03 · Use Cases & Verticals

Use cases & verticals.

From audience overlap to measurement, activation, enrichment, planning, suppression, and vertical-specific data collaboration.

The same architecture serves different business problems. Client type, vertical, and the specific decision being improved decide the right collaboration shape.

Scope this playbook → Run the 5-minute assessment →
A pegboard of buyers against use cases; a cord through the data-strategy row, the one buyer that owns activate, measure and decide ActivateMeasureDecide Media buyerData strategyExec Media buyer → Activate: reach + spend Data strategy → Activate: reach + spend Data strategy → Measure: proof + lift Data strategy → Decide: mix + plan Exec → Decide: mix + plan
Client type

Client type changes the collaboration path.

Data strength sets the starting play. The fuller the client's first-party data, the more the path shifts from insight toward governed match, measurement, and activation.

Client type changes the collaboration path Seven client types plotted by first-party data strength, each mapped to its primary collaboration path: brand with rich CRM, brand with limited first-party data, retailer or commerce, publisher or media, agency or holding company, platform / DSP / CDP, and data / identity partner. CLIENT TYPE DATA STRENGTH PRIMARY PATH Brand · rich CRM high Clean room match → measurement Brand · limited 1PD low Insight-first / private product Retailer / commerce very high Multi-party clean room Publisher / media high Advertiser-safe planning Agency / holdco medium Reusable workflow templates Platform / DSP / CDP high Governed activation loop Data / identity partner high Enrichment → model scoring
Representative — full detail in the cards below.
  1. High

    Large brand with rich CRM

    Data
    CRM, hashed email, purchase history, loyalty, web/app events, outcomes.
    Use case
    Audience match, suppression, measurement, LTV modeling.
    Pattern
    Clean room match → measurement → activation.
    Watch-out
    Do not skip governance and output policy just because the data is rich.
  2. Low to medium

    Brand with limited 1PD

    Data
    Campaign exposure, broad segments, site visits, agency-managed data, survey / panel data.
    Use case
    Planning insight, audience indexing, reach / frequency, partner enrichment.
    Pattern
    Insight-first clean room or private data product.
    Watch-out
    Do not over-promise activation if the brand lacks addressable 1PD.
  3. Very high

    Retailer / commerce platform

    Data
    Transactions, loyalty, SKU, basket, store visits, online / offline purchase, category data.
    Use case
    Closed-loop measurement, supplier collaboration, retail media optimization.
    Pattern
    Multi-party clean room with tiered access and strict output controls.
    Watch-out
    Supplier, agency, and media-network incentives must be defined upfront.
  4. Medium to high

    Publisher / media owner

    Data
    Logged-in users, content exposure, ad exposure, engagement, subscriptions, app / device behavior.
    Use case
    Audience overlap, reach / frequency, campaign proof, content affinity.
    Pattern
    Advertiser-safe planning → measurement → activation workflow.
    Watch-out
    Avoid making the clean room feel like a black box for advertisers.
  5. Varies by client

    Agency / holding company

    Data
    Client audience files, media spend, campaign logs, planning data, measurement results.
    Use case
    Reusable templates for planning, measurement, and partner collaboration.
    Pattern
    Workflow-led clean room templates across multiple clients and partners.
    Watch-out
    Client permissions, cross-client governance, and data separation must be explicit.
  6. High but governed

    Platform / DSP / CDP

    Data
    Device IDs, hashed email, campaign logs, segment data, conversion events, identity graph.
    Use case
    Activation match, suppression, frequency control, measurement feedback.
    Pattern
    Governed activation and feedback loop with strict destination policy.
    Watch-out
    Output rights and destination controls define what is commercially possible.
  7. High in one signal domain

    Data provider / identity partner

    Data
    Identity graph, attributes, household graph, intent, location, model scores.
    Use case
    Enrichment, identity resolution, lookalike modeling, attribution support.
    Pattern
    Enrichment → match quality → model scoring → DCR or native app output.
    Watch-out
    Define where the model output lives — raw data, DCR, app, or API.
Use cases

Five use cases, one orchestration layer.

  1. Audience matching

    Data strategy, MarTech, agency data team, privacy

    What overlap exists between my first-party audience and a partner's audience or inventory?

    • match rate
    • audience overlap
    • audience index
    • segment enrichment
    • planning insight

  2. Reach & frequency

    Marketing science, media analytics, agency measurement

    How often did my audience see the campaign across channels, platforms, or partners?

    • deduped reach
    • frequency curve
    • exposed / unexposed
    • partner contribution
    • incremental reach

  3. Measurement and incrementality

    Marketing science, analytics, finance

    Did exposure drive incremental outcome, and can we repeat this method?

