Salesforce Marketing Cloud best practices start with documentation and deployment discipline. That discipline is what separates a Marketing Cloud implementation that scales from one that turns into tribal knowledge and workarounds. Defining, documenting, and distributing plans ensures all parties are aligned on when, where, and how customer engagement happens. That was true when we first wrote this guide, and it's just as true now, arguably more so, because the platform itself has gotten more complex and AI has raised the cost of getting the basics wrong.
A quick note before we dive in: if you've looked at Salesforce's marketing portfolio recently, you've probably noticed the naming has changed, more than once. Here's the state of play as of mid-2026:
Salesforce describes the move from Engagement to Next as a "convergence," not a forced migration. Existing automations, segments, and Data Cloud work carry forward rather than requiring a rebuild. If you're currently on Engagement, you don't need to rush into a migration, but you do need a point of view on which platform you're building toward, because it changes some of the guidance below, particularly around sandboxes and deployment tooling.
So what does the platform actually do? At its core, Marketing Cloud is a marketing automation and engagement platform that enables personalized, cross-channel customer interactions across email, SMS, WhatsApp, mobile push, web, and increasingly two-way conversational messaging, all grounded in a unified customer profile. That unification is the biggest architectural shift since this guide was first written: Data Cloud (now often referred to as Data 360) has become the connective tissue that Journey Builder, Einstein, and Agentforce all sit on top of. Whether you're on Engagement or Next, your data foundation now matters more than it ever did.
AI is unavoidable in any Marketing Cloud conversation right now, and it should show up in your documentation and strategy. Salesforce has built real capability here: Einstein provides predictive features like Send Time Optimization, Engagement Scoring, Engagement Frequency, and Metrics Guard (bot filtering), while Agentforce adds generative and agentic capability, drafting campaign briefs, generating multichannel campaigns from a prompt, and increasingly running two-way conversational engagement.
That said, most organizations we work with aren't struggling because they lack access to AI. They're struggling because the fundamentals in this article (clean taxonomy, a documented data model, real governance, a working test environment) aren't in place yet. Agentforce is only as good as the data and guardrails it's grounded in: a campaign creation agent pointed at ungoverned data extensions and undocumented naming conventions will happily produce fast, confident, wrong output. We'd rather see organizations treat AI adoption as a crawl, walk, run progression layered on top of solid fundamentals than as a "run" strategy that tries to skip the fundamentals to get to agents faster. Concretely:
Where you land on that spectrum should be a deliberate decision documented in your marketing strategy, not something that happens by default because a feature was turned on in a release.
Below is an updated look at the documentation and process areas that matter most when implementing or optimizing your Marketing Cloud instance, whichever platform you're on.
Define the goals: brand awareness, lead generation, conversion, retention, and the channels and messaging that support them. Document decisions and share them with stakeholders, written from the perspective of what a new team member would need to understand the goals, process, and structure. This is also the place to document your AI stance: which use cases are approved for generative or agentic assistance, what level of human review is required, and what brand voice or compliance guardrails any agent must be grounded in. Documentation should include:
Create a detailed plan of all tasks: data migration, platform configuration, and customization. If you're newly implementing, decide deliberately between Marketing Cloud Engagement and Marketing Cloud Next (Growth or Advanced) based on your scale, existing Salesforce footprint, and appetite for the newer, still-maturing platform. If you're an existing Engagement customer, document your convergence plan and timeline rather than treating the decision as one and done.
Define the structure and relationships between data types (leads, customers, campaigns, email, etc.) stored in Marketing Cloud, including entities, attributes, and connections. This is also where Data Cloud/Data 360 modeling now belongs: Data Model Objects (DMOs), identity resolution rules, and the unified profile that Einstein and Agentforce both depend on. Define the data extensions, custom components, and automations used to create touchpoints, and create a complete inventory of all data being migrated with a map of where each data point resides. Treat this documentation as a prerequisite for AI adoption, not a parallel workstream. Ungoverned or duplicate identity data is the single most common reason agentic features underperform.
