Fullstory × Adobe × Accenture

Adobe Accenture Salesforce

Strategic Partnership · October 2026

Three playbooks.
One behavioral
data layer.

Fullstory is not a new capability — it's the ground truth your agents have been missing. Every interaction your digital customers have is already captured. Three playbooks show what that means for Adobe, Accenture, and Salesforce.

Playbook 1 — In-Session Intelligence

→

Playbook 2 — Intersession Behavioral Intelligence

→

Playbook 3 — Autonomous, Closed-Loop Agentic Systems

→

The Ask

→

57M

Events — zero preconfiguration

103

Behavioral archetypes discovered

3

Joint playbooks, ready to pilot

Today's Agenda

What is Fullstory

The behavioral data layer — Bronze, Silver, Gold. What it captures, how it's activated.

PB1 · In-Session Intelligence

Fullstory for Customer Agents — behavioral session context via MCP. Fewer ticket escalations, higher CSAT, smarter Agentforce and Intercom Fin.

PB2 · Intersession Behavioral Intelligence

91K users → 20 archetypes → CDP-ready profiles feeding Adobe RT-CDP and Salesforce Data Cloud.

PB3 · Autonomous, Closed-Loop Agentic Systems

The architecture that bridges observability and APM — Fullstory as the evidentiary layer, Accenture as the delivery partner.

The Ask

2–3 overlap accounts, named champions, one follow-up.

What is Fullstory · 1 of 2

The Intelligent Digital
Experience Company.

Fullstory automatically captures every user interaction across every digital surface — no manual tagging, no sampling, no configuration required. Three product lines, one behavioral data foundation.

Customer Journey

Fullstory Analytics

Uncover digital behaviors, remove friction, improve the user experience at scale. Session replay, heatmaps, journey maps, funnels, frustration signals.

Employee Experience

Fullstory Workforce

The same behavioral capture applied to internal employee apps — improving the workflows that shape how employees get work done.

Data Activation

Fullstory Anywhere

Activate behavioral data across any downstream system in real time — CDP, CRM, email platform, agent stack. The hub that powers all three playbooks.

Customers who already know this platform

Adobe ★ JetBluePizza HutTaco Bell CaesarsCarMaxKeyBank GrammarlyDuolingoFanatics Vivid SeatsPatagoniaChipotle ServiceTitanFreshBooks

★ Adobe is a Fullstory customer — they know this platform firsthand.

What is Fullstory · 2 of 2

From capture to semantic infrastructure.

What Fullstory has always been

Session replay — the full film of every user's experience
Heatmaps — where attention goes, across any page
Journey maps & funnels — where users drop off and why
Rage click & frustration signals — friction surfaced automatically, no configuration
User segments & dashboards — behavioral cohorts at any scale

What powers the three playbooks

Customer Agent MCP — pipes session context into Salesforce Einstein, Intercom, Zendesk
Fullstory MCP — agentic session review for engineering and coding agents
Anywhere: Warehouse + Activation — behavioral data to any downstream system in real time
Session Summary API — structured behavioral summaries for AI consumption (AJO, Agentforce)

The Medallion Model — where Fullstory sits in the data stack

Bronze

Raw capture — Fullcapture

Every click, navigation, form change, error, and network event. Zero configuration. Zero sampling. The complete behavioral record every playbook below is built on.

Silver

Cooked & indexed behavioral events

Analytically precise, privacy-filtered, structured behavioral data. Powers session replay, Analytics, StoryAI, MCP, and Fullcapture Warehouse. Complete — but not yet semantic.

Gold

Meaning — mined and round-tripped into the data

Behavioral sequences and episode clusters mined from Silver, written back as first-class semantic events. An agent consuming Gold inherits computed meaning for free — no re-derivation on every query.

Data Activation

Fullstory as a hub. Every agent stack as a spoke.

Four distinct channels move behavioral data — and structured behavioral intelligence — to wherever it needs to go. Every playbook uses a different combination.

