — and a habit into a trial. iOS first; Android follows.
Spec is with the developers. We'll test timing, offer and copy on the paywall itself, measured directly in RevenueCat. This is the cleanest shot at cost per trial.
Every new sign-up gets a note from me with an offer to jump on a 15-minute call. Cheap, personal, and it's how you learn what's actually stopping people. Win-back emails to lapsed trials ship alongside it.
Two new ad formats in production this week, plus a new sound and brainwave library in the app — the white-noise style content our competitors lean on, which widens who we can win at the top of the funnel.
05 — Distribution beyond ads
The second engine, quietly starting
Paid ads are the engine we can measure to the penny. B2B is the one that runs in the background at near-zero cost: a rack card in a hotel room, an insert in a box, a perk in a benefits program. Each one is small on its own. Stacked, they're the difference between a good subscription business and a great one — and they're starting to stack. My working target once the hotel deals are live is a floor of 50 new subscribers a month from B2B, before we spend a dollar on ads.
06 — Where you can help
Hotels and hospitality. If you know anyone at a hotel group, a resort, or a short-term rental operator, that's my highest-value intro right now. The Sooke Harbour House playbook is a one-day setup we can repeat anywhere.
Employer benefits. HR leads or benefits brokers who'd trial a sleep perk. Same low-cost, high-intent logic.
The round. The pre-seed is still open on the same terms you came in on. If you'd like to add, or know someone who should see this, reply and I'll set up a call with me and the team to walk the full model. BC residents still get the 30% refundable EBC credit.