Salesforce · Data 360 · AI

Salesforce Data Cloud AI: How Data 360 Uncovers New Revenue Opportunities

Most of the buying signals, renewal risks and upsell openings a business needs are already in its data, spread across systems that don't talk to each other. Here is how Salesforce Data 360 (formerly Data Cloud) and its AI features find those opportunities, act on them, and what it takes to get the first one live without overspending.

Key Takeaways

  • Salesforce Data Cloud was renamed Data 360 on October 14, 2025. Salesforce says the functionality did not change, so "Data Cloud AI" and "Data 360 AI" describe the same thing.
  • Data 360 connects data from Salesforce and other systems, unifies it into profiles for people and accounts, and calculates metrics such as lifetime value. AI Models, Flow and Agentforce then turn that unified data into predictions and actions.
  • The opportunities it uncovers are concrete: accounts likely to buy, customers at risk before renewal, inactive accounts worth re-engaging, precise audiences, and better-informed reps and agents.
  • Data 360 is priced on consumption. Bringing data in is free; segments, insights, queries and real-time features use credits. Design for one use case first.
  • Start with one measurable opportunity, the minimum data it needs, and one automated action. Expand once it beats the baseline.

Salesforce Data Cloud AI is the combination of Salesforce Data 360 (the new name for Data Cloud) and the AI that runs on it: predictive models, generative prompts and Agentforce agents. Data 360 brings scattered customer data into one place, and the AI finds the patterns in it: who is ready to buy, who is about to leave, and what to do next. We first published this article in September 2024. Salesforce has renamed and extended much of the product since, so we have rewritten it from current Salesforce documentation.

This guide is for revenue, marketing and operations leaders at growing and mid-market companies, and the Salesforce admins and architects who work with them. It covers what Data 360 does, six specific opportunities it can surface, how predictions turn into automated actions, what it costs, and how to start small.

Section 01 · Short Answer

What Is Salesforce Data Cloud AI?

Salesforce Data Cloud AI means using Data 360 (formerly Data Cloud) to unify customer data from every system, then applying Salesforce AI to it: insights that calculate metrics, models that predict outcomes, and flows and agents that act on the results. The data layer makes the AI accurate; the AI makes the data useful.

Salesforce renamed Data Cloud to Data 360 on October 14, 2025, and its help pages state that "the functionality and content remains unchanged". Some screens and features still carry the old name, such as the Data Cloud-Triggered Flow. The table maps the parts of the product mentioned in this article.

PartWhat it doesOlder name you may see
Data 360Connects, harmonises and unifies data from Salesforce and other systemsData Cloud, Customer Data Platform
Identity resolutionLinks records for the same person or account into a unified profileUnchanged
Calculated, streaming and real-time insightsCompute metrics such as lifetime value, engagement or open casesUnchanged
AI ModelsBuilds predictive models with clicks, or connects external modelsEinstein Studio, Model Builder
Data actions and Data 360-triggered flowsStart automation when data or an insight changesData Cloud-Triggered Flow
Agentforce Data LibraryIndexes content so AI agents can answer from itUnchanged
Section 02 · The Problem

Why Do Companies Miss the Opportunities in Their Own Data?

Because the signals that point to an opportunity sit in different systems. A buying signal on the website, a rise in support cases and an unpaid invoice may all concern the same account, but no one sees them together. Salesforce's own research finds that leaders believe their most valuable insights are in data they cannot use.

Salesforce's State of Data and Analytics report (second edition, 2025, a survey of 3,800 data and analytics leaders and 3,852 business leaders) puts numbers on the problem:

  • 76% of business leaders say they are under growing pressure to drive business value with data.
  • Less than half (49%) say they can reliably generate timely insights.
  • Data and analytics leaders estimate that 19% of their company's data is siloed, inaccessible or otherwise unusable, and 70% believe their most valuable insights are inside that 19%.
  • 89% say a strong data foundation is the most critical factor for successful AI.

These are vendor survey figures, so read them as a direction rather than a benchmark. The pattern matches what we see in growing and mid-market companies: CRM, marketing automation, billing, product usage and support each hold part of the customer, and the reports that join them up are built by hand once a quarter. By the time the spreadsheet shows that an account is at risk, the renewal has already been lost.

AI does not fix that on its own. A lead score trained only on CRM fields cannot see product usage, and an agent grounded only in Knowledge articles cannot see an open order. Uncovering new opportunities starts with putting the signals in one place, which is the job Data 360 does.

