Pika Retailer App

Pika Insights builds technology for Africa’s informal retail: neighbourhood shops, kiosks and market stalls that move a large share of everyday goods but are hard for brands to see. Three products work together. Retailers run their shops on the Retailer App. Field agents use the Agent App to onboard traders, run store surveys and record sales for them. Every sale and restock then feeds PikaView, the market intelligence dashboard FMCG companies pay for.Pika Insights builds technology for Africa’s informal retail: neighbourhood shops, kiosks and market stalls that move a large share of everyday goods but are hard for brands to see. Three products work together. Retailers run their shops on the Retailer App. Field agents use the Agent App to onboard traders, run store surveys and record sales for them. Every sale and restock then feeds PikaView, the market intelligence dashboard FMCG companies pay for.

So the Retailer App has two jobs at once: be useful enough that a shop owner opens it every day, and capture what happens at the counter accurately enough to become data a brand can trust.So the Retailer App has two jobs at once: be useful enough that a shop owner opens it every day, and capture what happens at the counter accurately enough to become data a brand can trust.

I joined Pika in December 2024 to design PikaView, the FMCG dashboard, while our previous designer worked on the Retailer App. When he left in December 2025, I took over design for all four of Pika’s digital products: the Retailer App, the Agent App, PikaView and Godview, and in April 2026 I began redesigning the Retailer App.I joined Pika in December 2024 to design PikaView, the FMCG dashboard, while our previous designer worked on the Retailer App. When he left in December 2025, I took over design for all four of Pika’s digital products: the Retailer App, the Agent App, PikaView and Godview, and in April 2026 I began redesigning the Retailer App.

The app I inherited opened on a “My Shop” dashboard with weekly revenue, items sold and a top customer. Recording a sale meant searching a master list of more than 20,000 products and setting quantities with a slider, and inventory was organised by fixed categories.The app I inherited opened on a “My Shop” dashboard with weekly revenue, items sold and a top customer. Recording a sale meant searching a master list of more than 20,000 products and setting quantities with a slider, and inventory was organised by fixed categories.

I am the only designer across Pika’s digital products, so I own the experience end to end: research with the retailer success team, flows and interface, the design system, and handoff to engineering. Working closely with retailer success is what keeps the design honest, because they hear from retailers every day.I am the only designer across Pika’s digital products, so I own the experience end to end: research with the retailer success team, flows and interface, the design system, and handoff to engineering. Working closely with retailer success is what keeps the design honest, because they hear from retailers every day.

“Being able to bridge between retailer success and product has really helped you refine your understanding of the user request and the informal retail sector. Your work cuts across product, retailer success and engineering.” Justin Mark, Retailer Insights Manager“Being able to bridge between retailer success and product has really helped you refine your understanding of the user request and the informal retail sector. Your work cuts across product, retailer success and engineering.” Justin Mark, Retailer Insights Manager

Before redesigning screens, I rebuilt the foundations. A small Pika UI kit became Pika Insights Foundations: the single source of truth for colour, type, spacing, shape, elevation and motion across the Retailer and Agent apps, with every value bound to a Figma variable. Change the token, and the system follows.Before redesigning screens, I rebuilt the foundations. A small Pika UI kit became Pika Insights Foundations: the single source of truth for colour, type, spacing, shape, elevation and motion across the Retailer and Agent apps, with every value bound to a Figma variable. Change the token, and the system follows.

Colour works in two layers, global primitives and semantic tokens, so a screen keeps working when the theme or brand changes.Colour works in two layers, global primitives and semantic tokens, so a screen keeps working when the theme or brand changes.

The home screen now leads with the two ways a retailer records business: “Start a new sale in seconds” and “Snap your logbook page”. Shortcuts to add products, manage expenses, view insights and find customers sit one tap away, with recent sales, the Pika Score and offers below.The home screen now leads with the two ways a retailer records business: “Start a new sale in seconds” and “Snap your logbook page”. Shortcuts to add products, manage expenses, view insights and find customers sit one tap away, with recent sales, the Pika Score and offers below.

Settings moved out of the bottom navigation into the profile picture, freeing the bar for daily work, and a short product tour shows returning retailers what moved: “Settings used to be in the menu below. It lives here now.”Settings moved out of the bottom navigation into the profile picture, freeing the bar for daily work, and a short product tour shows returning retailers what moved: “Settings used to be in the menu below. It lives here now.”

Recording a sale is the most important thing the app does: it turns a moment at the counter into a record. The earlier version was one long form. The redesign splits it into the steps a sale really has: find the product (search or barcode), build the cart, take payment, review, then share a receipt or invoice.Recording a sale is the most important thing the app does: it turns a moment at the counter into a record. The earlier version was one long form. The redesign splits it into the steps a sale really has: find the product (search or barcode), build the cart, take payment, review, then share a receipt or invoice.

Real counters are messy, so the flow is too. A product that isn’t in inventory can be added mid-sale. Prices and quantities can be edited at review. A sale can be fully paid, partly paid or unpaid, with what the customer owes recorded against them.Real counters are messy, so the flow is too. A product that isn’t in inventory can be added mid-sale. Prices and quantities can be edited at review. A sale can be fully paid, partly paid or unpaid, with what the customer owes recorded against them.

Retailers add products by photographing the front and back of the pack to auto-fill the details, set stock in pieces or packs, add an optional cost price, and set a low-stock alert that feeds notifications.Retailers add products by photographing the front and back of the pack to auto-fill the details, set stock in pieces or packs, add an optional cost price, and set a low-stock alert that feeds notifications.

