WolfSellers — Adobe Experience Cloud Partner en México

Article

Adobe Real-Time CDP Use Cases: 10 Audience Examples

Ten Adobe Real-Time CDP audiences to build: cart abandonment, suppression, high-value, win-back and B2B accounts, each with rule, evaluation and destination.

By WolfSellers··16 min read
Adobe Real-Time CDP Use Cases: 10 Audience Examples
On this page

An audience in Adobe Real-Time CDP is a group of profiles that match a rule, such as "bought in the last 30 days." A destination is the connection that sends that audience to another platform: Google, Meta, Adobe Target, your email tool or your data warehouse. If you already know what Real-Time CDP is and want to know which audiences to build first and where to send them, this is the playbook: ten audiences, each with its rule, evaluation method, destination, KPI and the mistake we see most often.

At WolfSellers we implement Adobe Real-Time CDP on Adobe Experience Platform (AEP) for companies across North America and Mexico. If you're still evaluating the platform, start with what Adobe Real-Time CDP is and how it unifies your data. Everything we say about Adobe features comes from Adobe's official Experience League documentation, reviewed in November 2026.


How data flows through Real-Time CDP: from source to destination

Sources bring data into AEP and destinations send it out. The profile and the audience sit in between:

  1. Source. Your store, CRM, app, loyalty platform or point of sale sends data in streaming or in batches, modeled in XDM.
  2. Profile. AEP stitches together the identifiers of one person. A merge policy decides which value wins when two sources disagree.
  3. Audience. You define the rule in Audience Portal, with Audience Builder or Audience Composition.
  4. Evaluation. AEP works out who qualifies: in batch, in streaming or on the edge.
  5. Activation. You connect a destination, pick the audiences, map identities and attributes, and schedule the export.

If you worked with older versions: Adobe no longer says "segments." The interface calls them audiences, and the old Segment Builder experience is being retired in early 2027.

One detail shapes many designs: you can only export profile attributes to a destination, not event fields. If you want to send the last product viewed or six-month spend, turn it into a computed attribute first.

The 10 Real-Time CDP audiences at a glance

# Audience Goal Sources Evaluation Destinations Primary KPI
1 Cart abandoners Win back the sale with ads Web SDK, store Streaming Meta, Google Customer Match Recovered carts
2 Recent buyers (suppression) Stop paying to acquire customers Store, ERP, POS orders Batch + streaming Google Customer Match, Meta Spend avoided
3 High-value customers Retain and seed lookalikes Orders, CRM, loyalty Batch Google Customer Match, Meta Match rate
4 Buyer look-alikes Prospect within your profiles Full profile Batch (daily model) Email, AJO, ads Conversion vs. the rest of the base
5 Purchase or churn propensity Prioritize budget Events, scores Batch Ads, email, CRM Lift vs. control
6 Lapsed customers (win-back) Reactivate customers Orders, email Batch Meta, Google Customer Match % who buy again
7 Category interest this session Personalize the home page Web SDK Edge Adobe Target, Custom Personalization Conversion vs. default
8 Loyalty tier Tier benefits everywhere Loyalty platform Batch, edge or streaming Target, Adobe Commerce, Meta, email Retention by tier
9 List for CRM or data warehouse Get it to sales or BI Full profile Batch Amazon S3, SFTP, Azure Blob Contacts reached on time
10 B2B accounts showing intent Account-based marketing CRM, Marketo Engage, web Batch (accounts) (Companies) LinkedIn, Demandbase Opportunities created

The time windows and thresholds in each example (30, 60, 120 days) are our starting points, not Adobe rules: tune them to your buying cycle.

Data engineer reviews on his laptop when and how CDP audiences get evaluated.

Batch, streaming or edge: how is an audience evaluated?

The evaluation method decides how fast a profile enters and leaves an audience. AEP assigns it based on the rule: if an audience qualifies for streaming or edge, you can pick that method in Audience Builder.

