Real prompts, real outputs, real data shapes. Copy one, or use it as a starting point for your own question. Everything here runs on Free — your data, not a demo.
A weekly report showing top movers and outliers — kept as a Monday-morning digest.
"Summarize last week's sales by category and flag outliers." Classic cohort retention, in one sentence. Works on any events table.
"Show retention by signup month for the last 12 cohorts." When conversions dip, find out which step broke first.
"Why did conversions drop this week? Show me which step of the funnel moved." Before you analyze, know what's broken. Nulls, duplicates, schema drift.
"Clean this CSV and tell me what's off — nulls, duplicates, and anything that looks wrong." Which acquisition cohort is leaking the most? The agent finds the pattern.
"Find which signup cohort or acquisition channel has the worst 90-day retention, and tell me what they have in common." Real net margin per SKU. Revenue minus COGS, ad spend, fees, shipping.
"Show net margin by SKU for the last 30 days. Include COGS, ad spend, platform fees, and shipping. Flag anything below 15%." Turn any question into a scheduled report. Delivered to Slack or email.
"Every Monday morning, email me last week's top 10 customers by revenue." Do the analysis once. Rerun it on the next client's data with one click.
"Save this cohort analysis as a reusable script — I want to run it on different client datasets." A pre-built month-end routine: GL checks, AR aging, revenue reconciliation.
"Run my month-end close checklist on this month's data." Catch fraud, glitches, and freakishly large orders before they cause a refund.
"Watch every new Shopify order. Flag anything that's 3× the average order size, multiple orders from the same IP in 10 minutes, or unusual shipping addresses." Forward purchases to Meta CAPI / GA4 MP / TikTok Events. Close the iOS 14+ attribution gap.
"Whenever a Shopify order lands, forward a deduplicated purchase event to Meta CAPI, GA4 Measurement Protocol, and TikTok Events with proper event matching." Auto-draft the metrics section of your weekly board update.
"Every Friday, draft my weekly investor update — MRR, runway, top wins, top concerns — using last week's data." A real admin panel, generated from a sentence. No React required.
"Build a CRUD admin for the users table with a search bar and role filter." A shareable live view for your team. Updates on its own.
"Make a dashboard for my team with MRR, churn, active users, and net new revenue." Let non-technical teammates explore the data themselves — without writing SQL.
"Let non-technical people on my team explore this table with guardrails." Take any saved analysis and give it a shareable public URL.
"Publish my weekly sales summary as a public dashboard at a shareable URL." A working reorder tool: forecast stock-out, suggest reorder quantity, log decisions.
"Build me an inventory app that forecasts when each SKU will run out, suggests reorder quantity based on the last 90 days, and logs my decisions." A daily summary of what your warehouse camera saw — without watching the footage.
"Take my warehouse camera feed and give me a daily digest of unusual events — bin movements at odd hours, doors left open, unattended packages." Ping me when refunds cross a threshold. Silent when they don't.
"Ping me on Slack if my refund rate goes above 3% on any SKU." For warehouses and kitchens running MQTT sensors. Pings you when temperature drifts.
"Ping me if any freezer sensor goes above -15°C for more than 5 minutes overnight." Catch budget runaways and zero-conversion campaigns before they burn $1K.
"Watch my Google Ads and Meta Ads. Alert me if any campaign's daily spend doubles unexpectedly, or if any active campaign has zero conversions for 48 hours despite > $100 spend." Get pinged the moment any SKU drops below a reorder threshold.
"Alert me when any of my top 50 SKUs drops below 14 days of inventory based on the last 30 days of sales velocity." Auto-detect when conversion drops meaningfully on any segment, and open an incident brief.
"Watch my conversion rate. If it drops more than 15% on any traffic source for an hour, fire off an incident brief explaining what changed." For small factories with PLC or MQTT-connected machines. Watches readings, pings on anomaly.
"Watch all my MQTT sensor streams. If any reading drifts more than 3 standard deviations from its 7-day baseline for 10 minutes, ping me and start logging at higher frequency." Missing the one you need? Send us your question — we'll publish good starter packs here when they fit.
Every starter pack above runs on Free. Drop in your own data and see how Tablize answers — before you decide whether to keep it around.