For Nelson. Pick a SKYZER product, see how often it's selling alone vs being bought together with other things, then play with the sliders to see how much extra SKYZER money you'd make if customers added it to more device purchases. Based on 260,041 customer receipts a year (116,910 of those had a device like a phone, watch, headphone, etc).
Three snapshots of today's situation. The 3rd box (device attach) is the one your target slider + forecast act on — it's the cleanest "extra SKYZER units sold" lever.
For each device type, this shows how often SKYZER currently rides along, and how much extra money you'd make if you pushed it to your target rate. Top of the list = biggest opportunity. Don't add up the columns — a basket with both a phone AND earbuds appears in both rows, so a column-sum overcounts. Use the headline numbers above for true totals.
| Device type | Buyers per year | % who buy SKYZER | % given SKYZER free | % no SKYZER (the gap) | Extra $/yr at target | Extra profit/yr |
|---|
Source: retail_sales table, YTD 2026 (1 Jan – 24 Jun), annualised ×2.12 (172 days → 365).
All paying baskets: 260,041/yr. Main-device baskets: 116,910/yr — receipts containing at least one main product (phone, watch, earbud, console, camera, etc., 28 categories).
Penetration (% of all baskets): how often this SKYZER family shows up in any paying receipt. Includes standalone customers who came just for a cable.
PWP attach (% of device baskets): how often this family rides along when a customer buys a main device. THIS is the lever for additive sales — standalone buyers were going to buy SKYZER anyway, attach buyers are net-new revenue.
v2 fix (Jun-25): earlier version summed per-main attach counts to get totals, which double-counted receipts containing multiple mains. Fixed by computing unique-basket totals via JOIN. Per-main rows are still correct individually for ranking, but never sum them to get a total — use the deduplicated headline numbers.
Lift formula: incremental units = (target % − current paid %) × main-device baskets · revenue = units × avg ticket · profit = revenue × margin.
Honest caveats: (a) Margin defaults are placeholders — Nelson should override with real per-family cost. (b) Lift assumes new attachers behave like current paid attachers (same avg ticket). (c) GWP "cost" not modelled (no cost data). (d) Doesn't model display real-estate competition with Joel's new sourced items.