Every AI marketing use case — personalization, insights, audience building, measurement — runs on data. The data has to be connected, trusted and lawfully usable.
Connected and trusted
- One view of the customer: shared IDs across CRM, web, app, commerce and service, so AI isn't guessing which records belong together.
- Clean taxonomy: consistent naming for campaigns, products, channels and content, so results can be compared.
- Known quality: field completeness and freshness for the fields your priority use cases need.
Lawfully usable
Consent and opt-out rules differ by market (for example GDPR in Europe, state privacy laws in the US). The principles are similar: tell people what you do, honour their choices, and don't use data for purposes they wouldn't expect.
Enforcement is real. In August 2022 California's Attorney General announced a $1.2 million settlement with Sephora, alleging it failed to disclose that it sold consumers' personal information and failed to honour opt-out requests sent through the Global Privacy Control browser signal.
Cookies: don't wait for a deadline
Google spent years planning to phase out third-party cookies in Chrome, then changed course. In April 2025 it said it would keep the current approach — users choose in Chrome's settings — and would not roll out a new standalone cookie prompt. The lesson for marketers: build on first-party data and consent you control, rather than on any platform's timetable.
What AI adds
- AI tools may copy data into new systems (vendor clouds, model providers). Check where data goes and whether it's used for training.
- Generated audiences and lookalikes must respect the same consent as the source data.
- Sensitive inferences (health, finances, children) need extra care, even if you never collected that data directly.