Last week a national insurer sent Dinasha an email that missed on three basics. First, it addressed her by her maiden name, a name she legally changed 20 years ago. Second, it promoted a motorcycle policy, even though she’s never held a C-class license or ridden a motorcycle. And finally, it showed no awareness that she switched her insurance to a competitor a decade ago.
This type of message is not rare, and most of us are used to swiping to delete after seeing one. But with the current promise of agentic marketing and 1:1 personalization, receiving a message with these errors gave us pause.
When the foundation is cracked, adding more “personalization”with AI on top doesn’t rescue the experience. It scales the bad experience. The order of operations matters: audit your data and then let agents help you move faster.
Make sure you have good profiles before launching personalization
True 1:1 personalization begins with a unified, accurate customer profile. In large organizations, records live in CRM and commerce, service and web analytics, advertising platforms, and data warehouses. Each system updates on its own time. Older dynamic fields linger in templates. IDs aren’t resolved across the board. Over months and years, those small misalignments become big customer errors.
A deep dive into agentic personalization
Heading to San Francisco next week? Dinasha Cellura and Blake Miller will chat about how Typeface and Salesforce execute 1:1 personalization. This session will be on Oct 14 at 12:30pm PST in the Marketing Cloud Campground.



Fixing this is not a “nice to have.” It’s the precondition for everything else. Resolve duplicates and conflicts. Merge the signals that matter: consent, preferences, lifecycle stage, product ownership, service history. If the profile is wrong or outdated, every decision downstream inherits the mistake.
Before marketers can use Agentforce or Typeface agents to deliver true personalization, their first step must be data hygiene. Validating and enriching audience segments in Salesforce Data Cloud ensures that every Agentforce and Typeface agentic workflow runs on clean, unified customer profiles. This turns potential erroneous data into meaningful action.
Take the time to “spring clean” your workflows
Many organizations admit they’re still running triggered campaigns on outdated workflows built years ago by SIs or operations team systems that no longer align with current conversion goals or performance metrics. Revisiting and cleaning these workflows may be a heavy lift, but it’s a high-value investment that prevents far greater risk: sending erroneous or irrelevant messages to valued customers and losing hard-earned mindshare.
Replace one-off data extracts with live event streams in Data Cloud that capture the few critical signals that should always fire, such as new purchasers, churn indicators, service moments and key digital behaviors. Dedicate and train teams to perform this foundational work — the kind of judgment and prioritization that only humans can do — to ensure every campaign is powered by clean, current and contextually correct data.
Turn your campaigns into conversations
With a solid foundation, agents help teams move from mass email sends to bolder, personal conversations. Journeys stop being monoliths and become modular, able to guide people through the perfect experience as their context changes. Messages respond to what a person actually does, not just to the date on a calendar or attribute in the “First Name” field. Audience selection, creative assembly, activation and learning happen in one continuous loop instead of four disconnected projects.
The role of the marketer changes, too. You set the strategy, the goals, the constraints and the approval points. Agents carry the work across steps, propose alternatives and keep iterating as results come in. The payoff is speed with safety: fewer handoffs, shorter cycle times and no more “Dear Wrong Name” moments.
Gauge your readiness by answering five simple questions:
- Do we have a single source of truth for identity and consent, and is it current?
- Can we trace every field our campaigns rely on back to a governed system of record?
- Are our journeys responsive enough to adapt to customer moments cleanly as their context changes?
- Can we approve or block agent actions based on risk for offers, segments and channels?
- When something goes wrong, can we quickly see what changed and why?
If you can’t answer “yes” to most of these, you need to rebuild your foundation. Agents will only magnify the disconnected data and processes.
What personalization success looks like
Let’s consider a practical case. You’re targeting lapsed mid-market accounts. After the foundation reset, Agentforce surfaces ex-customers who match your ideal profile. Consent rules ensure suppressions aren’t overridden. For opted-in lookalikes, an agent proposes two cohorts and a plan: a short win-back path for recent evaluators and a longer educational path for dormant accounts.
The agent produces an on-brand email, landing page copy and ad variants, including multilingual options within your guardrails. You approve the campaign in Slack. The agent hands off journeys automatically based on real behavior: open, reply, ignore, download and performance signals feed the next iteration without your staff needing to wrangle spreadsheets..
Personalization fails when profiles are not valid and journeys are fractured. Fix the foundation identity, consent and orchestration and then let agents scale the right things. Companies that will get it right with agentic marketing will have fortified their foundation, before bolting on AI, will have the most success moving forward.
Behold the magic of Dreamforce 2025 virtually!
Join us online for three days of learning, inspiration and fun. There’s nothing else like it.