Live · Warehouse

Anywhere: Warehouse

Behavioral data streamed into CDP, CRM, BI, and warehouse. Adobe RT-CDP, Salesforce Data Cloud, Snowflake, BigQuery. Feeds Playbooks 2 and 3.

Live · Activation

Anywhere: Activation

Real-time behavioral signals power personalization across web, email, and in-app — triggered by what users are doing right now. Feeds Playbook 1.

Fullstory

Behavioral
Data Layer

Live · AI · PB1 + PB3

Customer Agent MCP

Delivers behavioral truth to external CX agents (Salesforce Einstein, Intercom, Zendesk). Tools: get_session_events, get_session_summary.

Live · Engineering Agents · PB3

Fullstory MCP

Token-efficient structured signal stream for coding agents. Every Fullstory customer gets this natively. The evidence layer for the agentic architecture.

Playbook 01 of 03

In-Session
Intelligence

What's happening right now — inside this session, at this moment — is invisible to every agent your customers interact with. Fullstory makes it visible.

Playbook 1 · Problem Statement

What the customer actually experienced — captured in full by Fullstory

Your CX agents are
answering from ticket text.

Every customer-facing AI agent starts from the same place: a support ticket, a CRM record, or a chat transcript. None of them can see what the customer actually did.

What CX agents see today

The ticket. Not the session.

Support ticket: "checkout isn't working"
CRM record: last purchase, loyalty tier, channel preference
Chat history: what the customer said was wrong

What's invisible to the agent

Rage-clicked submit 4 times against an error state
Network failure on POST /checkout at 20:52:14
Tried two payment methods before abandoning
DOM showed incorrect price for 3 seconds before reverting
14 prior sessions — never hit this before

Playbook 1 · Fullstory Enables the Solution

Fullstory Sessions tab · Salesforce Service Cloud · Case: "Trouble adding to cart"

Give support and AI agents
the behavioral context they're missing.

Fullstory streams AI-generated session summaries into your support platform via MCP. Every CX agent now answers with what actually happened — not ticket text.

Salesforce Einstein / Agentforce

get_session_events(session_id) · get_session_summary(session_id)

Not a guess. Not the ticket. The session. Fewer escalations, higher CSAT, faster resolution.

Intercom Fin · Zendesk AI · any MCP-compatible agent

Fin sees what the customer experienced in-session before asking a single clarifying question. First-contact resolution goes up.

⬡ Live Demo — adobe.fullstorydemo.com ⬡ Value Stories — adobe-value-stories.fullstorydemo.com

Adobe is a Fullstory customer. The people in this room work at a company that already bought this platform.

Playbook 1 · Accenture as the Hero

Who delivers this
to hundreds of clients?

Accenture's Adobe and Salesforce practices already sit inside enterprise clients configuring Experience Platform and deploying Agentforce. Adding Fullstory's behavioral layer is a practice extension — not a new engagement type. It's the data upgrade every Salesforce and Adobe implementation is already missing.

The gap in every deployment

Adobe and Salesforce implementations without behavioral context

Every Accenture client who has deployed AEP or Agentforce has agents answering from CRM records and ticket text. The session data that would make those agents meaningfully smarter exists — but no one connected it.

The Fullstory layer

Customer Agent MCP + AEP stream as a repeatable add-on

Fullstory provides the Customer Agent MCP configuration, the AEP data stream, and the Session Summary API integration pattern. Accenture provides the delivery — one proven integration pattern, deployed across the existing client base.

The Accenture advantage

Existing relationships + delivery expertise at enterprise scale

Accenture already has the trust, the architecture blueprints, and the implementation muscle inside these clients. Adding Fullstory behavioral context to every Adobe and Salesforce engagement makes Accenture the firm that finally closes the gap between agentic capability and agentic truth.

The practice play: "Behavioral Context for Agentic CX" — a named Accenture practice offering that wraps Fullstory Customer Agent MCP + AEP stream configuration into every Adobe and Salesforce agentic engagement. Fullstory provides the reference integration. Accenture delivers it at scale.