Method When it updates Rules it supports Limit per sandbox Use it for
Batch Every 24 hours automatically; on demand with a segment job or flexible audience evaluation Any rule 4,000 active audiences 30-, 90- or 180-day histories; attribute-only rules
Streaming As data arrives; up to 5 minutes to qualify a profile One or more events within 24 hours, with or without profile attributes 500, including edge; 1,500 events per second across all sandboxes Cart, "bought today," status changes
Edge On the Edge Network, up to 350 ms per event One event within 24 hours, profile-only rules (evaluated daily) or audiences of audiences 150; 1,500 events per second Same-page and next-page personalization

The rules that trip teams up most:

  • Streaming doesn't accept profile-only rules, a single event with a window longer than 24 hours, or "ignore year" time constraints. If you edit an audience so it no longer qualifies, AEP switches it to batch on its own.
  • Since May 2025, a rule that mixes batch and streaming data is evaluated in batch. To combine them, build a batch profile audience and a streaming event audience, then join them in an audience of audiences.
  • Streaming and edge require an Active-On-Edge merge policy. Streaming also requires scheduled segmentation to be turned on for your organization in Audience Portal.
  • Flexible audience evaluation runs batch audiences on demand: up to twice a day and 50 times a year per production sandbox, with at most 20 audiences per run.

Real-Time CDP destination types: when to use each one

The destinations catalog is grouped by what you want to achieve. These are the types we use most, with examples listed in the Experience League catalog:

Type What it receives Catalog examples Frequency Use it for
Advertising and social The audience, keyed on hashed email or phone Google Customer Match, Google Display & Video 360, Facebook (Meta), LinkedIn, TikTok, The Trade Desk, Amazon Ads Streaming Retargeting, suppression, seeds
Email and engagement Profiles with the attributes you map Adobe Campaign, Salesforce Marketing Cloud (API or files), Oracle Eloqua, Braze API or files, depending on the connector Campaign lists
Personalization Audiences and, optionally, attributes Adobe Target, Custom Personalization, Adobe Commerce With Target and Custom Personalization: same page, next page or next session Home page, banners, promotions
Cloud storage CSV files with profiles and attributes Amazon S3, Azure Blob, Google Cloud Storage, SFTP, Data Landing Zone Daily full file or incremental every 3, 6, 8 or 12 hours Data warehouse, CRM, contact center
Enterprise streaming Profiles in near real time HTTP API, Amazon Kinesis, Azure Event Hubs Streaming Internal systems; Real-Time CDP Ultimate only

The activation limits Adobe documents:

  1. 100 destinations per sandbox, a system limit. Up to 10 of them can be Custom Personalization.
  2. 250 audiences per destination as a recommendation, and no more than 50 activated at once. Adobe suggests batches of 20.
  3. 50 mapped attributes per destination, or 30 for a personalization destination.
  4. 5 million rows per file. AEP splits larger exports into numbered files.
  5. Removing a profile from an audience removes it from the destination on the same cadence as evaluation, though each platform takes its own time to process it.

Google and Meta don't accept personal data in clear text. If emails or phone numbers sit in an unhashed namespace, AEP applies SHA256 on activation; when mapping from attributes, check Apply transformation. Phone numbers go in E.164 format. And some connectors are maintained by the partner, such as TikTok, so a fix may depend on that vendor.

Shopper on her living room sofa with her phone, about to abandon an unpaid cart.

10 Real-Time CDP audience and destination examples

1. Cart abandoners on paid media

What it's for: showing products again to people who added them to the cart and didn't check out, especially if they don't open your emails.

Sources: website events (Web SDK) or app events (Mobile SDK), such as commerce.productListAdds and commerce.purchases, plus the identity of a signed-in customer.

Rule: added to cart in the last 24 hours and then didn't buy.

Evaluation: streaming. Adobe supports multi-event rules in a strict order within 24 hours, even with a negated event, as long as the whole rule isn't a negation. The profile qualifies within minutes and leaves as soon as they buy.