Section 03 · How It Works

How Does Data 360 Turn Scattered Data Into AI-Ready Profiles?

Data 360 works in layers. It brings data in or federates it from a warehouse, maps it to a common data model, links records for the same person or account through identity resolution, and calculates insights on the result. AI, segments and automation all read from that unified layer rather than from each source.

Layer 1 · Connect

Ingest or Federate Data From Every Source

Salesforce clouds, web and mobile events, files and external systems stream in through connectors. With zero copy, data in Snowflake, Databricks, Redshift or BigQuery can be queried where it lives instead of being copied.

↓ mapped to the Data 360 data model ↓
Layer 2 · Harmonise

Map Every Source to One Data Model

Raw data lands in data lake objects and is mapped to data model objects (DMOs), so "customer", "order" and "case" mean the same thing whatever the source system.

↓ match and reconciliation rules ↓
Layer 3 · Unify

Resolve Identities Into Unified Profiles

Identity resolution rulesets can unify individuals, accounts, leads or households. Match rules decide which records belong together; reconciliation rules (Last Updated, Most Frequent, Source Priority) decide which value to show in a unified field.

↓ metrics on the unified data ↓
Layer 4 · Insights

Calculate the Metrics That Signal an Opportunity

Calculated insights compute metrics such as customer lifetime value or satisfaction score and can rank customers by spend or engagement. Streaming insights work on event windows as short as one minute, and real-time insights return results in milliseconds.

↓ predictions, segments and actions ↓
Layer 5 · AI and Action

Predict, Target and Act

AI Models score the profiles, segments publish audiences to activation targets, flows and data actions respond to changes, and Agentforce agents use the unified data to answer and act.

One detail shapes every project. Salesforce states that unified profiles "aren't golden records": identity resolution links source records like keys on a ring, and it "doesn't pick winning values or override any existing data". Duplicates and bad values in your source systems remain there, so data cleanup is still part of the work. Salesforce's Data 360 architecture guide covers the full design.

Section 04 · Opportunities

Which Revenue Opportunities Can Salesforce Data Cloud AI Uncover?

The most valuable opportunities are the ones a single system cannot see: accounts showing buying intent, customers drifting toward churn, inactive accounts worth re-engaging, audiences precise enough to convert, and conversations where reps and agents need the whole history. Each one maps to a specific Data 360 feature.

1. Accounts Ready to Buy

A propensity-to-buy model in AI Models scores unified profiles on purchase likelihood, so sales teams can focus on the prospects most likely to convert.

2. Renewals at Risk

A calculated insight such as open cases per account, copied onto the Account record, shows reps which customers need attention before a renewal conversation.

3. Inactive Accounts to Re-engage

Streaming insights watch engagement in near real time and can trigger a data action when an account goes quiet, so re-engagement starts before the account is lost.

4. Audiences That Convert

Segments built on unified profiles and insights publish to Marketing Cloud, Commerce Cloud or ad platforms such as Amazon Ads, without exporting lists by hand.

5. Full Context in the CRM

Related lists and data graphs bring orders, usage and engagement from other systems into Sales and Service records, so reps see the whole relationship in one place.

6. Agents That Answer Accurately

Data 360 indexes documents and records for retrieval, so Agentforce agents answer from your own content and customer data rather than general knowledge.

The table shows which feature delivers each opportunity and where people see the result.

OpportunityData 360 featureWhere it shows up
Accounts ready to buyAI Models (predictive model on unified profiles)Score on the record, in a flow, a segment or a prompt
Renewals at riskCalculated insight plus copy field enrichmentFields on the Account record and in reports
Inactive accountsStreaming insight plus data actionA flow that creates a task, a case or a journey entry
Audiences that convertSegments and activationMarketing Cloud, Commerce Cloud, ad platforms, file storage
Full context in the CRMRelated lists and data graphsContact, Account and Case pages
Accurate agentsSearch index, retriever and Agentforce Data LibraryAgent answers and prompt templates

Salesforce's copy field enrichment unit uses the second example almost exactly: a calculated insight of total and open cases per account is copied onto Account fields, so a sales rep "can quickly see which customers most need to purchase a new software pack". The same insight could just as easily flag accounts with too many open cases before a renewal.

Section 05 · Predictive AI

How Does Data 360 Predict Which Customers Will Buy or Leave?

