Diagnosing conversions

Where are you losing potential customers, and which improvements should you prioritise based on verified data?

1960 € per session, excl. VAT

In short

This seven-hour day, excluding breaks, provides a structure for diagnosing an acquisition journey, from first contact to conversion. You use generative AI to organise observations, test hypotheses against data and produce a prioritised optimisation plan, with indicators, attribution limitations and verification rules.

Why this training?

A satisfactory volume of visits can mask poorly qualified enquiries or drop-off after the form. Without shared definitions of the stages, teams compare incompatible indicators and focus their efforts on the wrong point in the journey. Generative AI helps to structure a journey map and formulate possible explanations. It proves neither causality nor attribution: every recommendation must remain linked to a verified observation and a way of measuring its effect. Continuing case study: Nacre Mobilier, a fictitious organisation selling professional furniture, attracts visitors through organic search and paid campaigns. Between browsing the catalogue, requesting a quotation and sales qualification, it does not know where losses are concentrated. Fictitious exports and sales feedback make it possible to examine these drop-offs. Workplace application: Apply the framework to a real journey using authorised data, have the definitions validated by the marketing and sales teams, then monitor a priority improvement over a period defined before its launch.

Objectives of this training

  • Mapping the stages and touchpoints through to qualification
  • Checking the consistency of events, periods and populations
  • Calculating intermediate conversions and losses by segment
  • Distinguishing observations, explanatory hypotheses and attribution limitations
  • Prioritising improvements with indicators and monitoring arrangements

Who is it for and what are the prerequisites?

Roles and tasks concerned

  • Acquisition and digital marketing managers overseeing multiple channels
  • Marketing analysts and CRM managers working with sales teams

Prior skills

  • Using a generative AI assistant and structuring a contextualised request
  • Knowing the acquisition channels and indicators for your activity
  • Being able to filter a table and calculate a ratio

What you will work on

  1. Defining the relevant journey

    Objective: Link the stages to an acquisition decision. Target conversion, micro-conversions, touchpoints and handover to sales.

  2. Making the available records reliable

    Objective: Rule out misleading comparisons. Duplicates, missing events, users or sessions as units of measurement, time windows and confidentiality.

  3. Measuring transitions

    Objective: Locate quantitative losses. Denominators, intermediate conversions, cohorts and breakdowns by channel or device.

  4. Explaining without overinterpreting

    Objective: Develop substantiated hypotheses. Intentions, form friction, lead quality, selection bias and multichannel attribution.

  5. Weighing up improvements

    Objective: Choose actions proportionate to the evidence. Prioritisation by the significance of the drop-off, the strength of the evidence, effort and dependencies.

  6. Organising measurement

    Objective: Make the diagnosis actionable and open to revision. Baseline, primary indicator, lead quality, seasonality and limitations of before-and-after monitoring.

What you take away

  • Journey map and event dictionary
  • Checked conversion table and anomaly log
  • Source-backed diagnosis of drop-offs and hypotheses
  • Prioritised optimisation plan with a measurement protocol

Method and assessment

  • Short instructional inputs followed by immediate practice using the continuing case study.
  • Individual work followed by work in pairs, with peer review of the outputs.
  • Guided group discussion: what works, what needs correcting and what remains a human decision.
  • Preparation: Describe an acquisition journey and its target conversion, without personal data. Check which tools are authorised by the organisation; the exercises use fictitious or anonymised data.

Observable criteria: Each stage has an explicit definition, source and population; Key ratios are recalculated outside the generated response; The selected drop-offs are substantiated by traceable observations; Hypotheses and attribution limitations are flagged; Each action has an owner, an indicator and a review deadline.

Frequently asked questions

Does an advertising platform need to be connected?

No, the fictitious exports are sufficient to practise the calculations and diagnosis.

Does AI calculate all the indicators?

It can help formulate the calculations; key results are checked in the spreadsheet.

Can a long journey be analysed?

Yes, by defining a cohort and a window compatible with the time to conversion.

Does a drop in conversion prove there is friction?

No: targeting, measurement or seasonality may explain it. The diagnosis documents these alternatives.

Last updated: