Archetypearchetype.aep-heavy-enterprise-evaluating-composable

AEP-Heavy Enterprise Evaluating Composable

A large enterprise with AEP as the incumbent CDP, a growing data warehouse footprint, and escalating friction from AEP's export and query limits. Evaluating whether to extend, supplement, or replace AEP with composable alternatives.

confidence 79%v2reviewed May 28, 2026archetype, aep, enterprise, composable, evaluation, egress

Archetype — AEP-Heavy Enterprise Evaluating Composable

Organizational profile

Large enterprise (typically >$1B revenue, >500K customer profiles in AEP). AEP is the primary CDP, often alongside Adobe Campaign, Marketo Engage, or Target in the same licensing agreement. IT and Marketing each hold separate Responsible roles; the Accountable is typically a VP Marketing Ops, CDO, or CTO.

A cloud data warehouse (Snowflake, Databricks, or BigQuery) has grown organically — driven by data science and finance, not marketing. The marketing team is increasingly aware the CDW exists and wants access to what's in it.

Trigger pattern

The evaluation is usually precipitated by one or more of:

AEP egress limits (constraint.aep-first-gen-export-500kb, constraint.aep-second-gen-export-1500kb) blocking analyst self-service reporting.
Query timeouts (constraint.aep-adhoc-query-timeout-10min) frustrating data scientists who expected SQL access to be "like Snowflake."
A new use case (personalized pricing, ML propensity modeling) that the CDW team wants to own end-to-end but that marketing needs for activation.
Second-gen Data Access API ceiling (constraint.aep-second-gen-data-access-api-200kb): when second-gen profile data exceeds 200 KB/profile/year in export volume via the Data Access API path, the Destination SDK constraint creates an architectural fork — accept AEP ecosystem lock-in for the higher (1500 KB) limit, or adopt the AEP-as-edge-node pattern.
Batch activation scheduling congestion (constraint.aep-batch-audiences-per-hour-100, constraint.aep-simultaneous-activation-per-destination-50): When the enterprise has >100 batch-activated audiences, scheduling them in the same export hour produces throughput degradation. The 100-audiences-per-hour performance guardrail and the 50-simultaneous-activations-per-destination guardrail interact: high-density scheduling windows exceed both limits, creating delayed export delivery and missed activation windows. This friction is often discovered during the first production export scale-up.
Ad-hoc activation bottlenecks (constraint.aep-adhoc-activation-max-audiences-80, constraint.aep-adhoc-concurrent-job-per-audience-1): Time-sensitive campaign launches requiring immediate activation of many audiences — flash sales, emergency suppressions, go-live events — are constrained by the 80-audience-per-ad-hoc-job limit and the 1-concurrent-job-per-audience serialization. Multi-destination rapid-fire activation for the same audience queues sequentially rather than executing in parallel. These limits become operational bottlenecks for enterprises with promotional marketing at scale.

Common presenting symptoms

"We can't get our own data out of AEP fast enough."
"Our data science team won't use AEP Query Service — they just run everything in Snowflake."
"We're building the same audience twice: once in AEP for activation, once in the CDW for the ML team."
"Our batch exports don't all deliver by the time our morning email campaign fires — something in the AEP scheduling is backing up."

Recommended direction

pattern.aep-as-edge-node: move audience computation and profile assembly to the CDW; keep AEP as the activation-side consumer. Use reverse-ETL to push CDW-assembled audiences into AEP for activation. This avoids ripping out AEP (expensive and politically difficult) while relieving data-engineering from AEP's compute constraints.

Key tradeoffs

tradeoff.center-of-gravity: CDW becomes the source of truth; AEP becomes a consumer.
tradeoff.data-egress: Egress pressure is relieved once audiences are computed outside AEP.
tradeoff.integration-philosophy: A reverse-ETL tool (vendor-neutral alternatives: Hightouch, Fivetran Activations (formerly Census, acquired May 2025), RudderStack) becomes a new system to operate and govern.
tradeoff.cost-predictability: CDW costs are consumption-based and require FinOps discipline.

Sources

source.packaged-vs-composable-md— AEP-as-Edge-Node pattern; The Activation Tax · agent:org-review · Tue May 05 2026 00:00:00 GMT+0000 (Coordinated Universal Time)
source.helpx-adobe-com.legal-product-descriptions-aep-activation-2026 — AEP Second-Generation Data — Data Access API export entitlement (concrete trigger) · agent:tech-considerations-reviewer · Sun May 10 2026 00:00:00 GMT+0000 (Coordinated Universal Time)
source.experienceleague-adobe-com.en-docs-experience-platform-destinations-guardrails-2026 — AEP batch activation guardrails: 100 audiences/hour throughput, 50 simultaneous activations/destination, 50 mapped attributes/destination, 5M records/file, 80 audiences/ad-hoc job, 1 concurrent ad-hoc job/audience — all grounding the activation-scheduling and file-export trigger cluster for this archetype · agent:org-considerations-reviewer · Thu May 28 2026 00:00:00 GMT+0000 (Coordinated Universal Time)

