Ask the way you'd ask a colleague.
No exports and no query language. Suggested questions get you started.
Ask SignalAsk your store a question. Get the working too.
Ask in plain English, in Signal or from your own AI tools. Every answer shows how it was measured and opens onto the records behind it.
In every planWhich acquisition sources produce the most repeat buyers?
Newsletter signups and the Meta dog-walking ads produce the most repeat buyers.
scanned 61,204 orders · 520ms · n=11,420
Sample dataOverview
What is Ask Signal?
Ask Signal is superintelligence for your store's data. You ask in plain English, and it reads your orders, sessions and segments to answer. Each answer names its population, timeframe, sample size, sources and limits.
See how each answer was measured.
The steps, the rule it ran and the sample size sit beside the answer.
Save the answer as a segment.
The people behind an answer become a segment you can act on.
In the app
A question, its answer and its working.
- Reading shopify.orders · Jan 2023 to Aug 12, 2026
- Finding each customer's first touch
- Counting second orders within 90 days
- Checking sample sizes · smallest group 1,116
ASK-2240
Newsletter signups and the Meta dog-walking ads produce the most repeat buyers. 41% of customers who first arrived through the newsletter ordered again within 90 days, against 19% for Google search ads.
| First touch | Customers | Ordered again | Share |
|---|---|---|---|
| Newsletter signup | 1,310 | 537 | 41% |
| Meta · dog-walking ads | 3,920 | 1,294 | 33% |
| Organic search | 2,204 | 507 | 23% |
| Meta · other ads | 2,870 | 603 | 21% |
| Google search ads | 1,116 | 212 | 19% |
customers WHERE first_order < 2026-08-12 AND first_touch IS known GROUP BY first_touch COUNT second_order within 90 days → 11,420 customers, 5 sources
scanned 61,204 orders · 520ms · n=11,420
Save as segmentPin findingSee these people
How it was measured
- Population
- 11,420 customers
- First order
- Before Aug 12, 2026
- Timeframe
- 2023-01-01..2026-08-12
- Compared
- 5 first-touch sources
- Smallest group
- 1,116
- Sources
- shopify.orders, storefront.sessions
- Cause
- Not established
- Reading shopify.orders · all time
- Ranking customers by lifetime spend · top 10%
- Comparing first touch and first two orders with all customers
- Checking sample sizes
ASK-2231
Your top 10% by lifetime spend are 1,348 customers. 62% first arrived from a Meta ad about dog walking, against 29% of all 13,480 customers, and 71% bought dog gear in their first two orders. Cause isn't established.
customers WHERE lifetime_spend >= p90 COMPARE first_touch, first_two_orders WITH all customers → 1,348 vs 13,480
scanned 61,204 orders · 340ms
Save as segmentPin findingSee these people
- Reading SEG-0044 · 6,318 members
- Splitting on orders ≥ 1 · 2,140 vs 4,178
- Comparing return gap, products viewed, first touch
- Checking sample sizes
ASK-2236
Buyers came back sooner and looked at the Trail Lead first. 71% of buyers returned within 3 days of their first visit, against 22% of the others. Cause isn't established.
| Behavior | Bought | Didn't |
|---|---|---|
| Returned within 3 days | 71% | 22% |
| Viewed the Trail Lead | 64% | 18% |
| First touch was a Meta ad | 58% | 61% |
| Sessions before ordering or leaving | 3.1 | 1.6 |
SEG-0044 SPLIT BY orders >= 1 COMPARE return_gap, viewed, first_touch, sessions → 2,140 vs 4,178
scanned 19,870 sessions · 410ms · n=6,318 · last 60 days
Save as segmentPin findingSee these people
Every pattern is labeled as a correlation to test.
Upside is shown as a scenario, with its assumptions.
Every answer shows how fresh its data is.
Illustrative data from a sample store.
The details
Your own tools can ask too.
01
A permission-controlled MCP connection
Connect the AI tools you already use. They ask Signal the way you do.
02
The same definitions everywhere
An answer in Signal and an answer over MCP use the same definitions and the same store-level access.
03
Correlation is labeled as correlation
A pattern is something to test. Signal says so on the answer, and says when cause isn't established.
Related
Goes with Ask Signal.
Segments
Describe the people you want in plain words. Signal builds the rule and keeps the segment current.
Customer profiles
Follow each shopper from first visit to order, with every session, ad and email on one timeline.
Autopilot
Test one change at a time, keep the winner, and see what Signal added.
Questions
Your questions, answered.
What can I ask?
Questions about your customers, your visitors and your store. For example, which acquisition sources produce the most repeat buyers, or what your highest-spending customers have in common.
How do I know an answer is right?
Each answer shows the steps Signal took, the rule it ran, the sample size and its sources, and opens onto the records behind it.
Can I ask from my own AI tools?
Yes. Signal has a permission-controlled MCP connection. Your tools use the same definitions and the same store-level access as the app.
How do I get started, and what does it cost?
Ask Signal and the MCP connection are in every plan. Starter is $47 a month plus 3¢ per unique visitor per month, with a 7-day free trial.
Ask the question you've been exporting for.
Every plan starts with a 7-day free trial. Want to talk first? Book a call.