Skip to content

How Aliveo Understands Your Data

To answer "Why did CPA go up on our Brand campaigns?", Aliveo needs more than the numbers. It needs to know what CPA means for you, which columns hold cost and conversions, and which campaigns count as Brand. It keeps this knowledge in a data model for each connected account, and every chat and agent reads it before touching your data.


The Data Model for Each Account

Each account's data model records:

  • Your KPIs: the metrics you judge performance by. A KPI can be a single metric, such as conversions, or a ratio of two, such as cost ÷ conversions for CPA.
  • Your funnels: the path from upstream to downstream metrics, such as impressions → clicks → installs → purchases, with the cost metric that feeds it. Aliveo suggests funnels from the metrics in your account, and you refine them.
  • How each metric adds up: whether a metric is summed (spend, clicks) or averaged, and how ratio metrics such as CTR are built from their parts. This keeps totals across campaigns or weeks correct instead of averaging averages.
  • The account's currency, so spend and budgets are reported in the right units.

You can review an account's data model at any time. Go to My Account → Business and click Review Data Graph on a connected account. The steps walk you through your KPIs, your funnels, the account's context and its campaign tags. Custom datasets and Google Analytics have no campaign tag step.


Column Types and Descriptions

For data you bring yourself, such as a CSV upload or a Google Sheet, every column carries three pieces of information:

  • A type: Date, ID, Metric or Category. Aliveo suggests one for each column, and you can change it.
  • An aggregation for metrics: Sum or Mean.
  • A description in your own words.

The AI reads these descriptions alongside the data. If a column is called gross_take and you describe it as "net revenue after refunds", the AI uses it when you ask about revenue. Metric columns are also the ones the AI compares by default when it breaks results down by campaign or segment.

For ad platforms, Aliveo already knows what each column means, so there is nothing to set up.


Metric Hints: Sending Each Question to the Right Account

Aliveo keeps a short summary of which metrics each connected account and dataset tracks. These are its metric hints. When you ask a question without naming an account, such as "How is ROAS trending this month?", Aliveo uses them to pick the accounts and the dataset that actually track ROAS. For a custom dataset, the hints include its metric, ID and date columns.


Campaign Tags

Most teams build meaning into campaign names, as in US_Brand_tCPA_Search. When you connect an account, Aliveo reads your campaign names, works out the naming convention, and proposes tag categories such as country, brand vs. non-brand, or bid strategy. You review and adjust them in the Define Campaign Tags step of Review Data Graph.

  • New campaigns are tagged automatically as they appear, using the same categories.
  • Tag categories are shared across the accounts in an organization, so "Brand" means the same thing on every account.
  • Tags that every campaign in an account carries are kept but given less weight. They can't tell one campaign from another, but they often tell one account from another, such as two countries.

Campaign tags work on every supported ad platform. They let you ask "Compare Brand vs. Non-brand CPA by country" without listing campaigns one by one. When your question names specific campaigns, Aliveo also looks up the campaigns whose names best match it.


How It Comes Together

When you ask a question, the AI receives the data model for every account in scope: KPIs and funnels, metric formulas, column descriptions and campaign tags. It also receives your saved business context (see How the Context System Works). So it works from a picture of your business, not from raw column names. To see how it then runs the analysis, read The Analysis Agent.