Destinations: Facebook (Meta), which covers Facebook, Instagram, Audience Network and Messenger, and Google Customer Match. Email and push are orchestrated in Adobe Journey Optimizer (AJO), as we show in 10 journey examples.

KPI: recovered carts and match rate on each platform.

Common mistake: expecting anonymous visitors to get in. Google and Meta match on hashed email or phone: no known identity, no match. And if you stretch the window to 3 days, the audience moves to batch.

2. Suppressing recent buyers from ads

What it's for: no longer paying acquisition ads to reach existing customers. It's one of the paid media activation patterns Adobe documents.

Sources: orders from your online store, ERP or point of sale, tied to the profile by email or customer ID.

Rule: two audiences. "Bought in the last 30 days" runs in batch, because the window is longer than 24 hours. "Bought today" runs in streaming and covers the gap between one daily evaluation and the next.

Evaluation: batch every 24 hours, plus streaming.

Destinations: Google Customer Match and Facebook (Meta). The exclusion itself is set in the campaign inside the ad platform: AEP only delivers the list.

KPI: spend avoided. Adobe suggests estimating it as estimated CPM times the size of the suppressed audience.

Common mistake: using only the batch audience. Someone who buys in the morning keeps seeing "buy now" ads until the next evaluation.

3. High-value customers for Customer Match and lookalikes

What it's for: reaching your best customers on Google (Search, Shopping and Gmail) and using the list as a seed to find similar people, a pattern Adobe describes for Google, Meta and The Trade Desk.

Sources: orders, CRM and loyalty platform.

Rule: six-month spend above your threshold, built with a computed attribute (SUM function).

Evaluation: batch.

Destinations: Google Customer Match and Facebook (Meta). For YouTube inventory, Adobe points to the Google Customer Match + DV360 destination.

KPI: match rate and return on the campaigns that use the seed.

Common mistake: asking a computed attribute for more history than it has. It looks back up to 6 months and, at that length, refreshes monthly; with fast refresh, daily. For 12- or 24-month value, compute the attribute with Adobe Data Distiller.

4. Buyer look-alikes inside AEP

What it's for: finding, among the profiles you already have (subscribers, registered accounts that haven't bought), the ones most similar to your best customers.

Sources: AEP's model uses attributes, events and audience memberships from the last 30 days.

Rule: a base audience of "bought twice or more in 90 days." You create the look-alike audience by picking a similarity percentage.

Evaluation: the model retrains every 24 hours. The base audience must be built in AEP, use the default merge policy and avoid fields restricted by data governance.

Destinations: your email tool, AJO or paid media.

KPI: look-alike conversion compared with the rest of your base.

Common mistake: confusing it with Google or Meta lookalikes: AEP only searches your own profiles. You also can't build a look-alike from another look-alike, and your allowance depends on your package: 5 active in production with Prime and 20 with Ultimate.

5. Purchase or churn propensity

What it's for: focusing budget and incentives on the people most likely to buy or to leave.

Sources: consumer experience events and, if your contract includes it, Customer AI, which writes conversion or churn propensity scores to the profile. A score from your own model, loaded as an attribute, also works.

Rule: high churn score and at least one purchase in the last 12 months.

Evaluation: batch, after each model run.

Destinations: paid media for retention, AJO for messages and a file to the CRM so sales can call.

KPI: lift against a control group that doesn't get the action.

Common mistake: asking Customer AI for things Adobe says it doesn't do: it doesn't recommend products, doesn't set prices and can't tell whether a discount will make someone buy.

6. Lapsed customer win-back

What it's for: reactivating people who stopped buying without relying on email alone.

Sources: orders and email engagement.

Rule: bought at least once and hasn't bought in 120 days.

Evaluation: batch.

Destinations: Facebook (Meta) and Google Customer Match for people who no longer open emails. The email journey that goes with this audience is in our AJO examples.

KPI: share who buy again within 30 days.

Common mistake: one threshold for the whole catalog. Someone who buys a refrigerator doesn't have the same cycle as someone who buys pet food: build one audience per category.

