A B2C personalization engine optimizes for one decision: this individual visitor likes this thing, show them more of it. A B2B engine cannot work that way. The visitor on the page is rarely the only person involved in the buying decision. They are part of a buying group, that group is part of an account, and the account is the unit that ultimately purchases. Marketing Cloud Personalization supports account-based personalization, but the patterns differ from the B2C defaults in ways that matter.
What account-based actually means in MCP
In B2B mode, MCP can be configured to attach visitor profiles not only to individual user profiles but also to account profiles. The account is a separate entity in MCP, fed from the CRM (Sales Cloud most commonly), with its own attribute set: industry, company size, sales stage, deal value, account owner.
When a visitor identifies as belonging to an account, MCP attaches the visitor's behavior to both the user profile and the account profile. Recommendations and segments can read from any combination of the three: visitor (anonymous browsing), user (this individual's history), account (the buying group's collective signal).
This is the key shift. In B2B, the most valuable personalization signal is often the account level, not the individual level. A visitor who has never been to the site before may still warrant aggressive personalization because their account has been engaged for months and the deal is in late stage. A visitor with extensive personal browsing history may warrant cautious personalization because their account just churned.
Identifying accounts from visitor behavior
The integration challenge is mapping visitors to accounts when the visitor has not signed in or is on a corporate network without explicit account linkage. Three patterns work, with different reliability:
1. Authenticated session. The strongest signal. The visitor signs in, the user profile already maps to an account record in CRM, MCP attaches the visit to that account immediately. This is the goal state. Programs that make sign-in friction-free get the most account-level data.
2. Reverse IP lookup. The visitor's IP resolves to a known company. MCP can be configured (often via third-party data providers) to associate the visit with the matching account. Useful for top-of-funnel traffic where sign-in is unrealistic. The accuracy depends on the data provider and falls off as remote work fragments office IPs.
3. Email match from form fills. The visitor fills a form (gated content download, demo request) with a corporate email. The domain maps to an account. Subsequent visits from that browser link to the account.
Most B2B programs use a combination. Authenticated when possible. Reverse IP for prospect identification. Email match for marketing-qualified visitors. Each method has its own confidence level, which the personalization layer should reflect.
Buying group representation
A single account contains multiple buying personas. The CFO who cares about ROI, the IT lead who cares about security, the end user who cares about workflow. A B2B personalization engine that treats all of them the same misses the point.
MCP supports persona-level segmentation within an account. Two implementation patterns:
- Persona attributes on the user profile. Each individual user is tagged with their role and seniority, populated from CRM contact records. Personalization reads role-specific content selection rules.
- Persona inference from behavior. The user has been visiting security-related pages exclusively. The platform infers a security persona, even without a CRM-supplied tag.
Behavior-based inference is more sophisticated and requires enough behavioral data per individual to be reliable. Most production B2B programs combine: CRM-supplied persona for known contacts, behavior inference for anonymous or sparse-data visitors.
Segment design that uses account signals
The most useful B2B segments are not "high-affinity visitors" or "frequent browsers." They are combinations of account-level and individual-level state:
- Active deal in named account, end user persona browsing pricing. Aggressive enablement content, sales notification.
- Cold account, decision-maker persona, first visit. Top-of-funnel education content.
- Closed-won account, post-purchase, expansion-relevant browsing. Cross-sell and upsell content.
- Renewal-window account, primary user persona, support content browsing. Health-check content, customer success outreach.
Each of these segments uses three pieces of information: account state from CRM, persona from CRM or inference, recent behavior from the visitor profile. The intersection produces actionable segments. Single-axis segments rarely do.
Integrating MCP with Sales Cloud and Service Cloud
The account profile in MCP is downstream of the CRM. Account creation, opportunity stage transitions, contact role updates all flow from Sales Cloud into MCP via the Salesforce Data Cloud connector or direct API integration.
Two integration patterns are common:
- Real-time event-driven. Opportunity stage changes fire an event into MCP, account attributes update within seconds. Best for programs where real-time personalization based on deal state matters.
- Scheduled sync. Daily or hourly batch updates of account state. Simpler to operate, sufficient for most use cases.
Service Cloud integration adds support and ticket data to the account record. A visitor whose account has an open priority-1 support ticket should not see "Try our new product" upsell content. The personalization layer can read service state and suppress accordingly.
What does not work the same as B2C
Programs that come from a B2C background often try patterns that do not translate:
- High-frequency holdback testing. B2C lift can be measured in weeks because traffic is high. B2B accounts produce a few visits per month at most. Holdback tests need to run for months. Cohort effects matter more than test design. Plan for slower iteration.
- Recipe-driven recommendation strips. B2C product recommendation strips work because there are thousands of products and visitors choose from many. B2B catalogs may have 20 products, often with explicit ownership boundaries (the prospect's industry only sees relevant content). Custom recipes with manual curation often outperform algorithmic ones at this scale.
- Click-through optimization. B2C metrics maximize click-through rate. B2B metrics often want fewer, higher-quality clicks. A visitor who clicks five things and bounces is worse than a visitor who clicks one thing and reads it for 10 minutes.
When account-based is worth the effort
The pattern works for programs that have:
- A defined account list, sourced from CRM, refreshed regularly.
- Buying cycles long enough that personalizing across multiple visits has compounding value.
- Marketing and sales aligned on what personalization should optimize for at each account stage.
- Enough technical capacity to maintain the CRM-to-MCP integration over time.
The pattern does not work for programs treating MCP as a marginal lift on top of self-service e-commerce. The configuration overhead is real and the lift only emerges when the account-level signal is genuinely actionable.
For programs in the right context, account-based personalization is the difference between MCP being a sophisticated B2B engagement tool and MCP being an oversized recommendation widget. Sapota's Salesforce team scopes account-based work explicitly during program design rather than treating it as an afterthought added later.
Designing or scaling account-based personalization in Marketing Cloud Personalization? Sapota's Salesforce team handles B2B integration patterns, account graph design, and Sales Cloud sync on production engagements. Get in touch ->
See our full platform services for the stack we cover.
