Four stages: audit against research rather than opinion, price every leak in dollars, ship the fixes in Shopify, then reconcile what was claimed against what came back. Thirty days, start to reconciliation.
Every leak on the ledger has to be traceable to something outside my own opinion. That’s the whole point of running it as a ledger.
Your store gets scored across all 10 page types — homepage, navigation, collection, product, cart, checkout, post-purchase, mobile, technical and accessibility — against the body of conversion research that already exists. Baymard’s 150,000 hours of UX research. Conversion.com’s repository of 20,000 experiments. Invesp’s 36,600. The industry has been documenting what works on a DTC product page and checkout for fifteen years; the work here is establishing which of it applies to you.
Then a competitive benchmark: three direct competitors, compared on the specific elements that drive the purchase decision — not a general vibe check. When someone leaves your page for the brand next to yours, this is the section that tells you what they found there.
You get the ledger inside 72 hours of kickoff. Not three weeks. It’s a diagnosis, and a diagnosis you wait a month for is a bad one.
A list of UX suggestions asks you to rank problems by taste. A priced ledger ranks them by recoverable revenue, which is the only ranking a founder can actually act on.
Each line is built from your own numbers: how many sessions hit that step, what share drop out there, what a recovered session is worth at your AOV, and how much of that gap the documented fix has been shown to close. The arithmetic is on the page for every line, so you can argue with any of it.
This is the part nobody under $7,500 a month does, and it is the entire reason the previous audit you paid for didn’t move revenue.
You see a before/after mockup of every priority fix first, so nothing ships that you haven’t seen. You approve or reject — rejecting is fine, it just comes off the ledger. Then I implement in a duplicated theme, review it with you, and publish.
Your total obligation across the 30 days: one kickoff call, and one approval pass on the mockups. No dev ticket, no sprint, no “let me check with our developer.”
You also get GA4 funnel tracking and a revenue-per-session baseline installed properly. That stays yours permanently, whether or not we ever work together again — it’s the instrumentation that makes the next 12 months legible.
Revenue per session on the pages I touched, 30 days after launch against 30 days before. Line by line, what I said it was worth and what it actually returned.
Some lines will beat the estimate and some will miss it. Both go in the report. A reconciliation that only shows the winners isn’t a reconciliation, it’s a case study — and you can already tell the difference.
If revenue per session hasn’t increased, this is also the call where I refund you in full. The conditions are published, in advance, so there is nothing to negotiate at the end.
I’m not against testing. I’m against selling it as the entry product to a store that hasn’t paid down its known debt yet — and to a store that mathematically cannot reach significance on the metric being tested.
At 95% confidence and 80% power, detecting a 10% lift needs roughly this much traffic. The floor moves dramatically depending on which metric you test:
| What you measure | Typical baseline | Sessions needed · 10% MDE |
|---|---|---|
| Session → purchase | 2.5% | ~125,000 |
| Product page → add to cart | ~10% | ~29,000 |
| Checkout step → complete | ~45% | ~4,000 |
| Revenue | Sessions / mo | Orders / mo | Purchase-rate test at 10% MDE |
|---|---|---|---|
| $1M | ~48,000 | ~1,200 | 2.6 months — not viable |
| $2M | ~95,000 | ~2,400 | ~1.3 months — marginal |
| $3M | ~143,000 | ~3,600 | ~1 month — workable |
| $5M | ~238,000 | ~6,000 | ~2 weeks — viable |
| $10M | ~476,000 | ~12,000 | ~1 week — real programme |
Assumes roughly $1.75 revenue per session (2.5% conversion at $70 AOV). Verticals vary — food and beverage runs nearer $1.18, electronics nearer $1.95 — so this gets recomputed for your store, not assumed.
Two consequences. First: below roughly $3M, a sitewide purchase-rate test takes so long that the answer arrives after the season it was meant to inform. Second: proximate metrics are far cheaper to test — but they can move opposite to revenue, more add-to-carts and fewer purchases. So a proximate win never counts until revenue per session confirms it didn’t regress.
Once the known debt is paid down and you’re above roughly 140,000 sessions a month, you’re at the frontier and testing is the correct next move. That’s the tier-2 programme — and if you’re not there yet, I’ll say so.
One kickoff call, Shopify and GA4 access, and the first version of your ledger lands three days later — priced, ranked, and with the arithmetic shown.