7. Home page personalization on the edge

What it's for: changing the main banner based on what the visitor viewed this session or on their profile.

Sources: Web SDK events and profile attributes, with a datastream that has Adobe Target and AEP configured as services.

Rule: viewed the "electronics" category today, meaning one event within 24 hours.

Evaluation: edge. Adobe documents up to 350 ms per event and 150 edge audiences per sandbox. A profile-only rule also runs on the edge, but it's evaluated once a day.

Destinations: Adobe Target, for the same page or the next one, or Custom Personalization if another tool makes the decision. If your store runs Adobe Commerce (formerly Magento) 2.4.4 or later and you have Real-Time CDP Prime or Ultimate, the Adobe Commerce destination uses the audience in banners and promotions.

KPI: conversion against the default experience, measured with an A/B test. How to run one is in Adobe Target and A/B testing.

Common mistake: leaving the audience on a different merge policy. Edge destinations only evaluate audiences on Active-On-Edge. Without a datastream (with at.js, for example), Target can only personalize the next session.

8. Customers by loyalty tier

What it's for: showing each tier its benefits on the site, in email and in ads.

Sources: tier, points balance and point expiration from your loyalty platform.

Rule: one audience per tier ("tier = Gold") and another for the event "moved up a tier today."

Evaluation: the tier audience is profile-only, so it runs in batch, or on the edge with daily evaluation. The tier change is one event within 24 hours, so it can run in streaming.

Destinations: Adobe Target and Adobe Commerce for the site, Facebook (Meta) for ads and your email tool.

KPI: retention and redemption by tier.

Common mistake: expecting an upgrade to land within minutes with a profile-only rule: streaming doesn't accept it. We cover program design in digital loyalty programs.

9. Working list for your CRM or data warehouse

What it's for: getting the audience into the systems used by sales, the contact center or BI. For example, high-value customers with no purchase in 60 days, so an account rep can call them.

Sources: the full profile.

Rule: whatever the receiving team needs, with the attributes they'll use: name, phone, assigned rep.

Evaluation: batch, with the export scheduled to run right after the daily evaluation.

Destinations: Amazon S3, Azure Blob, Google Cloud Storage or SFTP, with one full file a day or incremental files every 3, 6, 8 or 12 hours. Real-Time CDP Ultimate adds near-real-time destinations: HTTP API, Amazon Kinesis and Azure Event Hubs.

KPI: share of contacts reached within the agreed time.

Common mistake: scheduling the first export a minute after activation. Activation can take up to an hour to propagate, and for a new audience Adobe says to start the next day.

10. B2B accounts showing intent

What it's for: focusing ads and sales effort on accounts where several people show interest (account-based marketing).

Sources: CRM (accounts, contacts and opportunities), Adobe Marketo Engage and web events.

Rule: accounts with at least one person in the audience "visited the pricing page in the last 30 days" and no open opportunity. Audience Builder lets you use people audiences inside an account audience.

Evaluation: batch every 24 hours, with up to 50 account audiences per sandbox. It's only available in the B2B and B2P editions of Real-Time CDP.

Destinations: (Companies) LinkedIn, Demandbase, Bombora or cloud storage files. For the people in those accounts, the Marketo Engage destination creates static lists.

KPI: opportunities created in the activated accounts.

Common mistake: looking for it in the B2C edition, which has no account audiences. How to prioritize those people is in B2B lead scoring with Marketo Engage and AEP.

Real-Time CDP checks data governance at the moment you activate. Four pieces come into play:

  1. Data usage labels on fields and datasets, such as C2 (no export to third parties) or I1 (data that directly identifies a person). Audiences inherit the labels of the fields they use, and computed attributes inherit the labels of their sources.
  2. Policies that block a marketing action on data carrying certain labels. All of them, including Adobe's core policies, are off by default: you have to enable them.
  3. The destination's marketing action, which you choose when you connect it. If it conflicts with an enabled policy, AEP flags the violation and won't let you activate.
  4. Consent policies, which drop profiles that haven't consented to that action. They're only available with Adobe Privacy & Security Shield or Adobe Healthcare Shield. Without them, Adobe's Google destination documentation says to filter out non-consented profiles in the audience rule.

