Data activation is the step that turns stored customer data into something an advertising system can act on. A retailer may hold millions of loyalty records, purchase histories and email addresses in a customer relationship management (CRM) system or a cloud data warehouse. None of that changes which ads run until the records, or segments built from them, reach a platform that buys or sells media. Activation is that delivery: uploading an audience to Google, Meta or a demand-side platform (DSP), sending conversion events back to a bidding algorithm, or attaching a seller's audience to a deal.

The term exists because collecting data and using it are separate jobs, often run by separate teams on separate systems. PPC Land's explainer on first-party data places activation at the end of a chain that begins with collection and consent and passes through unification and identity matching. It is where most of the friction sits. Experian's 2025 Media Buying in Transition report, cited when the company partnered with London-based AUDIENCES on July 30, 2026, found that 72% of marketers struggled to turn first-party data into media activation.

How a record becomes an audience

Activation generally runs in five stages. The first is selection. A marketer defines a segment from source data, such as customers who bought in the past 90 days, subscribers who lapsed, or high-value households. The query may run in a customer data platform (CDP), a CRM or directly in a warehouse such as Snowflake, Databricks or BigQuery.

The second is preparation. Identifiers are normalised so identical values look identical: email addresses trimmed and lowercased, phone numbers converted to the E.164 international format. They are then hashed with SHA-256, which produces a 64-character string. A formatting error is rarely rejected outright. The record simply fails to match later, which makes preparation one of the quieter sources of loss.

The third is matching. The destination compares the hashed values with its own logged-in users, or an intermediary translates them into an identifier the destination recognises. The share of records recognised is the match rate. Figures vary widely. Google says the reachable share of a list can range from 1% to 99%. An ISBA and PwC study published on May 6, 2020 matched 31 million of 267 million impressions end to end, about 11.6%.

The fourth is delivery of the matched audience into a usable object: a Customer Match list in Google Ads, a Custom Audience at Meta or OpenAI, a segment in a DSP, or a deal on a supply-side platform (SSP). The fifth is refresh. Audiences decay as customers churn, so most pipelines re-sync on a schedule. According to Google's Customer Match help pages, memberships expire after 540 days, and a list stays eligible only if at least 100 members are added or updated within that window.

Platforms set floors on audience size. Google standardised a minimum of 100 active users across its networks after cutting the Search threshold from 1,000. OpenAI requires 25,000 matched users for targeting in ChatGPT ads. The arithmetic matters: a 50,000-person file at a 40% match yields 20,000 people, too few to target there.

Who operates it

On the buy side, activation is run by advertisers' CRM and data teams, agency data units and the DSPs themselves. The Trade Desk's Galileo, announced on January 5, 2023, was built to ingest first-party data through CRM, CDP and clean room partners including Adobe, LiveRamp, Salesforce and Snowflake. StackAdapt launched a Snowflake Native App on July 24, 2025 so segments could be selected inside the warehouse before syncing to the DSP.

On the sell side, publishers, retailers and SSPs activate their own audiences for buyers. Walmart Connect made its shopper segments available in Yahoo DSP on May 28, 2026, with Magnite moving the data and Walmart retaining control over its use. Magnite's self-serve ClearLine Activate added 27 named data partners on September 9, 2026, letting buyers attach audiences without negotiating a new deal ID for each one.

Between them sit intermediaries: onboarders such as LiveRamp, Experian and TransUnion, CDP vendors such as Adobe, Salesforce, Tealium and Treasure Data, and warehouse-native tools such as Hightouch and Census.

From onboarding to the warehouse

The practice predates the label. LiveRamp built its business on what it called onboarding, linking offline customer files to online identifiers. Acxiom announced on May 14, 2014 that it would buy LiveRamp for about $310 million in cash, according to its filing with the US Securities and Exchange Commission (SEC). The filing said LiveRamp connected to nearly 100 marketing technology companies. Acxiom sold its marketing solutions business to Interpublic Group for $2.3 billion in 2018, and the remaining company took the LiveRamp name.

Walled gardens built their own doors at the same time. Facebook launched Custom Audiences in September 2012, letting businesses upload hashed emails, phone numbers or user IDs, according to TechCrunch. Google followed with Customer Match on September 28, 2015, initially for email addresses only.

The CDP category arrived in between. The analyst David Raab is generally credited with naming it in April 2013. A CDP copies data from many systems into its own store, builds one profile per person and pushes segments to destinations through prebuilt connectors.

The next shift questioned the copying. Census, founded in 2018 in San Francisco, began publishing warehouse data back into business applications, a pattern the market labelled reverse ETL (extract, transform, load). Census co-founder Boris Jabes called the name "a misnomer because ETL has no direction," according to TechTarget. Hightouch, also founded in 2018, extended the idea into a composable CDP, which activates data directly from Snowflake or Databricks without copying it into a proprietary profile store. Gartner named Hightouch a Leader in its CDP Magic Quadrant for the first time in January 2026.

Google's pipes were consolidated in parallel. The Data Manager API, launched on December 9, 2025, replaced separate routes for Customer Match lists, conversions and Display & Video 360 audiences with one endpoint. Customer Match uploads through the older Google Ads API services stopped working on April 1, 2026.

Why it matters to marketers

Third-party cookies did not disappear from Chrome. Google abandoned its plan for a standalone cookie prompt on April 22, 2025. Yet Safari and Firefox block them by default, and Apple has required opt-in for its mobile advertising identifier since 2021. Data a company collects itself, under its own consent, has become the most dependable input for targeting and measurement. It is only worth something if it can be activated.