Playbook 02 of 03

Intersession
Behavioral Intelligence

CDPs know who your customers are. They don't know who they are behaviorally — across dozens of sessions, across weeks of intent signals. Fullstory does.

Playbook 2 · Problem Statement

Adobe RT-CDP and Salesforce Data Cloud
know who. Not who-they-are.

Every CDP implementation Accenture has delivered knows purchase history, loyalty tier, email engagement rate, and demographic segments. None of them know whether a user is a habitual direct buyer, a chronic window shopper, or a frustrated near-converter. That behavioral identity — the dimension that would make every downstream agent meaningfully smarter — is missing from every CDP today.

What RT-CDP / Data Cloud knows without Fullstory

Transactional identity — accurate but incomplete

Purchase history and order value
Email open rate and click rate
Loyalty tier and point balance
Last channel of purchase (app / web / in-store)

These fields drive segmentation that's demographic at its core. Audience Agent, Journey Agent, and Agentforce personalization are limited to what the CRM knows — not what the user's behavior actually signals about their intent.

What's missing from every CDP today

Behavioral identity — the intent layer

Archetype — "Habitual Direct Buyer" vs "Deal-Seeking Browser" vs "Delivery Tracker"
Purchase intent — signals across sessions: was this user trying to buy, or browse?
Promo sensitivity — genuinely driven by discounts vs. ignores them completely
Funnel depth pattern — consistently completes checkout vs. repeatedly abandons at cart
Channel preference behavior — not just last channel, but behavioral dominance across weeks

Playbook 2 · Fullstory Enables the Solution

103 behavioral archetypes.
Zero event configuration.

We ran this pipeline against a national quick-service restaurant brand. 57 million raw Fullstory events, 100,000+ users. No one pre-defined segments. No one configured events. The archetypes emerged from the behavioral record.

Step 1

Segment + Export

91,797 signed-in buyers · 30-day window · 1.3 GB NDJSON via Export API

→

Step 2

DuckDB

One SQL query: 1.3B events → 91K behavioral fingerprints

→

Step 3

Semantic layer

YAML maps raw FS fields → business concepts (page buckets, event tokens)

→

Step 4

Embeddings

all-MiniLM-L6-v2 · 384-dim behavioral vectors per user

→

Step 5

UMAP + HDBSCAN

3D reduction → density clustering → archetypes emerged

→

Step 6

Claude labeling

Opus analyzes centroid users → named archetype + narrative per cluster

→

Output

CDP-ready profiles

12 behavioral signals per user · ingested via Fullstory Anywhere into RT-CDP or Data Cloud

The "irony proof": A user flagged by a naive system as low-value turned out to be a high-frequency recurring buyer — misclassified because a hidden cart/store-binding bug was silently voiding their sessions. The behavioral composite exposed the bug and the lost revenue. A CRM record would never have surfaced this.

Rerunnable on any customer: The pipeline uses only Fullstory's raw NDJSON export — no pre-tagged events, no custom schema required. The same approach runs against any Fullstory customer's data in a few hours. This is not a bespoke engagement — it's a repeatable playbook.

Playbook 2 · The Behavioral Space

91,797 users — plotted by behavioral similarity. Each point is a person.

Proximity = behavioral similarity · 25K sampled for rendering · Click clusters to explore archetypes and their narratives

Playbook 2 · The Per-User CDP Record

From archetype to personalization
in Adobe Real-Time CDP.

Enriched user profile — ingested per user via Fullstory Anywhere

U13

fs_user · authenticated web buyer

🧠 Habitual Direct Buyers

fs_cluster_idU13

fs_behavioral_archetypeHabitual Direct Buyers

dominant_channelWeb/Mobile

sessions_in_window18

orders_in_window9

avg_order_value$24.80

is_high_frequencytrue

is_repeat_buyertrue

is_promo_userfalse

is_deal_seekerfalse

is_mobile_dominanttrue

deepest_funnel_stageconfirmation

12 signals per user. In RT-CDP as a custom profile attribute. Updated via Fullstory Anywhere on each pipeline run.