Two situations we see often:

  • The Web SDK only enforces the data collection permission automatically. Other permissions (email, ads, personalization) are stored on the profile, but you have to honor them in your rules or with policies.
  • Since June 2026, Meta blocks audiences built on data its terms restrict, such as health or financial information: those audiences stop receiving data from AEP. If you're a bank, insurer or pharma company, review the rule before activating.

We cover labels, identity and consent in depth in data governance in Adobe Experience Platform. For Mexican customers, what the LFPDPPP requires on privacy notices and consent is in data protection for ecommerce in Mexico.

Manager and analyst review a printed checklist before launching their first audiences.

A checklist for launching your first 3 audiences

At WolfSellers we usually start with recent-buyer suppression, cart abandoners and high-value customers: they run on data almost every store already has, and their impact shows up within weeks.

  1. One owner and one KPI per audience, with the control group defined before activation.
  2. Identity in shape: lowercase emails, phone numbers in E.164 and each identifier in the right namespace, in clear text or hashed.
  3. Merge policy reviewed, plus Active-On-Edge if you'll use streaming or edge.
  4. Scheduled segmentation turned on in Audience Portal; streaming won't run without it.
  5. Eligibility checked: windows of 24 hours or less for streaming; everything else in batch.
  6. Labels and policies enabled, with the right marketing action on each destination.
  7. Consent filtered in the rule or through consent policies.
  8. Ad accounts ready. On Meta, Adobe's business added as a partner and the Custom Audiences terms signed. On Google, an account that complies with the Customer Match policy.
  9. File exports scheduled at least an hour after activation.
  10. Monitoring: match rate, delivery rate and time to activation (the KPIs Adobe suggests), plus your business metric.

Real-Time CDP audiences and destinations: frequently asked questions

What's the difference between a source and a destination?

A source brings data into Adobe Experience Platform: your store, CRM or cloud storage files. A destination sends data out to another platform: Google, Meta, Adobe Target or your data warehouse. The same system can be both: your CRM feeds the profile and gets a list back.

How often does an audience update?

It depends on the method: batch every 24 hours, streaming within 5 minutes and edge within 350 ms. Then the destination matters: streaming destinations send each change as it happens, and file-based ones on your schedule, from every 3 hours to once a day. Each platform also takes its own time to process it.

Do I need Adobe Journey Optimizer to activate audiences?

No. Real-Time CDP activates audiences to destinations on its own. AJO comes in when you want to orchestrate messages per person, with waits and conditions, across email, SMS, push or WhatsApp. Both share the AEP profile, and audiences created in AJO can also be activated to destinations.

What happens to my CRM?

Real-Time CDP doesn't replace it: your CRM stays the system for sales and service, and it's usually a source for the profile. In the other direction, audiences reach it as scheduled files or, with Real-Time CDP Ultimate, through the HTTP API.

How many destinations and audiences can I have?

Per sandbox: 100 destinations and 4,000 active audiences, up to 500 of them streaming and edge (150 edge) and, in B2B, 50 account audiences. Per destination, Adobe recommends up to 250 audiences. The limits that apply to you are the ones in your sales order.

Can I activate audiences that weren't built with rules in AEP?

Yes. Audience Portal also takes audiences uploaded as CSV, composed with Audience Composition, federated from your data warehouse with Federated Audience Composition, or created in Customer Journey Analytics, Data Distiller or the look-alike model. The destinations we reviewed for this article, such as Google Customer Match, Facebook and Adobe Commerce, accept them.

If this topic is relevant to your business, these services from WolfSellers can help you implement it:

Want to dive deeper?

Let's talk.

We're an Adobe Gold Partner in Mexico with experience in Adobe implementations and integrations. If anything in this article applies to your operation, the first consultation is on us.

Or email us at contacto@wolfsellers.com

Keep reading

Chat with us on WhatsApp