SensingDesign complete (stages 1–6), not built, not validated

StoreSense

Camera-free shop analytics that tells you what to change

Counts who comes in, where they wait and how many buy, then sends the owner a Monday email naming one to three things to change, each with a dollar value, instead of a dashboard.

Who buys it: Independent specialty retailers (boutiques, gift, books, home decor, apparel) and small chains of 2–10 stores on Square, Shopify POS or Clover.
Why it pays: At 100 visitors/day, a $40 basket and 50% margin, +1 conversion point is worth about $600/month gross profit, and moving one dead staff hour a day about $480/month (est., research assumptions).
Target price: $149 door unit + $69 per zone node + $15/month (target)
First-minute wow: A Monday email that says: "Saturday 2–4 pm you lost about 38 buyers (about $1,520): the till queue averaged 4 people. Add one person at the till 1:30–4:30 pm."

See the CAD design Concept page

StoreSense concept: the idea and where each part goes — The concept. The CAD design is below, under Printable CAD.

What the research found

And how this design answers it

One ToF sensor is too narrow at the door header → two VL53L8CX splayed ±20° on the ceiling give an 85° field covering doors up to 2.0 m
Sun through glass weakens ToF → sun-tolerant VL53L8CX, heads-only tracking, a "sun mode" and flagged hours dropped from advice
Groups side by side merge into one blob → 16 columns across the door split heads; reports count shopping parties
Children, strollers and carts inflate counts → height-based child and stroller rules, not counted as buyers
Installs need an electrician → USB-C 5 V or PoE splitter, T-bar clips or VHB; zone nodes can run on 4×AA
Owners stop reading dashboards → the product is a weekly email with at most 3 actions, and next week checks whether the change worked

Use case

Saturday you lost 38 buyers

Shop ownerPart-time staffDoor unit and till zone nodeSquare POSStoreSense cloud

01Trigger: the Monday 7 am weekly email is built
02Detection: 210 shopping parties entered Saturday 2–4 pm, the till queue was 3+ people for 41 minutes, and Square shows 38 orders: 18% conversion against an 8-week median of 36%
03Alert: the email says about 38 buyers (about $1,520 in sales) were lost to the till queue and suggests one more person at the till 1:30–4:30 pm
04Action: the owner adds a staff member for 3 hours; the next Monday's email reports whether it worked

Outcome: If half the gap is recovered, about $1,430/month net gross profit after the extra labour (est.)

How it works

Architecture

01
Sense: twin VL53L8CX depth sensors at the door and VL53L7CX zone nodes over the till, an aisle or fitting rooms (8×8 depth values only, no camera)
02
Decide: on-device height map → heads → tracks → line-cross counts at 15 Hz; zone occupancy and dwell
03
Send: zone nodes over encrypted ESP-NOW to the door unit, then MQTT over TLS; POS hourly transaction counts polled from Square, Shopify or Clover
04
Tell: rules turn counts and sales into a Monday email with up to 3 dollar-valued actions, plus an app card and CSV export

Hardware

Real parts, buyable today

Bill of materials: about $39 per door unit and $21–27 per zone node at qty 1; about $86 for a 1-door + 2-zone kit (est.).

Block	Part	Why this part
Door sensors	2× ST VL53L8CX 8×8 multizone ToF	Best ToF range under ambient light; splayed ±20° for an 85° field
Door MCU	ESP32-S3-WROOM-1-N16R8	Runs the 15 Hz tracker and hosts the zone-node link
Zone sensor	ST VL53L7CX (60°)	Wide field for indoor queue and aisle zones
Zone MCU	ESP32-C6-WROOM-1-N8	Low-power ESP-NOW link to the door unit
Power	USB-C 5 V or PoE splitter; zone node optional 4×AA L91 + TPS62840	Low voltage only, no electrician

Mechanical

Printable CAD

12 parts, generated from code and checked as closed solids. Pick a part to inspect it, or download the STL files to print.

See the CAD parts render

StoreSense printable parts, rendered from the CAD — Render of the printable parts from the actual CAD. Not yet printed or photographed.

Loading the CAD…

Real CAD · drag to rotate · scroll to zoomReal CAD · swipe sideways

Open risks we are working on

Low sun through glass doors: ST rates the sensor at 5 klx, direct sun can be about 20× that; a bench test decides whether glass-front shops are supported
Splitting side-by-side heads at a 7–13 cm zone pitch is untested; the ≥90% overall accuracy gate depends on it
Go-to-market against Dor and Avia: the edge is the zone nodes and instruction email, which is software and can be copied
Battery zone nodes last only about 2–3.5 months (est.)
VL53L8CX lead time (22 weeks at DigiKey) and VL53L7CX stock are supply risks
Research verdict was NICHE; the 30-day, <$500 pilot gates still apply

Design stage only: nothing on this page is for sale. Specifications and prices are targets.