    • lift
    • incremental reach
    • matched market design
    • holdout / control
    • matched exposure table

  4. Activation and suppression

    Activation, growth, performance, lifecycle

    Which users should I reach, suppress, retain, or value differently?

    • activation audience
    • suppression audience
    • frequency control
    • lookalike
    • feedback to optimization

  5. Planning and insight

    Brand planning, agency planning, finance

    Where does my customer overlap with partner audiences, content, or commerce signals?

    • indexed planning view
    • partner overlap
    • media mix input
    • commerce signal
    • audience read

Each use case has to land somewhere — a CDP, DSP, retail media network, publisher / SSP, BI / MMM system, or the semantic layer. Those execution surfaces are detailed in the Ecosystem Surfaces cluster →.

Marketing analytics ladder

From business question to decision.

Most enterprise data collaboration does not start with a clean room. It starts with a business question. The marketing analytics ladder translates data into decisions across three rungs — understand, predict, and activate.

  1. Understand

    Explain what happened and where value is coming from.

    Use cases
    • Customer journey analytics
    • Trendspotting
    • Self-service analytics
    • Segmentation
    • Reach / frequency
    • Market contribution
    • Online-to-offline measurement

    Outputs: Dashboards · cohorts · audience maps · path analysis · overlap · contribution views

  2. Predict

    Estimate what is likely to happen next.

    Use cases
    • LTV prediction
    • Purchase propensity
    • Churn prediction
    • Lead / sales prediction
    • High-value customer discovery
    • Budget response curves

    Outputs: Scores · model features · cohorts · ranked audiences · model diagnostics · planning inputs

  3. Personalize

    Change what the customer sees or receives.

    Use cases
    • Predictive segmentation
    • Suppression
    • Next-best audience
    • Product recommendation
    • CRM triggers
    • Site / app personalization
    • Media activation

    Outputs: Segments · suppression lists · activation audiences · offer rules · recommendation logic · experiments

Meta examples — where Meta media is material
  • Understand: reach / frequency distribution · campaign overlap · path-to-purchase · audience overlap · creative and placement distribution
  • Predict: conversion propensity · value-based event quality · MMM contribution · budget response curves with Robyn
  • Optimize: Advantage+ learning signal · retargeting windows · suppression logic · custom audiences where eligible · creative sequencing

Open the Meta Advanced Analytics deep dive

Where it fits Google / GMP / ADH Amazon Marketing Cloud Snowflake Databricks Productization
Multi-party collaboration

Four collaboration patterns.

  1. 1 + 1 collaboration

    Brand + publisher, brand + retailer, brand + partner. Most common pattern.

  2. Hub-and-spoke

    One enterprise (retailer, publisher, platform) running multiple bilateral collaborations.

  3. Federated network

    Multi-party measurement or planning environment with shared rules.

  4. Closed-loop measurement

    Brand + retailer + publisher + measurement partner — all parties contribute, all parties learn.

Verticals

Same architecture, different business problem.

  1. Pharma

    HCP / patient data is highly regulated. Federated, neutral environments win over single-platform paths.

  2. CPG / FMCG

    Retailer collaboration is the prize. Retail media networks become the primary clean-room counterparty.

  3. Automotive

    Long sales cycle, dealer / OEM data split. Identity resolution + LTV modeling matter more than activation.

  4. Retail

    Retailer is the host, brand + agency are guests. Output policy and supplier governance define the deal.

  5. Entertainment

    Exposure + content + subscription data are the assets. Reach / frequency and content affinity dominate.

  6. Financial Services

    High consent + governance bar. Insight-first clean rooms with strict output controls.

  7. Travel

    Booking, loyalty, partner network. LTV + cross-sell + lookalike for thin-data partners.

  8. QSR

    Loyalty + store visits + retail media. Closed-loop measurement against media + foot traffic.

Example

Business-user analytics workflow.

The semantic + agent-ready layers turn business users into safe self-serve analysts — when the metadata, metric logic, and benchmark answers are in place.

business-user-analytics-workflow
data in
Sales, exposure, conversion, product, customer, campaign, and outcome data.
governed workflow
Catalog → metric definitions → dashboard → conversational analytics → benchmark questions → monitored feedback loop.
decision improved
Which business users can safely answer planning, measurement, optimization, and forecasting questions without creating shadow analytics?
watch-out
If metadata, metric logic, and benchmark answers are weak, the AI layer will sound confident but produce unreliable answers.
Next step

Use case clear? Package it as a sellable product.

Once the client type, use case, and vertical are defined, the next decision is whether you're selling a data product, model, native app, workflow, or enterprise capability.

Scope this playbook → Next: Productization → Back to overview