Activation also feeds automated bidding. Google's Smart Bidding uses Customer Match lists automatically, and suppression lists keep existing customers out of acquisition campaigns. A list that fails to sync can therefore distort spend in campaigns nobody thinks of as audience-based.

Loss at every hop is the central weakness. Each translation between identifier systems drops records, and vendors seldom publish rates. LiveRamp's ChatGPT integration, for instance, launched without a disclosed match rate. Vendor claims, such as Epsilon's 5% to 15% improvement on Pinterest, are self-reported.

Consent is the second constraint. An IAB Tech Lab perspective published in May 2026 argued that first-party data is usable only when collected through valid consent and that the consent layer should mature before activation tooling. Norway's data protection authority, Datatilsynet, made the point in a decision dated June 1, 2026. It fined two Elkjop companies NOK 20 million, about EUR 1.85 million, partly because Elkjop planned to reuse loyalty club data for Customer Match without assessing whether the new purpose was compatible with the original one.

Hashing is not anonymisation. The US Federal Trade Commission (FTC) said so in July 2024. In November 2024 its technical staff warned that clean rooms "are not rooms, do not clean data," and that unlawful disclosure stays unlawful regardless of whether a clean room is involved.

Concentration is a quieter dispute. Onboarding has long been described as neutral plumbing. Publicis agreed on May 17, 2026 to buy LiveRamp for $2.5 billion, placing a widely used activation layer inside a holding company that competes with other agencies relying on it.

Not the same as

Data onboarding. Onboarding is the translation of offline records into online identifiers. It is one route to activation, not the whole of it. Data already keyed to a platform's users, such as hashed emails sent straight to Customer Match, is activated without an onboarder.

Identity resolution. Identity resolution decides which records belong to the same person and assigns them a shared identifier. Activation consumes that output; it does not create it.

Audience marketplace. A marketplace is where segments, usually from third-party data providers, are listed and licensed. Activation is the act of delivering a segment, from any source, to where media is bought.

Data enrichment. Enrichment appends attributes to existing records. It changes what a profile contains, not where it can be used.

Recent developments

Destinations keep multiplying. LiveRamp connected brands' customer lists to ChatGPT ads in 11 markets on September 28, 2026, weeks after OpenAI dropped the 25,000-user floor for exclusion audiences. Databricks launched CustomerLake, which it calls an agentic CDP, on June 16, 2026, keeping profiles inside its own platform.

Inputs are widening too. Google began accepting IP addresses through the Data Manager API on May 28, 2026, though not for Display & Video 360, and extended the option to Google Ads file uploads in September 2026, excluding users in the European Economic Area, the UK and Switzerland. Each addition raises match rates on paper. Whether the matches are accurate, and lawful, remains the harder question.

Timeline

  • September 2012: Facebook launches Custom Audiences, accepting hashed emails, phone numbers and user IDs
  • April 2013: David Raab names the customer data platform category
  • May 14, 2014: Acxiom announces it will acquire LiveRamp for about $310 million
  • September 28, 2015: Google launches Customer Match for email addresses
  • 2018: Census and Hightouch are founded; Acxiom sells its marketing solutions unit to IPG and the remaining company becomes LiveRamp
  • January 5, 2023: The Trade Desk announces Galileo for first-party data
  • July 24, 2024: FTC states that hashing does not make data anonymous
  • November 13, 2024: FTC staff warn on data clean room privacy claims
  • April 22, 2025: Google drops its plan for a standalone third-party cookie prompt in Chrome
  • July 24, 2025: StackAdapt launches a Snowflake Native App
  • December 9, 2025: Google launches the Data Manager API
  • January 2026: Gartner names Hightouch a Leader in its CDP Magic Quadrant
  • April 1, 2026: Customer Match uploads through older Google Ads API services stop working
  • May 17, 2026: Publicis agrees to acquire LiveRamp for $2.5 billion
  • May 28, 2026: Google adds IP addresses to Data Manager API Customer Match; Walmart Connect audiences reach Yahoo DSP
  • June 1, 2026: Datatilsynet fines Elkjop NOK 20 million
  • June 16, 2026: Databricks launches CustomerLake
  • July 30, 2026: Experian and AUDIENCES announce partnership
  • September 9, 2026: Magnite adds 27 data partners to ClearLine Activate
  • September 28, 2026: LiveRamp connects customer lists to ChatGPT ads in 11 markets

Summary

Who: Advertisers, agencies and their data teams on the buy side; publishers, retailers and SSPs on the sell side; and intermediaries including onboarders such as LiveRamp, Experian and TransUnion, CDP vendors such as Adobe and Salesforce, and warehouse-native tools such as Hightouch and Census.

What: The process of delivering customer data, or segments built from it, into platforms where media is bought or sold, so it can be used for targeting, suppression, bidding and measurement.

When: Platform-level activation began with Facebook Custom Audiences in 2012 and Google Customer Match in 2015, alongside the onboarding business LiveRamp built before Acxiom acquired it in 2014. Warehouse-native activation spread from 2018, and Google consolidated its intake through the Data Manager API from December 2025.

Where: Between a company's data stores (CRM, CDP or cloud warehouse) and destinations including Google, Meta, OpenAI, DSPs and SSPs, often through an identity intermediary or clean room.

Why: First-party data has no effect on advertising until a buying or selling system can recognise it. Activation bridges that gap, at the cost of match-rate loss at each hop and legal exposure where consent does not cover the new use.