Adobe RT-CDP — Audience Agent

Build the U13 audience — no rule-writing

fs_behavioral_archetype = "Habitual Direct Buyers"
AND is_promo_user = false AND is_high_frequency = true

Audience Agent segments this cohort automatically. U13 users get a streamlined re-order experience — no promotional banners, mobile-first layout. Personalization driven by behavioral identity, not demographic inference.

Adobe Journey Optimizer — Journey Agent

Trigger at the right behavioral signal

U13 user hasn't ordered in 7 days (unusual for their archetype) — Journey Agent triggers a direct re-order push. Not a promo code (U13 never uses them). A streamlined re-order link matching their fastest checkout path. Behavioral pattern drives the trigger, not a time-based campaign.

For Salesforce / Agentforce: the same per-user archetype record in Data Cloud gives Agentforce the composite intent layer to distinguish a frustrated near-converter from a chronic window shopper before it escalates. Not "what just happened" — who this person behaviorally is.

Playbook 2 · Accenture as the Hero

The CDP partner who brings
the behavioral dimension.

Accenture has implemented more Adobe RT-CDP and Salesforce Data Cloud environments than almost any other firm. Every one of those deployments is missing the behavioral archetype layer. Accenture is the firm that adds it — and earns the right to run the pipeline for the client's entire user base on day one.

Every CDP deployment today

Segmentation built on transactional signals only

Accenture's clients have invested heavily in RT-CDP and Data Cloud. Their audience definitions are demographics, purchase history, and recency. Their agents personalize from those signals — which means they personalize for who a customer was, not who they behaviorally are.

The Fullstory layer

Behavioral archetypes as a native CDP dimension

Fullstory runs the behavioral clustering pipeline. The resulting archetype profiles are ingested via Fullstory Anywhere into the client's existing RT-CDP or Data Cloud as custom profile attributes. No new platform. No new instrumentation. A richer signal on top of what the client already has.

The Accenture advantage

CDP implementation expertise + client trust + Fullstory's pipeline

Accenture already owns the RT-CDP implementation at the client. Adding behavioral archetypes is a practice extension, not a new sale. Accenture brings the delivery relationship; Fullstory brings the behavioral data science. The client gets a meaningfully smarter CDP — and Accenture gets a differentiator in every CDP engagement going forward.

Playbook 03 of 03

Autonomous,
Closed-Loop
Agentic Systems

Every enterprise wants agentic infrastructure. Every enterprise spends too much time chasing bugs. This is the playbook that delivers both — and makes Fullstory's behavioral data layer the foundation that makes agents trustworthy.

Playbook 3 · Problem Statement

Every enterprise wants agentic AI.
Every enterprise is chasing bugs manually.

Every customer Accenture works with is trying to stand up an agentic practice — to use AI as enterprise infrastructure, not just a productivity tool. And every one of those same customers spends too much time, too much engineering headcount, and too much money chasing reported bugs and manually investigating root cause — a process that takes days when it should take minutes.

The manual loop today — 48+ hours

From "something's broken" to "we fixed it"

User reports a problem — support ticket, survey verbatim, or nothing at all
Engineer manually searches for the session, the trace, the log
Impact is estimated, not measured — "how many users hit this?"
Root cause is guessed from partial evidence across disconnected tools
Days pass before a fix reaches production — if it ever does

The gap between observability and APM

Two systems with no bridge between them

APM tools (Dynatrace, New Relic, Sentry) see the server-side failure signature — the trace, the exception class, the status code. They know what broke in the code.

Fullstory sees the user-side behavioral evidence — what the customer actually experienced, how many were affected, and what the dollar impact is. It knows whether it mattered and to whom.

Neither source alone authorizes a code change. And today, no system connects them.

The highest-value bugs are the silent ones — the failures that never throw a server error and never produce a survey response. A price that renders wrong. A date that silently reverts. A balance that shows as unavailable. These are the bugs that cost the most and are found last.

