What is Salesforce Personalization?
Salesforce Personalization (SP) is a Customer 360 application built on top of Data Cloud. It delivers personalized experiences — such as product recommendations, targeted banners, and custom content — across web, mobile, email, and agentic channels.
At its core, SP uses:
- Real-time behavioral data captured through the Salesforce Interactions SDK (Web SDK and Mobile SDK) to understand what individuals are doing right now.
- Unified customer profiles created through Data Cloud's Identity Resolution (IR), which stitches together anonymous and known interactions into a single view of each person.
- Data Graphs — pre-calculated views of Data Model Objects (DMOs) — that give the decisioning engine fast access to profile data, engagement history, segment memberships, and item catalogs.
- Calculated Insights (CIs) — multidimensional metrics computed over your data — that power rules-based recommendations and targeting logic.
- Machine-learning-driven recommendations that select the most relevant items for each individual based on configured business objectives (maximize revenue, maximize clicks, etc.).
How Does It Work?
The personalization data flow follows five key steps:
- Ingest — User interaction data is captured by the Salesforce Interactions SDK and sent to Data Cloud. Data flows into two processing layers simultaneously: a real-time layer for immediate processing and a standard layer for regular batch processing.
- Identify — Data Cloud's Identity Resolution performs real-time matching to identify the user profile. If the user is unknown, a new anonymous profile is created.
- Update — The known or anonymous profile in the real-time data graph is updated with the ingested engagement and profile event data.
- Compute — Real-time insights and segments are calculated, and the real-time profile data graph is updated with metrics and segmentation results.
- Decide — When a personalization decision is needed, the Personalization service calls the Data Cloud profile API to obtain the updated individual profile from the real-time profile data graph, evaluates targeting rules, and returns the winning decision.
What You'll Learn
This guide is organized into seven sections, each building on the previous one:
| Section | What You'll Do |
|---|---|
| Setup & Permissions | Configure licenses, permission sets, deploy the Personalization datakit, and plan your implementation blueprint |
| Data Capturing & Modeling | Install the Web and Mobile SDKs, build sitemaps, map data into Data Cloud, configure Identity Resolution, and create Data Graphs |
| Web Implementation | Configure recommenders, response templates, personalization points, decisions, experiments, web templates, and the Web Personalization Manager |
| Mobile Implementation | Deliver personalized experiences in native iOS and Android apps |
| Personalization API | Use the Decisioning API for server-side and headless personalization |
| Experimentation | Set up A/B tests, define metrics, allocate traffic, and analyze results |
| Batch Personalization | Run personalization at scale for email, push, and offline channels |
Additional Resources
Salient Features
Real-Time Data Ingestion
Capture user behavior as it happens using the Web SDK (for websites) and Mobile SDK (for native iOS and Android apps). Every page view, product browse, cart action, and purchase is streamed into Data Cloud in real time.
Unified Customer Profiles
Identity Resolution stitches together anonymous browsing sessions, email addresses, phone numbers, login credentials, and other identifiers to create a single, unified view of each individual — across channels and devices.
Profile and Item Data Graphs
Data Graphs are pre-calculated views built from your Data Model Objects. A Profile Data Graph combines a unified individual with their engagement history, segment memberships, and calculated insights. An Item Data Graph organizes business objects — products, articles, events — so they can be used in recommendations.
Manual Content and ML-Powered Recommendations
Two personalization types are available:
- Manual Content — Return static, targeted text or images (banners, infobars, pop-ups) without needing item data. Great for quick wins.
- Recommendations — Return dynamic, ML-driven item recommendations powered by recommenders. Recommenders come in two flavors:
- Rules-Based — Driven by Calculated Insights (top sellers, most viewed, co-browse, co-buy).
- Objective-Based — ML-driven, configured around business outcomes like maximizing revenue or clicks.
Web Personalization Manager
The Web Personalization Manager (WPM) is a WYSIWYG, no-code experience builder. Append ?sf_personalization_wpm to your website URL, and business users can visually add, configure, preview, and publish personalized experiences — without writing any code.
Experimentation with A/B Testing
Test different decisioning strategies against each other using Experiments. Define primary and secondary metrics, create cohorts with customizable traffic allocations, optionally include a control group, and measure which approach performs best.
Attribution Analytics and Pipeline Intelligence
Track and understand personalization performance with two analytics frameworks:
- Pipeline Intelligence — Operational metrics on the Personalization request pipeline: decision requests, personalization points served, unique individuals targeted.
- Attribution Service — Business metrics showing how visitor engagement with personalized content drives revenue, product orders, and other business outcomes.
Decisioning API
Use the Decisioning API to request personalization decisions programmatically from any channel — server-side applications, headless commerce platforms, kiosks, or any system that can make REST API calls. Personalization as a Service.
Batch Personalization
Generate personalized recommendations at scale on a schedule for offline channels — email campaigns, push notifications, CRM-driven next-best-action workflows — without requiring real-time decisioning.
Agentforce Integration via Invocable Actions
Extend personalization into the Agentforce ecosystem using invocable actions:
- Get Recommendations — Inject individualized, ML-powered recommendations into agentic conversations.
- Get Context — Provide agents with the relevant customer data they need to address questions intelligently.
Found this guide helpful?
Buy Me a Latte ☕