%202.png)
How Ipsos South Africa turned WhatsApp photos of menu boards into a live price tracker across Kenya, Ghana and Zambia for a global quick-service restaurant brand.
How Ipsos South Africa turned WhatsApp photos of menu boards into a live price tracker across Kenya, Ghana and Zambia for a global quick-service restaurant brand.
In short
Ipsos South Africa tracks menu prices every month for a global quick-service restaurant brand in Kenya, Ghana and Zambia. Photos of menu boards are sent on WhatsApp, Yazi's AI reads every item and price, and Ipsos analysts check only what the system flags. The end client renewed the tracker for 2026, with a live dashboard in place of monthly Excel pivots.
Ipsos South Africa tracks menu prices for a global quick-service restaurant brand in three African markets. Each month, people in the field visit the brand's restaurants and its competitors, photograph the menu boards and send the photos in. The prices on those boards become the tracker: what each item costs, what changed, and which promotions are running.
Ipsos runs the programme with a fieldwork partner on the ground, and brought in Yazi to collect the photos on WhatsApp and turn them into data. In 2025 the photos came from a consumer panel. For 2026 the tracker moved to dedicated field auditors and a live dashboard.
Every item on every menu board had to be read and captured. The first country alone took hours.
The end client received the tracker as Excel pivot tables, once a month.
Blurry, partial or wrong photos had to be caught one submission at a time.
Menus differ by country, with combos and specials that don't map neatly to a product list.
Field auditors send menu-board photos to Yazi on WhatsApp, up to ten photos per restaurant visit in one session. Yazi's AI reads each photo and turns it into line items, capturing combos and time-limited specials as items of their own. Ipsos analysts review anything the system can't match in an admin view, and the results flow into a live dashboard for the end client.
How the monthly cycle changed from 2025 to 2026
Each visit is a short WhatsApp session, checked as it arrives.
Illustrative screen showing the field auditor flow. Not real menu data.
The difference showed in the first months. Ipsos's account lead at the time captured the first country's menu photos by hand, which took two to three hours. For the next two countries, with Yazi reading the photos, he set up the full report and analysis in less time than that first capture took.
Accuracy held up under Ipsos's own checks. After QC on a month of Zambia data, the Ipsos analyst reported a low error rate. In 2026 the fieldwork partner tested deliberately poor photos and found the system still read them.
The QC view: analysts only review items the system flags, such as new items and specials. Illustrative layout, no client data shown.
For 2026 the end client renewed the tracker for a second year, with a live dashboard and WhatsApp quality checks added. Ipsos's QC team has since signed off the August and September 2026 data.
Two years of the tracker
“Comparing it to what we were doing before, this is far better.”
“Images I thought weren't zoomed in or close enough, the whole thing picked up perfectly.”
| Tracker in 2025 | Tracker in 2026 | |
|---|---|---|
| Photo collection | A consumer panel on WhatsApp | Field auditors, up to ten photos per visit |
| Reading prices | Read by AI, processed by analysts each month | Read by AI as the photos arrive |
| Specials and combos | Handled case by case | Captured as their own line items |
| Quality checks | Reviewed at month end | Analysts review only flagged items |
| Client reporting | Monthly Excel pivots | A live dashboard |
A tracker that takes less analyst time each month, catches problems at the point of upload, and was renewed for a second year with a live dashboard added.
Menu prices for its own restaurants and its competitors in three markets, in a live dashboard instead of a monthly spreadsheet.
It will make it a lot easier for everyone. We don't have to do manual dashboard updates every morning, send feedback, check for the photos. It will just make it so much smoother.
The fieldwork in a price tracker was never the hard part. Reading thousands of menu photos was. With collection on WhatsApp and the reading done by Yazi's AI, Ipsos turned a monthly manual job into a live tracker, and kept its analysts for the judgement calls.
Field auditors send up to ten photos per restaurant visit on WhatsApp. Yazi's AI reads each photo and turns it into line items with prices, capturing combos and specials as items of their own. Ipsos analysts review only the items the system flags, such as new items.
Kenya, Ghana and Zambia, every month, for a global quick-service restaurant brand and its competitors.
For 2026 the photos moved from a consumer panel to dedicated field auditors, WhatsApp quality checks were added, and a live dashboard replaced the monthly Excel pivots the end client used to receive.
Turning photos into data by hand?
Book a 30-minute demo →