Take the number apart until it makes sense.
Ask in plain English. Answers come with the chart and the metric behind them. Then keep pulling. Slice it another way, drop the campaign you suspect, ask why it disagrees with what the platform reported. Stop when the difference has a location. Pin it. You'll run this one again.
Opens from any dashboard panel or Signal · Every answer names the metric behind it · Pin what's worth keeping
Most questions don't start in a chat box.
They start at a chart that looks wrong, or a line in Monday's briefing saying paid social spend moved 18%. Click either one and chat opens with the question already assembled: same metric, same window, same filters that produced the thing you were looking at. You begin on the second question instead of retyping the first.
A chat box that opens empty assumes you already know how to phrase what you noticed. You didn't notice a phrase. You noticed a chart.
- From any dashboard panel. The panel's metric, date range and filters carry over.
- From any Signal. The alert opens with its own numbers already loaded.
- Nothing to re-select. You don't rebuild the view you were already looking at.
- Paid social spend moved 18%Open in chat
- Acquisition & Media · nCAC by channelOpen in chat
- Same metric
- Same window
- Same filters
Two numbers disagree. The interesting part is why.
Ad platforms count on their own terms: their attribution window, their view of what a conversion was. So their number and yours part ways, and the only useful question is where. Every answer names the metric behind it and the formula it used, so you can see exactly what's being counted. Then narrow it until it's small enough to see. One campaign, one week. A few questions in, the gap has a name.
- Every answer names its metric. The definition behind it is one click away.
- Chart, table or SQL. Switch views on any result; the SQL is there when someone wants to check it.
- CSV export on any result, including for teammates with view-only access.
Transcript pending: this slot takes real product output, not placeholder figures.
Figures throughout come from a sample store modelled on real brand patterns. The gaps they illustrate are real; the sizes will be different on yours.
Pin the ones you'll run again.
Some checks are one-offs. Others come back every month, and it's usually the same handful: the channel that always looks better than it should, the SKU nobody can explain. Pin an answer and the whole thread stays. The question, the number, the chart, the metric it came from. Next time it comes up you're reading last month's working instead of rebuilding it. A pinned answer holds the day it was asked. If you want it watched continuously, that's a dashboard.
- The question is kept with the answer. Chart and metric included.
- Hand it to someone else, including teammates with view-only access.
- What did discounts cost us last quarter?MER After Promo5.05×
- Show contribution margin against gross margin.Contribution Margin Pct31.9%
A wrong number is worse than no number.
Truefold answers from a defined set of metrics, each with a formula and a stated scope. Ask something those metrics can't support and it says so, along with the nearest question it can answer. It won't improvise. A plausible wrong number doesn't cost you anything today. It costs you in a meeting three weeks from now, in someone else's slide.
- It tells you what it can't do, and why.
- A refusal points somewhere. The closest question it can answer.
- No improvised figures. Every answer comes from a metric with a defined formula.
Break net profit down by channel.
I can't, and I'd rather say so than guess.
Overhead isn't allocated to a channel on purpose — rent isn't incurred by paid search, and splitting it by revenue share makes a channel look worse in the month it sold more. Contribution margin by channel is the closest honest answer.
Read-only on every source.
Truefold pulls; it never writes. Nothing in your store, your ad accounts or your flows changes because you connected us.
Isolated, and minimised on the way in.
One BigQuery dataset per organisation, used by nobody else. Customer names, email addresses, phone numbers and shipping addresses are dropped before anything is written.
The definitions are yours to adjust.
Rename metrics to your team's language, correct anything that doesn't match how you run the business. Version-controlled, with a name against every change. We don't train AI on your business data.
See it against your own numbers.
Connect Shopify, your ad accounts and Klaviyo. Then ask it something you already know the answer to.