This node →

applies-whenorg-dim.org-chart-type.matrixed-enterprise— The AEP-heavy evaluating-composable archetype is characteristically a large enterprise with a matrixed org structure — IT and Marketing hold separate Responsible roles, with a VP Marketing Ops, CDO, or CTO as Accountable. Matrixed enterprise org structure is a prerequisite condition: the dual-ownership dynamic (marketing on AEP, data science on CDW) is the organizational root cause of the architectural tension this archetype describes.
applies-whentech-dim.dev-team.backend-data-warehouse— The archetype fires only when a CDW team (backend-data-warehouse dimension) exists and has grown organically alongside AEP. Without a CDW capability, there is no composable alternative to evaluate — the organization has no destination for CDW-computed audiences. The CDW dev team is the enabling condition for the composable evaluation itself.
recommendspattern.aep-as-edge-node— The AEP-as-edge-node pattern is the primary recommended direction for this archetype: move audience computation to the CDW, retain AEP as an activation-edge consumer fed by reverse-ETL. This avoids displacing AEP entirely while relieving the data-engineering team from AEP's compute and egress constraints.
involves-vendorvendor.adobe-experience-platform— OC-070. Adobe Experience Platform (AEP) is the incumbent CDP that defines this archetype — the organization has AEP licensed and deployed, and the composable evaluation is triggered by AEP's export limits, query timeouts, and CDW-team friction with AEP Query Service. The recommended direction (pattern.aep-as-edge-node) is explicitly a strategy for organizations that cannot displace AEP due to licensing and political cost. Mirrors OC-044 and OC-069 pattern.
governed-byconstraint.aep-batch-destination-per-destination-audience-250— OC-078. The 250-audience-per-destination performance guardrail is a direct architectural planning constraint for AEP-heavy enterprise evaluators: organizations with large segment catalogs (the defining characteristic of this archetype) are most likely to approach this ceiling. The guardrail informs the composable-augmentation evaluation — exceeding 250 audiences per destination is one of the operational friction points that drives the composable exploration.
governed-byconstraint.aep-destinations-per-sandbox-100— OC-078. The 100-destinations-per-sandbox system-enforced limit is an enterprise-scale architectural constraint for AEP-heavy evaluators: multi-brand, multi-region, or highly diversified channel stacks — exactly the enterprise scale this archetype represents — can approach this ceiling. Multi-sandbox strategy is an architectural response that affects the evaluation of composable alternatives (which avoid the sandbox limit by activating from the CDW directly).
governed-byconstraint.aep-batch-file-max-records-5m— OC-079. Enterprise organizations in this archetype (>500K profiles, often >5M at large enterprise scale) regularly export large audiences to file-based batch destinations (S3, Azure Blob, GCS). The 5M-record-per-file system limit requires deliberate file-partitioning strategy and downstream ingestion pipeline design — a planning constraint architects in this archetype profile encounter during activation stack design.
governed-byconstraint.aep-adhoc-activation-max-audiences-80— OC-079. Enterprise-scale campaign launches requiring immediate on-demand activation of many audiences (product launches, flash sales, emergency suppression events) are a defining operational pattern for this archetype. The 80-audience-per-job system limit creates operational choreography requirements for large time-sensitive activations: launches requiring >80 audiences must be split across sequential ad-hoc jobs.
governed-byconstraint.aep-adhoc-concurrent-job-per-audience-1— OC-079. Complements constraint.aep-adhoc-activation-max-audiences-80 for the AEP-heavy enterprise. When the same audience must be activated to multiple destinations in rapid succession via ad-hoc jobs (e.g., suppression sync to 10 ad platforms simultaneously), the 1-concurrent-job-per-audience serialization limit extends total activation time linearly with destination count. Multi-destination rapid-fire activation is a common enterprise requirement during go-live events and emergency suppression workflows.
governed-byconstraint.aep-simultaneous-activation-per-destination-50— OC-079. Large activation catalogs (a defining characteristic of this archetype) increase the probability that many audiences share overlapping export windows, approaching the 50-simultaneous-activations-per-destination performance guardrail. Organizations evaluating composable alternatives often discover this guardrail during export schedule audits — it is one of the operational friction points that drives the composable evaluation.
governed-byconstraint.aep-mapped-attributes-per-destination-50— OC-079. Enterprise personalization use cases — personalization engines, email content rendering platforms, dynamic display — often require rich profile payloads (40-60 profile attributes: identity fields, loyalty tier, propensity scores, product affinity, RFM decile). The 50-mapped-attributes performance guardrail is a planning constraint for enterprises building attribute-rich activation configurations.
governed-byconstraint.aep-batch-audiences-per-hour-100— OC-079. Enterprise organizations with large batch activation catalogs — potentially 100-300 audiences across file-based destinations — are most likely to hit the 100-audiences-per-hour throughput guardrail when all exports are clustered at the top of the hour. The hourly throughput constraint requires deliberate export schedule staggering at enterprise scale. Discovering this constraint during operational scaling is a common trigger for the composable-augmentation evaluation (Hightouch activation avoids AEP's hourly throughput ceiling by operating at the CDW layer).

← Referenced by

contrasts-witharchetype.aep-locked-healthcare-cx-evaluator— The two AEP-incumbent archetypes address the same buyer domain — healthcare CX vs. enterprise-composable — via different distinguishing signals; do not conflate. AEP-heavy-enterprise: >$1B, large staffing, composable evaluation driven by AEP export cost/limits, Accountable VP Marketing Ops / CDO. AEP-locked-healthcare-CX: healthcare org, VP Engineering Accountable, HIPAA data residency as the primary constraint.
contrasts-witharchetype.marketing-led-mid-market-composable— OC-065. Mid-market composable (budget $100K–$400K/yr, CMO/VP-Marketing Accountable, no CDW dependency) vs. AEP-heavy enterprise (IT+Marketing joint governance, enterprise scale, AEP incumbent, composable evaluation driven by AEP export cost/limits). Both involve composable CDP consideration; differentiated by scale, governance, and incumbent vendor situation.
contrasts-witharchetype.engineering-led-mid-market-composable-evaluator— OC-108. Both are engineering-accountable composable evaluations; distinguished by budget tier (~$400K vs enterprise AEP licensing) and absence vs presence of an incumbent enterprise CDP to protect or augment.