What is AVA?
AVA is an AI-powered data exploration assistant built into YUMBI Reports. Your team can ask questions in plain English and get structured results immediately in a clean table that’s ready to copy-paste or download as CSV for Excel or Google Sheets. That means faster reporting cycles, fewer dependency bottlenecks on analysts, and a consistent way for marketing, ops, and franchisor leadership to self-serve answers, especially when decisions need to be made quickly.
Just as importantly, AVA keeps the output practical and usable: it returns tables only, so you’re not stuck translating narrative “insights” back into numbers or hunting for the underlying data. And because AVA only works with your YUMBI data, it stays focused on the metrics that actually move your business (stores, customers, orders, items, vouchers, campaigns), helping teams move from “What’s happening?” to “What do we do next?” with far less friction.
You can find AVA Explorer in YUMBI Reports under ‘Overview’ in the left-hand navigation. (see Figure 1).
Figure 1
AVA also keeps your question history for you to look back on. Click on ‘Conversation History’ next to the left-hand navigation bar which will bring up your previous conversations.
Figure 2
What questions can AVA answer?
AVA can currently answer questions across the following areas:
- Store and brand sales performance - revenue, order counts, growth, order type breakdowns
- Customer behaviour and traits - registrations, ordering patterns, retention
- Campaign performance - YUMBI campaign enrolment and activity metrics
- Web and app analytics - installs, sessions, events, user acquisition
- Store configuration - store details and settings
Note: Store trading hours are not yet available but are coming soon.
How do I write a good prompt?
The best prompts include three ingredients:
| Ingredient | What it means | Examples |
| Metric | What are you measuring? | number of orders, total revenue, % growth, new users |
| Dimension | How do you want the data split? | by day, by store, by channel, by brand |
| Filter | What to include or exclude? | last 6 months, January 2026, app and web only |
Vague vs. Specific
Try and be as specific as possible in order to get the best answer.
| Too vague | Much better |
| "How is Pollo Main Street doing?" | "Revenue per day for Pollo Main Street, web and app orders only, January 2026" |
| "Show me Pollo data" | "Total orders per store for Pollo South Africa, December 2025, broken down by delivery vs collect" |
What keywords should I use?
These keywords matter - AVA treats them differently:
- Online orders = orders placed via Web, Android or iOS channel. Use "online orders" or "app and web orders".
- Paid online = payment made digitally, regardless of channel. You must say "paid online" or "online payment" to filter for this.
- Order type = use "delivery", "collect" or "kerbside" to filter by fulfilment method.
- Time of day = use "breakfast", "lunch" or "dinner" - or say "morning", "afternoon", "evening" and AVA will map them correctly.
How do I use follow-up questions?
AVA remembers your conversation. Each follow-up question builds on the ones before it - you don't need to repeat yourself.
Example - three short questions that give the same result as one long one:
- "Give me the revenue for Store A for Jan 2026"
- "Break it down per day"
- "Online orders only"
This gives the same answer as: "Revenue per day for Store A for January 2026, web and app orders only"
You can ask up to 5 follow-up questions per conversation. Click Ask a new question to start fresh.
Figure 3
Here are some example prompts to try
- "Total revenue by brand for the last 12 months in South Africa, web and app orders only"
- "Revenue per day for Pollo Main Street for January 2026"
- "Actual vs forecast revenue for Pollo stores in Gauteng for February 2026"
- "Online payments revenue for all Pollo stores for December 2025"
Growth
- "Year-on-year order count growth by store for Pollo, current 12 months vs prior 12 months"
- "Month-on-month revenue growth for Pollo South Africa for 2025"
Customer Behaviour
- "How many customers registered in the last 3 months but haven't placed an order yet?"
- "What proportion of customers ordered for collect vs delivery in the last 7 days?"
- "How many customers bought chicken items last Friday?"
Items and Products
- "How many Sushi items were sold last month, broken down by store?"
- "Which stores sold the most Pizzas in the last 30 days?"
App and Web Analytics
- "How many app installs did Pollo get per month for the last 6 months?"
- "Compare unique users per month vs number of orders for Pollo for the last 12 months"
Order Failures
- "Which stores had the most order failures via POS in the last 30 days?"
- "Show me order failure reasons for Pollo South Africa for the last 7 days"
Advanced
- "Break Pollo stores into revenue deciles for January 2026 and pick 5 random stores from each decile"
- "How many brand A customers have registered in the last 3 months but haven't placed an order yet?"
Figure 4
Tips and Gotchas
- Be specific about dates - "last month" works, but "January 2026" is more reliable.
- Name the store or brand - if you know it, include it.
- Quoted phrases are treated as product searches - e.g. "Chicken Picanto" or "Two for Tuesday" will search for menu items, not filter by day of week.
- Growth is year-on-year by default - current 12 months vs prior 12 months, unless you specify otherwise.
- Unique users per month means total users within that month's slice, summed from the monthly analytics table.
- Results are capped at 1000 rows.
- Give feedback - use the thumbs up/down after each answer. This helps AVA improve over time. (see figure 5 and 6)
Figure 5
Figure 6
If you’d like help with AVA Explorer please don’t hesitate to reach out to the Customer Success Team.