The question usually arrives on a Tuesday: Which events actually carried margin last quarter? How do returns compare by region? When I slice revenue by rep, does it still tie to what I see at the event level?
Getting there used to mean pulling multiple reports, stitching spreadsheets, and hoping nobody fat-fingered a join. Data Insights is our answer to that grind: you type the question, and ZenShows proposes a read-only answer against a curated analytics layer—sales, events, returns, shipments, rep-day performance, forecasts vs. actuals (where you have forecasts), and the payroll or P&L pieces you’ve chosen to include.
Built for questions, not screenshots
The data underneath is organized so common roadshow questions map cleanly:
- Sales with pricing reconciled so comparisons are apples-to-apples.
- Event-level summaries so you can rank or filter the way you think about the business.
- Returns and shipments when you need operational follow-up, not just top-line revenue.
- Rep-day performance with pre-attributed revenue so you don’t double-count when you go per-rep versus per-event.
- Calculator items, datacollector fields, and P&L inputs you’ve opted into for analytics—so answers can reflect the definitions your team already trusts.
Monetary amounts are normalized for consistent comparisons. If the answer is larger than a quick on-screen preview, you can download the full result as CSV and share or archive it.
Every run is scoped to your vendor context. The assistant can’t run destructive operations or wander outside the approved analytics tables. Results stay private to your user account. If your team asks the same questions often, you can save reusable prompts so nobody starts from zero.
Why we’re confident in the foundation
Before we leaned into conversational analysis, we measured two things: how complete the analytics layer is on our production platform, and how early adopters have actually used saved answers.
| What we measured | Order of magnitude |
|---|---|
| Event-level summaries in analytics | 33,900+ |
| Sales line rows (item × day) | 1,270,000+ |
| Rep performance day rows | 472,000+ |
| Return line rows | 494,000+ |
| Forecast summaries (latest run per event) | 9,300+ |
| Saved query results (all time) | 620+ |
| Data rows returned in those results | 17,600+ |
That tells us the lake is deep enough for real questions—not a demo—and that vendors are already running repeatable analysis through the same path Data Insights uses.
What you’ll notice
Speed on exploratory work. You get a short explanation of what ran, plus a table. You didn’t have to spec a formal report first.
Fewer spreadsheet mistakes. Grain matters—revenue by rep isn’t the same roll-up as revenue by event. The system is guided toward the right shape so you’re less likely to mis-compare.
Your opt-ins matter. The metrics you expose through calculators, P&L, and datacollector configuration can show up in answers when you mark them for analytics—so plain-English questions return numbers that match how you run the business.
Exports when audits or shares need the full picture. Big answers don’t trap you in a preview window.
How to try it
- Access. Data Insights is available on vendor accounts when your role includes permission to view AI Reports / Data Insights. If you don’t see it under Labs, ask your ZenShows administrator to enable it for your role.
- Open it. From Labs, choose Data Insights (or open AI Reports from your ZenShows site).
- Ask precisely. Include the time range, metric, and how you want it sliced—e.g. “top 10 events by retail revenue last month” or “returns by product group this year.”
- Review and export. Read the description, scan the table, and grab CSV if you need every row.
- Tune inputs (optional). Use AI Reports configuration to opt calculator and datacollector fields into the analytics set so answers line up with the metrics you already rely on.
One more thing: bring your own AI assistant
We’re also testing something we’re excited about. The same analytics layer that powers Data Insights is now accessible through the Model Context Protocol (MCP)—an open standard supported by AI assistants like Claude, ChatGPT, and Cursor. In early beta, you can point your own AI assistant at ZenShows using your existing API key, and it can answer business questions against your data without you ever opening a browser. Same read-only safeguards, same vendor scoping. It’s early, and we’re learning alongside the teams trying it—but if you already live inside an AI assistant for other work, this is worth a look.
We’re continuing to expand what belongs in the lake based on how teams use it—if you have a question shape that’s painful today, that’s exactly what we want to hear about.