- May 31, 2026
- 6 min read
Pressure-Testing an Offer I Never Launched: A Heat-Pump Sensor Kit
Part 3 of the series. Part 1: the pipeline. Part 2: the first screen.
To test my landing blueprint I needed a real offer, not a “lorem ipsum SaaS”. I picked one close to what I was building anyway: sensors, a dashboard, and energy numbers.
The page built and could be deployed. The offer never launched. I kept experimenting with sensors and solar instead, which is a different story. This post is the review: what the blueprint made me write, and what I’d change.
The offer
Aerothermics Calibrated, tagline “Measure, don’t model.”
For homeowners comparing installer quotes for a heat pump plus solar, a $20k+ decision. The installer’s numbers come from a model of a generic house. The offer: measure your house instead.
- Order: pay, upload 12 months of electricity bills. A kit arrives with your Wi-Fi already set up: a Pico W gateway and three DHT22 temperature and humidity sensors.
- Measure: six weeks, unattended, with a dashboard to peek at.
- Report: send the kit back, receive a feasibility report (heat loss, expected efficiency, payback) and a 30-minute walkthrough.
About eight weeks from order to report. These were the test prices:
| Tier | What | Price |
|---|---|---|
| Entry | 30-minute call to pressure-test an installer’s quote | $235 |
| Core | Kit, six weeks of measurement, bills, report, walkthrough | $799 |
| Next step | Keep the kit, measure again for 8 weeks after the install | $1,499 |
The score: 34 out of 50
| Trait | Score | Why |
|---|---|---|
| Sticky | 4 | a household makes this decision once |
| Expensive | 7 | decent margin on the core tier |
| Expansion | 9 | heat-pump retrofits are growing across Europe |
| Low capital | 6 | kits are inventory, and they travel both ways |
| Unique | 8 | measured, not modelled, and no installer commissions |
No fatal disqualifiers fired. But by the blueprint’s own gate, 34 is “build, but say the weakness out loud”, not a clean go. At the time I wrote “proceed”, which is the kind of rounding up the gate exists to prevent.
The weakness is obvious: nobody buys this twice. The “next step” tier isn’t recurring either; it’s a deeper one-off. The realistic substitutes are referrals, and selling through energy consultants and installers who’d use the report with their own clients.
The best part: the damaging admissions
This is where the blueprint earned its keep. Forced to write who the kit is not for, I ended up with the most convincing section of the page:
- Under $5,000, don’t buy it. The free estimate is good enough at that scale. Save your money.
- It takes eight weeks. If your installer wants a decision next week, this isn’t for you.
- Not for hobbyists. The sensors cost a few euros each. You’re paying for the report, not the hardware.
- It costs more than a free quote. Their quote is a sales document; this is a measurement.
- It won’t pick a brand for you.
- No app, just a web page.
- Wi-Fi or nothing, mostly.
Each one removes a customer who’d be unhappy, and makes the rest more believable.
The FAQ had my favourite line, on why cheap sensors: about ±0.5 °C is enough for this calculation, and a sensor ten times the price wouldn’t change the report. Cheap sensors, smart math.
The weakest part: the guarantee
There were two promises:
- If the kit doesn’t deliver at least five weeks of valid data in the six-week window: full refund.
- If the installer’s quoted efficiency is more than 15% off my report’s prediction: full refund, “regardless of who was right”.
The first is a good guarantee: precise, measurable, and about something I control.
The second sounded bold, and it’s backwards. It pays out exactly when the report disagrees with the installer, which is when the report is most useful. It rewards me for agreeing with the quote. I’d replace it with promises about the work itself: data completeness, the report on time, one revision included, and a pro-rata refund if you stop halfway.
Also weak: the value stack
The core tier listed every component with a “value” (the report “$1,500”, the call “$235”, the firmware “$150”) adding up to about $2,850, “bundled at $799”.
Hormozi’s value stack works when the numbers are believable: what the customer would actually pay elsewhere. Mine were numbers I picked. A homeowner can’t compare “firmware value” to anything. A better anchor was already in the copy: $799 against a $20,000–50,000 decision. That’s the comparison people make anyway.
What I’d add
A season. Measuring how fast a house loses heat needs a real difference between inside and outside, and a known amount of heating going in. In late spring that difference is small. The offer as written ignored this; it’s really an autumn-and-winter product, and the page should say so. That’s the missing damaging admission.
Honest error bars. One FAQ answer said models can be 20% wrong and “measurements can’t”. Measurements have errors too; the difference is that you can see them. The report promised confidence intervals; the copy should have matched.
What the page looked like
Built with the “brutalist” preset from the blueprint: off-white paper, black type, one red accent, dense information. It fitted an offer whose whole pitch is “here are the numbers”. The default for a premium service would have been softer; the override was the right call.
- Headline: “Know if heat pump + solar pays off before $20,000 says it didn’t.”
- Without: an installer’s spreadsheet, guessing your heat loss, a retrofit you can’t return.
- One button, and “Five weeks of data or full refund. No installer fine print.” under it.
That first screen followed the checklist. What it didn’t have was proof: no testimonials, no case study. The plan was ten beta kits before raising the price. It never got that far, and the page was honest about it: the proof strip only said what was true.
Takeaways
- Test a blueprint on a real offer, ideally one you understand technically.
- Don’t round a borderline score up. 34 means “say the weakness out loud”.
- Damaging admissions sell. They were the strongest section.
- Guarantee what you control, not whether you disagree with someone else.
- Use anchors the buyer already has, not made-up component values.
- Check the physics before the copy. A measurement product has seasons.