Skip to content
Communication · Marketing · Artificial intelligence · Cultures2 November 2026
The ideas, skills and cultures
reshaping communication.
Magazine menu
Machine translation. The French edition is the reference. Original advertising archives may remain in French. Read in French
Home / Technologies
TechnologyAnalysis

AI-assisted data analytics: turn an answer into a verifiable marketing decision

Ask the assistant for the data scope, calculation and limitations of the explanation. A useful decision links an observed result to an action and a re-evaluation criterion.

Article prepared with artificial intelligence assistance. The sources and method are indicated in the text.

Visual of the source publication: Artificial intelligence-assisted analytics data: turning a response into a verifiable marketing decision
Visual published by blog.google · Source and context articleThis visual is that of the original publication; the model demonstrations may contain images generated by artificial intelligence.
In this folder
  1. Asking data without skipping context
  2. Define a question that the table can shed light on
  3. Control the calculation chain
  4. Separate recognition, hypothesis and evidence
  5. Write the decision and its control
  6. Sources and method
  7. Also to read
The decision to prepare

Ask the assistant for the data scope, calculation and limitations of the explanation. A useful decision links an observed result to an action and a re-evaluation criterion.

Asking data without skipping context

Google describes Advisor and Analytics Advisor as assistants who can help explore performance and pathways. This approach facilitates dialogue with a dataset. It does not automatically correct measured events, comparative periods or conversion definitions. The first question therefore remains the one of context: What data can the assistant read and what decision does the team seek to prepare?

Define a question that the table can shed light on

"Why do sales fall?" can go beyond what is contained in the advertising report. The team must distinguish between broadcasting, user response and commercial processing. A more precise question is about an observable difference: evolution of the rate of passage between two stages, decrease in volume on a medium or change in the cost of a defined action. This precision allows the calculations to be checked before proposing an explanation.

Control the calculation chain

Ask for fields, filters, period and reconciliation rules. Check duplicates, missing values and units. Aggregates must be able to be reconciled with sources. A percentage deserves a check of its denominator. Artificial intelligence can help prepare requests and synthesis; the responsibility of the balance sheet requires to return to the tables. An apparent variation may come from a change in measurement rather than a change in public behaviour.

Separate recognition, hypothesis and evidence

A finding describes what appears in the scope of the study. An hypothesis suggests an explanation to be examined. Evidence requires a mechanism adapted to the question. A campaign, promotion or product availability may evolve at the same time. The report must make these possibilities visible. The team is getting the missing information and the test that would allow for arbitration, rather than presenting a single cause with an assurance that the data do not allow.

Write the decision and its control

A recommendation must indicate the proposed action, its purpose, the person responsible and how to review it. For example, a pilot may seek to reduce form friction before extending media pressure. This scenario is a proposal for a method, not a guaranteed result. For several countries or languages, keep the differences in measurement and context. The assistant becomes useful when it makes reasoning more readable and the verification work easier.

Sources and method

The recommendations and grids of this article are proposed by Le Fils de Pub. The constructed examples are indicated; the advertisements and results attributed to an organization remain those of this source. Article prepared with artificial intelligence assistance.

Explore further

Continue with a practical method

Suggested routes from the story to briefing, production or measurement. These features are separate from the original publication.

  1. From brief to CRM: manage an AI-assisted digital campaign around business goals
  2. CRM and loyalty: use AI to identify a need, then organise the response

Log In

Forgot password?

Don't have an account? Register

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

To use social login you have to agree with the storage and handling of your data by this website.

Add to Collection

No Collections

Here you'll find all collections you've created before.

Directeur de la publication : Alexandre Frioui