Through AI Models, formerly Einstein Studio, which sits inside Data 360. Business users can build a predictive model "using clicks not code" on unified data, or connect a model their data science team already runs in Amazon SageMaker, Google Cloud Vertex AI or Databricks. The predictions can then be used in flows, segments, prompts and agents.

Salesforce's Trailhead module on AI Models describes three ways to work:

Build a model from scratch.
→ Train a predictive model on Data 360 data with clicks. Salesforce's own example is "Propensity to Buy Scoring", which predicts which prospects are most likely to purchase so sales teams can focus their effort.
Bring your own model.
→ Connect a predictive model built in Amazon SageMaker, Google Cloud Vertex AI or Databricks directly to Data 360, so existing data science work reaches the CRM.
Bring your own LLM.
→ Add a foundation model hosted on an external platform for generative use cases.

The value comes from what happens to the prediction. According to Salesforce Help on AI Models, model results can be used "in flows, prediction jobs, batch data transforms, agents, and prompt templates", and from Apex and REST. A churn-risk score, for example, could feed a segment for a retention campaign, a flow that alerts the account owner, and a prompt that drafts the outreach. Because AI Models runs inside Data 360, it also sits under the Einstein Trust Layer.

If your Salesforce org has enough conversion history, native Einstein Lead Scoring may cover the first need without a custom model. Our Einstein Lead Scoring service starts there. A custom model in Data 360 earns its cost when the prediction depends on data from outside the CRM, such as product usage, payments or web behaviour.

Section 06 · Automation

How Do You Act on Data 360 Insights Automatically?

With data actions and Data 360-triggered flows. A flow can start when a unified record changes or when a calculated insight crosses a threshold, and a data action can send the event to Salesforce, a webhook or Marketing Cloud. The insight becomes a task, a case, a journey or an alert without anyone running a report.

Salesforce's Trailhead unit on flows and agents gives two patterns worth copying:

  • React to a change in a record. A flow triggered by a change in a data model object, such as emailing a guest when their loyalty level becomes "Diamond".
  • React to a change in an insight. A flow that responds when a "Number of Abandoned Carts" calculated insight reaches three or more and creates a case so someone follows up.

Data actions extend this beyond Salesforce. Targets include the Salesforce event bus (for flows), webhooks, Amazon Kinesis and Marketing Cloud Engagement, and Salesforce notes that a new action starts receiving events in less than 15 minutes. Real-time data actions can trigger events from Flow Builder in milliseconds.

Know the limits before you design. Flows can read data model objects and calculated insight results, but not raw data lake objects. Streaming insights can drive data actions but cannot be used in segments or activation. Real-time data graphs, which the real-time features depend on, require the Real-Time SKU. For complex hand-offs to systems outside Salesforce, such as ERP or billing, an integration layer like MuleSoft usually carries the action the rest of the way.

Section 07 · Agentforce

How Does Data 360 Ground Agentforce and Generative AI?

Data 360 is the retrieval layer for Salesforce's generative AI. It chunks and vectorises documents into a search index, and a retriever passes the most relevant passages into a prompt or an agent. Salesforce requires Data 360 for all Agentforce use, and agents built on it can read unified customer profiles as well as content.

Salesforce's Trailhead unit on trusted agents defines the three parts: a vector data store holds embeddings, a search index stores the chunked and vectorised content, and a retriever is "the link between the prompt and the search index".

OptionWhat it sets upWhen it fits
Agentforce Data LibraryA preconfigured RAG setup: vector store, search index and retriever built with default settingsOne source per library: Knowledge articles, uploaded files, web search or a custom retriever. No zero copy.
Advanced Data 360 setupYour own search index, retrievers (including ensemble retrievers) and data graphsSeveral sources, hybrid search, grounding in structured customer data such as orders

Salesforce's guide to retrieval augmented generation covers both paths. The practical lesson is the same either way: an agent is only as good as the content it can find. Our Agentforce data and knowledge integration work tests retrieval before the agent itself is built, because poor answers usually trace back to the content and the index, not the model. For how agents reason and act, see our guide to how Agentforce works.

Section 08 · Industries

What Do Data 360 AI Use Cases Look Like by Industry?

The opportunity changes by industry, but the pattern stays the same: join the signals no single system holds, calculate the metric that matters, and act on it. These are typical use cases we scope with growing and mid-market companies. They are illustrations, not reported client results.

