Electro-Mart

Shop floor simulator 3D behaviour lab

Running · illustrative assumptions

00:00SIMULATED TIME

Scenario

Opening the shop floor…

In store0

Outside0

14.4 × 24 m schematic · space reserved between peopleTwo mobile assistants + three counter staff

0Purchases completed

0Visits completed

0Left early / gave up

0Helped by assistants

0Plan changes

—Mean wait to service¹

—Mean finished visit

Ten ways to shop

Green ring: activity time remainingAmber / red ring: patience remaining“?” badge: waiting for a mobile assistant

Model settings & observed data

Not calibrated: all inputs are assumptions. Changes affect new arrivals and activities.

Arrivals per hour Walking speed · m/s Browsing time · seconds Checkout time · seconds Service time · seconds Advice time · seconds Assistant help · seconds Patience · seconds Purchase propensity multiplier Finding-error probability · 0–1 Shared places per display · 1–6 Counter places incl. service · 1–8

Timing inputs are base means before persona differences. Capacity changes affect new admissions; existing customers keep their places. Each counter has one server. Restart retains current inputs; scenario presets begin a new run with assumed inputs.

Import real observations (Excel up to 5 MB; CSV up to 1 MB)

How the model works & CSV columns

Rows may contain any of these columns. Leave unknown values blank. Durations and walking speeds are sampled directly from imported values. Purchased is 0 or 1; a named persona gives a persona-specific base purchase probability. A blank persona applies to all shoppers who have browsed. Help and frustration still modify that probability.

interarrival_s, walking_m_s, browse_s, checkout_s, service_s, advice_s, help_s, patience_s, persona, purchased

Patience observations should measure an abandonment threshold, not the wait of a shopper who was served. Importing replaces previous samples; editing a timing input clears that field’s samples. Other inputs remain assumptions. Persona mix is balanced across each ten arrivals.

Displays have shared browsing positions. Checkout, service and advice use FIFO queues. Each mobile assistant has an independent request queue and moves towards demand. Shoppers can reconsider crowded displays, seek help after a product-finding error, abandon queues, continue unaided or leave. Movement uses 0.8 m reserved grid spaces; people cannot overlap, although movement and congestion are still simplified. There is no artificial customer headcount limit. Up to 16 people are drawn outside; the Outside count includes everyone waiting.

Paths record the whole visit. The last 50 departed shoppers remain selectable, plus any selected older shopper. Simulation time advances in fixed steps independently of display refresh. The 1× / 5× / 20× / 60× buttons change playback, not walking or service assumptions. Background tabs pause the clock.

¹ Mean wait includes completed waits at counters and for mobile assistance. Abandoned waits are excluded. Purchases excludes collection-only transactions. “Left early” includes customers who abandoned the entrance queue. Visit length includes normal and early departures after entering. This model has not been validated against a real store.