Playbook 3 · Fullstory Enables the Solution

From signal to shipped fix, without guessing.

query_sessions · detect_struggle_signals · conversion_delta
Fullstory MCP — behavioral evidence on demand

Phase 1 (0–60 days): Human in every loop. Beats the 48-hour manual baseline on time-to-evidence.

Autonomy expands by measured precision. Each stage clears its precision metrics before the next unlocks. Scope expansion is a board decision.

Control Plane

Triage Orchestrator — no write access, cannot assert root cause

Owns the case file. Routes to specialist agents. Enforces two-source reconciliation before any case advances. Deterministic code between agents — not another model call.

Fullstory MCP

Session Evidence Agent

Observe-only. System prompt rejects "because," "caused by," or a fix. Proves a failure happened. Cannot claim why.

Sizing

Cohort Quantification

100-session floor before publishing any impact figure. Every estimate ships with a sample size.

APM · Dynatrace / New Relic

Ops Correlation Agent

Server-side failure signature. This + evidence agent corroboration authorizes engineering work.

Remediation

Engineering Agent

Draft PRs only. Zero merge rights — enforced by platform permissions. A human always merges.

Playbook 3 · Accenture as the Hero

The firm that carries enterprises
into the agentic era.

Accenture's rich experience scoping and deploying enterprise infrastructure at this size and scale makes Accenture the obvious partner to build and deliver this architecture. No other firm has the combination of enterprise client trust, platform-agnostic delivery expertise, and agentic practice maturity to do this at scale.

What every enterprise needs

Agentic infrastructure that enterprises can actually trust

Every CIO and CTO in the enterprise wants AI as infrastructure — not a chatbot, not a copilot, but real autonomous systems that compress incident response from days to minutes. They can't get there with agents that hallucinate root cause and ship confident wrong fixes. The architecture has to be defensible before the board will approve it.

What Fullstory provides

The behavioral ground truth that makes the architecture honest

Fullstory provides: Fullstory MCP as the behavioral evidence source, the reference architecture and guardrail patterns, the golden-set evaluation methodology, and the Session Evidence Agent constraints that prevent the system from asserting unverified cause. The architecture is credible because it's honest about what behavioral data can and cannot prove.

The Accenture advantage

The only partner with the scale to deliver this across enterprise clients

Accenture brings the client discovery, architecture decisions, program governance, and the delivery of the agent fleet itself. Accenture has stood up APM, observability, and incident management infrastructure at the world's largest enterprises. This is the natural next chapter — autonomous agentic infrastructure built on that foundation, with Fullstory as the behavioral ground truth that makes every agent in the fleet defensible.

The practice play: "Trustworthy Agentic Delivery" — a named Accenture practice offering with Fullstory as the behavioral evidence layer. A staged autonomy model. A governance charter. A golden-set evaluation method. A deliverable that any client's risk and audit teams can sign off on. Accenture delivers the first one; the reference architecture makes every subsequent one repeatable.

The Ask

Three playbooks. Many shared clients.
One thing needs to change today.

We have the playbooks. We have the architecture. We have customers who are already buying from multiple people in this room. The only thing missing is the joint motion — taking these solutions to those shared clients together.

1

Name 2–3 overlap accounts as pilot candidates

Each side identifies shared clients where one or more of the three playbooks would land. We don't need a new logo — we need a friendly environment where we can prove the joint motion, build a repeatable playbook from it, and create the first co-branded deliverable.

2

Name a champion on each side

A named person from Adobe and a named person from Accenture who owns carrying this forward internally. The follow-up meeting needs an owner — not a to-do item on someone's list.

3

Agree on a format for what "co-branded" looks like

A joint solution brief. A co-presented workshop at a pilot client. A shared POC proposal. The right vehicle is a follow-up conversation — but the goal is a deliverable with all three logos on it, not a one-sided sell. We can build that together.

Lane Greer · Fullstory | lane@fullstory.com | Confidential · October 2026