IndustrySignals to unifyOpportunity it surfaces
SaaSProduct usage, support cases, billing, CRM opportunitiesExpansion-ready accounts and renewals at risk, flagged on the Account before the renewal date
Financial servicesAdvisor CRM, portfolio or account data from a warehouse, service historyClients due a review or showing signs of moving assets, routed to the right advisor
Real estateListing views, enquiries, viewings, past transactionsBuyers whose behaviour suggests they are ready to view or offer, and past clients likely to sell
HealthcarePatient or member engagement, appointments, service records (with strict access controls)Missed-follow-up and re-engagement lists for care and service teams
ManufacturingERP orders, distributor data, service and warranty casesReorder timing, accounts with falling order volume, and service-led upsell
Retail and e-commerceWeb and app events, orders, loyalty, serviceAbandoned-cart follow-up, high-value segments and next-purchase audiences

Regulated industries need extra care. Decide which data each user, segment and agent can see before you connect sensitive sources, and check consent and residency rules for every activation target. Data 360 provides the controls; the policy is yours to design.

Section 09 · Cost

What Does Data 360 Cost, and How Do You Control Consumption?

Data 360 is priced on consumption. Salesforce offers credit-based pricing, where Flex Credits are deducted as you use features, or profile-based pricing per 1,000 profiles a year. Bringing data in by batch and zero copy is free; segments, insights, queries, unstructured processing and real-time features use credits.

Salesforce's Data 360 pricing page, which it marks as "subject to change", listed these options when we checked in October 2026:

ModelList priceFits when
Flex Credits$500 per 100,000 credits, usable across Data 360 and AgentforceUsage varies, or you are starting with one use case
Profiles$240 per 1,000 profiles a year (includes 25 calculated insights, 25 transforms, 100 segments)You want a predictable, flat cost
Enterprise Profiles$420 per 1,000 profiles a year (100 insights, 100 transforms, 500 segments)Many segments and insights per profile

Salesforce also states that customers "can provision Data 360 for free with a limited amount of storage and consumption credits", which is enough for a proof of concept. The Core, Advanced and Max editions announced in September 2026 include Flex Credits, which the pricing page says can move between Data 360 and Agentforce. Your actual price depends on your contract.

Three habits keep consumption under control. Connect only the sources and fields the first use case needs. Set refresh schedules to the speed the decision actually needs, because real-time costs more than daily. Watch usage from the start in Digital Wallet. Zero copy is not automatically cheaper either: it moves the query compute to the connected warehouse, so that bill can rise. Our Data 360 services start with a cost model for exactly this reason.

Section 10 · Pitfalls

What Mistakes Stop Data 360 From Finding Opportunities?

Most stalled Data 360 projects fail on design, not technology. Teams connect every source before they choose a use case, trust identity resolution without testing it, or build insights that no flow, segment or person ever acts on. Each mistake costs credits and delays the first result.

1. Connecting everything first. Every stream, refresh and insight consumes credits. Start from one opportunity and connect only the data it needs.
2. Untested identity rules. Over-matching merges two different customers; under-matching leaves the duplicates you were trying to remove. Test match rules on a sample before you act on unified profiles.
3. Treating the unified profile as a golden record. Identity resolution links records but does not clean them. Fix source data quality as part of the project.
4. Insights with no action attached. A score that no one sees or uses changes nothing. Design the flow, segment or record field that uses the insight at the same time as the insight.
5. Choosing the wrong insight type. Streaming insights cannot be used in segments, and max, min and sum need calculated insights. Match the insight type to how you will use it.
6. No baseline. Without today's conversion, retention or response numbers, no one can prove the opportunity was real. Record them before go-live.
Section 11 · Getting Started

How Should a Growing Company Start With Data 360 AI?

Pick one opportunity with a measurable outcome, connect the minimum data it needs, test identity resolution, build one calculated insight, and wire it to one action. Prove the result against a baseline in a sandbox or the free tier, then put it into production and expand to the next use case.

Data 360 is likely worth it if the opportunity depends on data outside your CRM (product usage, orders, web behaviour, a warehouse), you need one view of a customer across several systems, or you plan to use Agentforce, which requires it.

It may not be needed yet if everything the use case needs is already in Sales Cloud, your CRM data is not yet trusted, or a native feature such as Einstein Lead Scoring or Marketing Cloud segmentation already answers the question.

