Copilot lets you ask questions about your Unleashed data in plain English. Rather than navigating to a specific dashboard and applying filters manually, you can describe what you want to know, and Copilot will find the right data, build the query, and return results as a table. You can then generate a chart, refine your question with follow-up prompts, or open the results in the Insights editor for further adjustment and export.
How Copilot works
Step 1: Give Copilot a prompt
From Unleashed or Access Evo, click on the white "access" button displayed in the top left corner of the page.
From the menu, select Copilot.
Select New Topic. If you haven't used Copilot before, skip to the next step.
Select an Evo Skil: Analytics
Select one of the auto-generated prompts or type in a question in the free-text field available.
Step 2: Search data views
Copilot scans all available data views and selects the one most relevant to your question. If multiple views could answer it, Copilot may ask which one to use. You can also ask, "Is there a data view that records [what I need]?" to discover which views are available.
Step 3: Build and execute a query
Copilot translates your question into a structured query with field selection, filters, grouping, and sorting. It shows you the data source and fields it has selected before presenting results, so you can see exactly what it has done.
Step 4: Return results
Results appear as a table with up to 10 rows shown. Two options appear below the table:
Show a chart to visualise the data as a bar or line chart
View this data in Analytics to open the full result set as an editable insight where you can add columns, apply filters, and export.
Important: Copilot queries one data view at a time. It cannot join or combine data across multiple views in a single question. If your question requires data from two views, ask in two steps, or redirect Copilot to a view that contains both dimensions. If Copilot picks the wrong view, follow up with: "Is there a data view that has [what I need]?"
Working with time-based questions
Some data views contain lifetime or pre-aggregated totals and cannot be filtered to a specific year or quarter. For time-bound analysis, redirect Copilot to one of these periodic views.
Unleashed_Customer_Monthly_Performance — monthly revenue, profit, and margin per customer. Use for queries like "customer revenue in 2025 only."
Unleashed_Product_Monthly_Performance — monthly revenue, cost, and profit per product. Use for queries like "product performance this quarter."
Unleashed_Sales_KPI — invoice-line level data with dates. Use for any date-filtered transaction query.
If Copilot uses a pre-aggregated view for a date-specific query, follow up with: "Is there a data view with periodic revenue per month?" Copilot will discover the monthly views and you can restate your question against the right source.
Tips for effective prompts
Be specific about what you want to see. "Show me the best products sorted by revenue" works better than "show me products."
Name filter values explicitly. "Products Needing Attention" matches the segment label exactly, helping Copilot filter accurately.
Use column names when you know them. "Where recency is under 30 days" is clearer than "recently sold products" because it maps directly to a field in the data.
Ask Copilot to discover views. "Is there a data view that records [what I need]?" prompts Copilot to search all available views and suggest the best match for your question.
For date-specific analysis, use monthly or periodic views. Pre-aggregated views contain totals only and cannot be filtered by date range; redirect to a monthly performance view for time-bound queries.
Simple lookups
"Show me all the best products sorted by revenue." A clear subject, filter, and sort instruction gives Copilot everything it needs to return a focused result.
Multi-condition filters
"Products where recency is under 30 days but frequency score is low." Copilot applies both conditions simultaneously, returning only the records that match all criteria.
Aggregation and grouping
"What is the average margin for each product segment?" Copilot groups by segment and returns the average. You can follow up with "and per product group?" to pivot the same data by a different dimension without starting a new conversation.
Follow-up refinement
Copilot maintains context within a conversation. After an initial query, you can add filters, change grouping, or switch dimensions without repeating the full question. For example: "and now only in 2025", "sort by profit instead", or "use a different view."
Prompts to get you started
Use case | Prompt | Data view |
Sales performance | Monthly revenue trend for the last 12 months | Unleashed_Sales_KPI |
Discount abuse | Invoice lines with a discount above 30% by the salesperson | Unleashed_Sales_KPI |
Margin issues | Negative margin invoice lines last six months | Unleashed_Sales_KPI |
Fulfilment speed | Average order-to-ship days by warehouse | Unleashed_Sales_KPI |
Assembly quality | Assemblies with wastage above 10% last quarter | Unleashed_Assembly_KPI_Timeseries |
Supplier performance | Average order-to-receipt time by supplier | Unleashed_Purchase_KPI_Timeseries |
Procurement spend | Monthly purchasing spend over the last 12 months | Unleashed_Purchase_KPI_Timeseries |
📌 Note: Copilot querying requires data views to be enabled for your Access Evo organisation. If a query returns no results, check with your administrator that the relevant data view has been turned on.