ShopeyeQ

Working on ShopeyeQ was one of the most rewarding projects I've taken on, because the answer came from learning an unfamiliar domain rather than from interface craft. The ask was to design an iOS app for FMCG field sales representatives in Germany — the people who walk into a supermarket and check whether the products a brand is contractually owed shelf space for are actually on the shelf.

As the lead product designer, I was responsible for research, information architecture, user flows, the design system, and every high-fidelity screen. I worked closely with a product manager, the engineering team, and the ShopeyeQ stakeholders. I created flows, wireframes, and clickable prototypes to share the vision, design principles, and content strategy. This helped to conceptualise ideas, gain alignment, and drive decision making.

Over the project timeline we ran regular design reviews with the PM and engineering, plus stakeholder reviews where higher-level business direction was set.

Salesforce and Zoho are CRM systems with a field-sales surface bolted on: the data model is the account, and a visit is an event attached to it. Repsly and Pepperi get closer to the field, but still treat a store visit as a form to complete. None of them treat the visit as the primary object, a bounded window of time, in a physical place, containing several different kinds of work with different completion criteria. That reframe drove the entire information architecture, and it is the single most useful thing the research gave us.

Grey boxes, real labels, no colour. The point of this pass was to argue about hierarchy and step count without anyone reacting to a shade of blue. I laid out the task list, PoS navigation, the scanning session, and the CRM handoff at this fidelity and walked them end to end. Iterating here was cheap, which is the whole argument for the stage, restructuring a scanning session takes minutes in grey boxes and half a day once type, colour, and components are attached.

Type, colour, iconography, and real component states. This is the pass where the delta stopped being a number in a list and became a red-badged tab, where travel time replaced distance on the map pins, and where the countdown found its home on the PoS sheet rather than the assignment list.

Clickable prototypes came out of this stage and went to stakeholders and reps. Most of the decisions in the next section were settled here, in front of someone tapping through a phone.

Map pins read 8 min, 10 min, 16 min. Not kilometres. Distance is the wrong unit for a decision about what to do next — a store 3 km away across the Elbe is farther, in the only sense that matters, than one 6 km away on the ring road. Assignment cards carry the same logic, with Duration and Travel sitting side by side so the rep compares the total cost of a visit rather than its position in a list.

The PoS sheet separates opening hours (09:00–22:00) from active hours (12:00–17:00) — the window when store staff actually permit shelf work. This distinction doesn't exist in general-purpose CRM tools, and it is the single most useful piece of domain knowledge in the product. Paired with it is a countdown in warning red, Closing in 02:30:55, ticking on the PoS card.

This is the centrepiece. When scanning finishes, results split into two tabs: Scanned — 85 in neutral grey, and Portfolio delta — 9 in error red. Eighty-five products are fine. Nine are contractually supposed to be on that shelf and aren't. The nine are the visit.

Everything in the treatment pushes attention there: red tab underline, red count badge, a number small enough to act on. The delta view groups missing items by portfolio and shows each as a product card with brand tag, image, and full SKU name, because “Messmer Apricot Peach Fruit Tea, 20 bags” is what the rep says to the store manager, not a product code. The two actions are Back to scanning and Submit, so a rep can re-enter the aisle, find what they missed, and return.

Route and upsell suggestions live in a distinct mode the rep switches on, not a permanent condition of the interface. Recommendation systems in field tools have a credibility problem: a rep who has worked a territory for seven years knows things the model doesn't, and a route suggestion that ignores the Tuesday loading-bay closure destroys trust in every suggestion after it. A toggle makes the feature earn its place, and gave us a clean way to compare behaviour with it on and off.

Barcode scan and AI shelf recognition exist as separate flows. Barcode is precise and slow, one SKU at a time, reliable in bad light. AI recognition is fast and probabilistic, good for sweeping a shelf facing. Rather than picking one, the app offers both, because shelf conditions vary more than any single capture method handles.

A full Break section with fourteen states sits in the bottom navigation beside Jobs, Learning, Products, and Profile. Putting rest at the same level as the core work was a stance. Field reps are frequently measured on utilisation, and a tool that treats rest as an exception state communicates something about the employer. Making it first-class also gives the rep clean documentation of their own hours.

Campaigns are typed, because “is the promotion running?” means something different for each: Multibuy, Temporary Price Reduction, Brand Awareness, Second Placement Only, Bundle. Media placement is modelled separately again — Aisle Fin, Radio, Digital Signage.

When a promotion isn't running, the rep files a missing-promo report: take photo, photo metadata, submit. Photo plus automatic metadata rather than a text field, because the brand paying for the campaign needs evidence with a timestamp and a location, not a rep's recollection.

Assortment history with its own portfolio delta, promotion history, upsell history, shelf audit history. A rep walking into a store sees what happened last time before they start. This is what makes the tool a record rather than a form. It is also the first thing that gets cut for scope, which is why I built it across all four types rather than as one generic activity log.