A sequence that works:

1. Choose one opportunity.
→ For example renewals at risk, expansion-ready accounts or abandoned carts, with an owner and a number to move.
2. Record the baseline.
→ Today's renewal rate, conversion rate or response time, before anything changes.
3. Connect the minimum data.
→ Usually Sales Cloud plus one other source. Use zero copy if the data already lives in a supported warehouse.
4. Test identity resolution.
→ Run the ruleset on a sample and check for over-matching and under-matching before anyone relies on the profiles.
5. Build one insight or prediction.
→ A calculated insight first; an AI Models prediction when a rule-based metric is not enough.
6. Wire one action, then measure.
→ A record field, a flow or a segment that uses the result. Compare against the baseline, then add the next use case.

If you already run Marketing Cloud, read our view on Marketing Cloud and Data 360 first: the honest answer to "do we need both?" is frequently no. If AI agents are the goal, our Agentforce readiness assessment checks processes, data and permissions before you commit to a build. For how Data 360 fits Salesforce's wider AI stack, see our guide to the Einstein 1 Platform and what it became.

Section 12 · Bottom Line

The Bottom Line

Data 360 does not create opportunities. It reveals the ones already in your data by putting the signals side by side, and Salesforce AI turns them into a score, a segment or an action someone can use. The companies that get value from it start narrow: one opportunity, the minimum data, one action and a number to beat.

  1. 1

    Data Cloud is now Data 360; the functionality is the same, and AI Models replaced the Einstein Studio name.

  2. 2

    Identity resolution and calculated insights are what make AI predictions trustworthy, and unified profiles are not golden records.

  3. 3

    Every insight needs an action: a record field, a flow, a segment or an agent.

  4. 4

    Consumption pricing rewards narrow, well-designed use cases.

  5. 5

    Start with one opportunity and a baseline, then expand.

Ready to Start

Find the First Opportunity Data 360 Should Uncover for You

On a 30-minute discovery call, a Twopir Salesforce architect will look at your data sources and goals, tell you whether Data 360 is the right tool, and outline the first use case and what it would consume.

Schedule a Discovery Call

FAQ

Frequently Asked Questions

Yes. Salesforce renamed Data Cloud to Data 360 on October 14, 2025, and states that the functionality is unchanged. Some screens and features still use the old name, such as the Data Cloud-Triggered Flow, so you will see both names in documentation and in your org.

Data 360 provides the unified data that Salesforce AI runs on, and some AI is built in: AI Models (formerly Einstein Studio) for predictive models, Einstein Segment Creation for building segments from plain language, and the search index and retrievers that ground generative AI and Agentforce agents.

Yes. Salesforce states that Data 360 must be provisioned and enabled for all Agentforce use, and that features such as the Agentforce Data Library and the Einstein Trust Layer do not work without it. You do not need to connect every data source to start, only what your agents use.

A calculated insight is a metric Data 360 computes from your unified data, such as customer lifetime value, satisfaction score or open cases per account. It can rank customers, recompute when the base data changes, and feed segments, activations, flows and Salesforce record fields.

Yes. Data 360 zero copy has built-in support for Snowflake, Databricks, Amazon Redshift and Google BigQuery, so data can be queried where it lives. The query compute then runs in the connected warehouse, so plan for that cost as well as Data 360 credits.

Data 360 is priced on consumption. Salesforce lists Flex Credits at $500 per 100,000 credits, or profile-based pricing from $240 per 1,000 profiles a year, subject to change. Batch ingestion and zero copy are free; insights, segments, queries and real-time features use credits. A free tier with limited credits exists for trials.

No. Salesforce states that unified profiles are not golden records. Identity resolution links source records for the same person or account and uses reconciliation rules to choose display values, but it does not change or delete source data. Cleaning duplicates and bad values in the source systems is still required.

It depends on the number of sources and how clean the data is. One use case with Sales Cloud and one other source can often be proven in weeks; Twopir's typical Data 360 engagement runs 8 to 16 weeks, including the data model, identity resolution, activation and cost controls.

TC

Twopir Consulting

Salesforce Partner · HubSpot Partner

Twopir Consulting helps growing and mid-market companies solve complex CRM, integration, and business system challenges. We implement, rescue, optimize, and integrate complex CRM and revenue systems, including Data 360, Einstein and Agentforce on Salesforce, for 500+ clients over 12+ years.