Concept case study · Two versions · Independent work
Quick Help
Closing the distance between needing help and getting it.
A customer-facing concept written from the floor of a large-format home improvement store: a way to raise your hand from where you're standing, and a way for the store to route that signal without adding headcount.
Developed independently while working at a national home improvement retailer. Brand, store and product names have been removed.
Role
Independent research, synthesis and concept design
Versions
V1 physical stations → V2 app-adoption model
Output
Journey map, three-surface model, wireframes, pilot measures
The insight
It's a discovery problem at the front door, not a technology problem on the roadmap.
The app already solves this. Customers just don't know it exists.
Associates live in the store app on every shift. Most customers have it installed and have never heard of its in-store mode. Front-of-store signage for it is blocked, small, or out of sight — at exactly the moment right before the frustration starts.
So the question stopped being “what should we build?” and became “where is the doorway, and what is the smallest thing that gets someone through it?”
Research
Five methods, one consistent coding scheme.
Everything observed was recorded in the same structure so separate shifts could be compared rather than merely accumulated.
- 01
Aisle observation
Shifts spent watching how customers behave in the minutes before they ask anyone.
- 02
Intercept conversations
Short, in-context questions at the moment of hesitation.
- 03
Associate reality check
Every idea tested against whether a person mid-task could actually run it.
- 04
Signal coding
Each request tagged by trigger, department, and how it resolved.
- 05
Revision
Version two rewritten against the objections version one couldn't answer.
Customer journey
Where the help gap opens.
Select a stage to see the emotional state, the friction, and the handoffs that shape the visit.
Moment
Identifies a home repair, maintenance or installation problem.
Motivated, slightly uncertain
Pain points
- Knows the goal, not the product language.
Opportunities
- Let people shop by project instead of requiring product expertise.
Moment of delightSomeone asks what they're building before asking what they need.
Pain points
Four failures that repeat across every visit.
P1
The silent wait
Customers search for several minutes before they consider asking anyone. Asking feels like an imposition.
P2
The unowned request
Verbal handoffs and overhead pages produce no acknowledgement, no owner and no clock.
P3
The wrong expert
The first available associate is frequently not the one who knows the answer.
P4
The lost signal
A request that ends in a walk-out leaves no trace in any system, so it can never be fixed.
Concept
Three surfaces. Each one does exactly one job.
No surface is asked to do something it's bad at. A static code carries its own location; the cart is the only surface that moves, so it gets the one job that doesn't require knowing where it is.
Cart QR
Open the app
Every customer touches a cart, and the app's geofence already knows which store they're in.
One universal code, fleet-wide. No per-store version.
Entrance station
Open the app and orient
Front of store, strongest signal, the moment before frustration. Store map and directory in one place.
Needs the store's own map.
Department QR
Request help, already located
The code is the geolocation. A scan in roofing arrives tagged “roofing.” No GPS, no beacons.
Static, one per department.
Routing help without new hardware
If a customer asks for help inside the app, ask them where they are. The store map already exists; they tap the aisle. People know where they're standing — the phone doesn't have to.
Version 1 → Version 2
The correction log, kept in public.
Version one proposed physical help stations. The pushback was fair, and the revision is more interesting than the original — so the changes stay visible.
01
The cart QR has one job: open the app.
A moving cart can't be geolocated, so it can't route help. That objection was right — so I changed the job rather than defending the idea.
02
Kiosks were tried. That's not what this is.
Locator maps and push-button help stations failed on adoption. All of them predate the app already in the customer's pocket.
03
Scan-for-help in locked bays already works.
It gets meaningfully more use than the call button. That's the mechanism, and it's already earning — this extends it, it doesn't invent it.
Persona & handoff
A guide with a name, whose most important behavior is knowing when to get a human.
The concept wraps the entry point in a warm, plainspoken guide persona — the associate who's rooting for your project. Every assistant has an adoption problem; a familiar, already-loved character is the cheapest way through it.
“That's a bit beyond me. Let's get a real expert over to you.”
It never dead-ends a customer. This is AI routing to associates, not replacing them — and the handoff line is the most important thing it says.
Bilingual service
Not a translation feature. The difference between being served and waiting.
Spanish-speaking customers wait for a bilingual associate to come free. In the store observed — comparatively well staffed for Spanish — it was still a bottleneck: a single mulch request pulled a second associate off task during a rush; another was resolved with a translation app, charades and luck.
Describe-and-visual search in both languages means a basic request stops consuming two associates, and stops depending on who happens to be scheduled.
Opportunity areas
Prioritized by impact against operational effort.
The concept deliberately targets the quick-win quadrant: high impact, low operational disruption, testable in a single store.
Wireframes
Three views, one shared signal.
Customer · request
Associate · inbound
Manager · patterns
Pilot measures
How we'd know the customer half worked, in one store.
The concept is only credible if it comes with the measures that would prove or kill it.
App opens by surface
Unique campaign parameters per surface — cart, entrance, department — so we can see which doorway actually works.
Leading · weekly
Time to responder
Minutes from a routed request to an associate physically arriving, with acknowledgement in between.
Leading · per request
Requests resolved
Share of help requests that close, and the recurring ones that graduate into a signage or adjacency fix.
Lagging · quarterly
Lessons learned
- Customers don't ask for help early because asking feels like an imposition. The interface has to lower social cost, not just physical distance.
- Routing is the hard problem, not capture. A signal nobody owns is worse than no signal at all.
- Adoption beats invention. The winning move was pointing at software that already existed rather than proposing new hardware.
- Any concept that adds a task without removing one will be quietly abandoned by the floor within two weeks.
Future vision
- Request data becomes a live findability heatmap merchandising can act on without commissioning a study.
- Recurring questions graduate into permanent fixes: signage, adjacency changes, shelf-level decision aids.
- Safety prompts and project guidance meet the customer at the shelf, before the wrong product goes in the cart.
- Resolution becomes a recognized operational outcome, measured alongside task completion rather than against it.
Frontline Signal is the research program this concept came out of.
Read Frontline Signal