Why does Omni's figure differ from the one on a screen?

Omni reads the same engine as Analytics. Check the period and the attribution model in the answer's Source line against the screen's before you compare.

Can Omni remember what I asked last week?

Each conversation keeps its own questions and answers. Open it again from the history to carry on, or start a new one for a new subject.

What does a question cost?

It runs on your own key for the model you chose, and each answer shows what it cost. Short questions that name what to look at cost the least.

Questions that work

Questions Omni answers well, grouped by what you want to find out, and how to word one so the answer is the one you need.

Omni answering How healthy is my tracking?: a Tracking Health card with Pixel signal, Order sync, the Meta, Google and TikTok spend syncs and Attribution match, each with a green check, and the Source line under it

Omni answers best when a question names what to look at, over which period, and how to judge it. The questions below are written the way it reads them, grouped by what you are trying to find out. Use them as they are, or put your own names and periods in.

How to word a question

Name the period. "Last 7 days", "yesterday", "this month", "this quarter" or two dates. Without one, Omni chooses, and says which in the answer.
Name things as your Library does. A hook, an angle or an avatar by its name, a campaign by the name Meta shows. Omni finds the item before it measures it.
Say how to judge. Revenue, orders, ROAS, POAS or CPA, and "against last month" when you want a comparison.
Name the attribution model when it matters, such as "under Linear". Otherwise Omni uses your store's.
Narrow to one platform with "on Meta", "on Google" or "on TikTok".
Keep going in the same conversation. Omni reads the conversation so far, so a follow-up such as "and by angle?" can lean on the question before it.

Omni answers about the store chosen in the store switcher, from the data Omnymous holds. It does not read competitors' ads (that is Reconnaissance), and it changes nothing: its buttons open the screen, and you decide there.

How the store is doing

How did revenue trend over the last 30 days?
What was my net profit last month, and what ate into it?
How do this month's orders and AOV compare with last month?
Show me ad spend against revenue, day by day, for the last 14 days.

What earned the money

Which hook, angle and avatar actually earned the money last month?
Which creatives are winning right now?
Which campaigns on Meta had the best POAS in the last 30 days?
Which landing pages convert the visits our ads send them?
Which hook works best with which avatar?

What the Lab has called

Which variable is winning, and how sure are we?
Tell me everything about the angle "The 3am Economy".
What tests am I running, and who is winning?
Which experiments have stalled, and why?

Two questions a new conversation draws from the store's data, each marked From your data: Which experiments have stalled, and why? and Which variable is winning, and how sure are we?

A new conversation offers questions like these marked From your data: Omni has found something in your store worth asking about.

Your buyers

Where do buyers drop between my ads and the order?
Which paths do buyers take from an ad to the order?
How much is a customer from TikTok worth after 90 days, against one from Meta?
How do customers who first bought in July come back?
Show me the journey of order 22648, and its credit under each model.

Your production and your tracking

How many creatives did we make this month, how many launched, and how many won?
How healthy is my tracking?
What has the Insights feed found this week?
What is aMER, and how is it different from MER?

What to do next

What should we test next?
Which avatar should we double down on, and with which angles?

These end in buttons under Suggested Next Steps, such as one that opens Studio seeded with the avatar the answer found. See How Omni answers.