Empty and no-result states guide instead of dead-ending: “Your shelves are empty. Add products to start tracking what’s in stock.”Empty and no-result states guide instead of dead-ending: “Your shelves are empty. Add products to start tracking what’s in stock.”

A dedicated notifications screen gathers what needs a retailer’s attention:A dedicated notifications screen gathers what needs a retailer’s attention:

• Stock: “3 products are low in stock. Restock now to avoid losing sales.”
• Money: a customer’s payment received, with the new wallet balance.
• Habit: a daily sales summary, and a gentle reminder on a day with no sales recorded.
• Growth: Pika Score changes and new rewards in PikaFlex.
• Product: new features, and an empty state, “Nothing new for now”, when there is nothing to show.

Many retailers already keep a paper logbook, and asking them to abandon it fails. PikaBook meets them there: photograph the page, and Pika reads the sales on it, matches each product to inventory, and asks the retailer to confirm anything uncertain before it is logged.Many retailers already keep a paper logbook, and asking them to abandon it fails. PikaBook meets them there: photograph the page, and Pika reads the sales on it, matches each product to inventory, and asks the retailer to confirm anything uncertain before it is logged.

The page stays the source of truth, so the hard part was everything that can go wrong, designed state by state: a blurry photo (“Hold your phone steady and try again”), a book with several pages, a page that may already be logged, a price Pika can’t read clearly (“Please confirm it matches the page”), a product it can’t find, and a page it can’t process at all.The page stays the source of truth, so the hard part was everything that can go wrong, designed state by state: a blurry photo (“Hold your phone steady and try again”), a book with several pages, a page that may already be logged, a price Pika can’t read clearly (“Please confirm it matches the page”), a product it can’t find, and a page it can’t process at all.

PikaNet takes the app beyond one shop by connecting retailers with wholesalers nearby. Wholesalers post deals on their stock. Retailers search by product, wholesaler or LGA, see how far each one is, and reserve stock before they make the trip.PikaNet takes the app beyond one shop by connecting retailers with wholesalers nearby. Wholesalers post deals on their stock. Retailers search by product, wholesaler or LGA, see how far each one is, and reserve stock before they make the trip.

A reservation holds the stock for a set time, shows the exact total to pay at pickup, and is shown at the store to collect. Its status runs from reserved to confirmed, awaiting the visit, then purchased, and a retailer can contact the wholesaler or cancel, releasing the stock. I explored five layouts for the reservation summary before choosing one. An earlier Business Network concept was folded into PikaNet.A reservation holds the stock for a set time, shows the exact total to pay at pickup, and is shown at the store to collect. Its status runs from reserved to confirmed, awaiting the visit, then purchased, and a retailer can contact the wholesaler or cancel, releasing the stock. I explored five layouts for the reservation summary before choosing one. An earlier Business Network concept was folded into PikaNet.

Every sale and restock raises a retailer’s Pika Score, and PikaFlex turns that score into real value: stock credit, working capital, shop and stock insurance, and cashback from partners. This is where recording sales pays back most directly.Every sale and restock raises a retailer’s Pika Score, and PikaFlex turns that score into real value: stock credit, working capital, shop and stock insurance, and cashback from partners. This is where recording sales pays back most directly.

Partners run their own applications, and the app says so plainly: Pika opens the partner’s page, and the retailer fills in the partner’s own form.Partners run their own applications, and the app says so plainly: Pika opens the partner’s page, and the retailer fills in the partner’s own form.

Simi is Pika’s in-app assistant. I specified its button states (resting, new message, unread count, thinking) so it can signal without getting in the way of the counter.Simi is Pika’s in-app assistant. I specified its button states (resting, new message, unread count, thinking) so it can signal without getting in the way of the counter.

Field agents are how many retailers first meet Pika. The Agent App, built on the same design system, lets an agent onboard traders, record sales and add products on a retailer’s behalf, and run store surveys for brands: chillers, planogram and pricing, stock and expiry, shelf placement and competitors.Field agents are how many retailers first meet Pika. The Agent App, built on the same design system, lets an agent onboard traders, record sales and add products on a retailer’s behalf, and run store surveys for brands: chillers, planogram and pricing, stock and expiry, shelf placement and competitors.

Design didn’t stop at release. I read PostHog behaviour and weekly sales cohorts (May to August 2026) to see where retailers spent their time. Early August was an agent-led onboarding drive in one location, so I read those weeks as their own group, not as a trend for every trader. Each signal became the next change:Design didn’t stop at release. I read PostHog behaviour and weekly sales cohorts (May to August 2026) to see where retailers spent their time. Early August was an agent-led onboarding drive in one location, so I read those weeks as their own group, not as a trend for every trader. Each signal became the next change:

• Record a Sale is where the work happens: the most-used screen after Home, and the longest high-traffic one, at over a minute a visit. Every second saved there is saved dozens of times a day.
• Retailers are moving from typing to choosing. In July they typed items in manually more often than they picked them; among the retailers onboarded in early August, picking from inventory had already overtaken typing, and search was used far more. So search is the next investment: recent items, typo tolerance, stock levels and barcode or image shortcuts.
• The success screen became a launch pad. Taps on “Start another sale” kept climbing, more than doubling during the August drive, so it becomes the primary action, with the last customer and payment remembered.
• Reviewing a sale is fast and submissions are climbing, so the risk is no longer time but failed submissions. The next step is logging every failure by reason, app version, device and network.
• Not everything went up. Completed product creation dropped among newly onboarded retailers after photo capture and scanning arrived, so the new creation flow is being audited step by step.
• Cohorts acquired in June were the healthiest: half of one June cohort recorded a sale in their first week, and more than one in five were still selling six weeks later.