Ask why, not just how much.
The assistant reads your data — campaigns, orders, margins — and answers in plain language, in English and Italian, with the numbers and campaigns that back up the answer: every figure comes from your data, read at that moment.
The problem
The answer exists, but it lives in six different places.
To understand why ROAS dropped you need the Google export, the Meta export, the order spreadsheet and half an hour of cross-referencing. By the time you finish, the data is already stale.
The assistant runs those queries for you: it reads the metrics, compares periods, cross-references channels and answers in one sentence, attaching the numbers and campaigns it based the answer on. And before concluding it digs into the cause, rather than stopping at the first figure.
How it works
Seven tools, one conversation.
Metrics and comparisons
Queries spend, revenue, ROAS and CPA over any period and compares them without you opening a single export.
Cross-channel reports
All platforms in a single currency: spend, revenue and ROAS comparable channel by channel.
Recommendations in context
It reads the open recommendations from the rules engine — statistical thresholds, no alarms on noise — and relates them to your question: what to change, where, with what expected impact.
It knows what it can ask
Before answering it consults the data dictionary of your workspace: it knows which metrics, periods and sources really exist, and builds the answer on those.
Off-script questions
For what the other tools don't cover it runs additional queries on your data, always read-only: it answers, it doesn't modify.
Working memory
Remembers targets, seasonality and constraints you explained once, per user and per workspace.
Saved conversations
Your last thirty conversations stay available, with an automatic title, and the quick questions you use most often are one click away, in English and Italian.
Real questions
"Why did ROAS drop in September?"
It isolates the channels and campaigns that lost ground, compares them with August and tells you how much of the drop is new spend and how much is lost conversion.
"Which keywords are eating my budget?"
It lists spend on zero-conversion Google keywords over the period you choose, sorted by cost: the list ready for cleaning up the account.
"What is August's true ROAS, email costs included?"
It adds up ad spend and off-platform costs, compares them with real order revenue and tells you how far that is from the ROAS the platforms declare.
What it doesn't do
We also tell you where it stops.
It doesn't edit campaigns on its own
It analyzes, explains and proposes. The changes — budget, pause, target, new campaigns — you make yourself from Thor, when you decide.
It doesn't calculate per-product profit
Spend is not attributed by category: we talk about margin, MER and true ROAS.
It doesn't use real-time exchange rates
Rates are static and stated next to the numbers, with per-workspace overrides.
Put your first question to your data, not ours.
Connect your accounts via OAuth — on your own, or with us during the first call — and ask your first question the same day. In English and Italian.