Create detailed documentation of settings for email sending, data extensions, property definitions, user roles, and user permissions. On Marketing Cloud Next, also document Business Unit setup and how Einstein features and Agentforce agents are scoped per unit.
Detail integration with other systems in the marketing automation process, such as analytics, CMS, and CRM, whether Salesforce-native or third-party. Document how Agentforce agents are governed within these integrations: which actions they're permitted to take, which systems they can write to, and how access controls and audit logging are configured. Salesforce's Einstein Trust Layer and agent permission sets are the relevant controls here, and they deserve the same documentation rigor as any other integration.
If the strategy includes custom email templates, API integrations, or custom Agentforce agent topics and actions built on Flow, document the specifications in detail, including any custom Flows powering agent actions like "Draft Campaign Brief" or "Generate Campaign from Brief."
Define templates and create content for communication across email, mobile, social, and web. If you're using Agentforce's Content Builder or Campaign Creation agents to draft copy, document the brand grounding inputs (tone, voice, approved messaging, CMS brand references) that keep generated content on brand, and be explicit about what still requires human review before it ships.
This is the area that has changed the most since we first wrote this guide. Historically, Marketing Cloud had no real sandbox concept, and organizations either duplicated data extensions, journeys, and emails within production (fast, but risky and manual) or purchased a second production instance to use as a test environment.
That's still the reality for Marketing Cloud Engagement: there is no native sandbox, and the guidance from a few years ago largely still holds. A separate Business Unit or a second purchased instance remains the standard workaround, with the same considerations:
Marketing Cloud Next, however, introduced native Sandbox environments for both the Growth and Advanced editions, a meaningful gap closed compared to Engagement. If you're on Next, you should be provisioning and using a true sandbox rather than reaching for the workarounds above. Note that sandboxes require their own dedicated sending domains and don't share domain configuration with production. If you're evaluating a move to Next, native sandbox support is one of the strongest arguments in favor of making the jump sooner rather than later.
Build user manuals and quick-start guides for each user group to drive adoption. In 2026, this should explicitly include training on how to work with Agentforce agents: writing effective campaign briefs (brief quality directly determines output quality), reviewing AI-generated content and campaign previews before approval, and knowing when to escalate to a human-only workflow.
Document the testing approach and expected outcomes for campaigns, journeys, and integrations. Extend this to AI-generated output: define what "good" looks like for an Agentforce-drafted campaign brief or email, who reviews it, and what checks catch inaccurate personalization, off-brand tone, or hallucinated content before anything reaches a customer.
When you have separate development/testing and production environments, you need a process for promoting components. Marketing Cloud Engagement still relies on the Package Manager (under the Platform tab) for this, and the mechanics we described previously are largely unchanged: there's no direct deployment between instances. In the source environment, Package Manager creates a package of the components to deploy (Journeys, Automations, Assets, Data Extensions, and Attribute Groups), which is exported as a zip file and then imported into the target instance via Package Manager, with a summary of results shown on completion. A Deployment Manager tool has also become available for Engagement customers and is worth evaluating alongside Package Manager, though it doesn't yet cover every component type either.
Pre-deployment considerations (largely unchanged):
Post-deployment considerations:
If you're on Marketing Cloud Next, deployment between your new sandbox and production uses a different toolchain than legacy Package Manager, reflecting the platform's native Salesforce metadata model. Document it separately rather than assuming Engagement-era deployment guidance carries over directly.
Beyond traditional Split (A/B) Testing and Analytics Builder reports, including open rates, conversion rates, NPS/CSAT, and social engagement, Einstein now adds a predictive layer worth building into your standard reporting: Engagement Scoring, Send Time Optimization, Engagement Frequency, and Metrics Guard (which filters bot-driven opens and clicks out of your metrics, an increasingly important correction as automated inbox tools proliferate). On Marketing Cloud Next, Agentforce can also generate campaign performance summaries and optimization recommendations conversationally, a useful addition to, not a replacement for, your standard analytics review cadence.
Finding the right partner to meet your organization's needs is essential to realizing success faster and with confidence. The experts at Kenway Consulting act as an extension of your team, setting you up for success and filling in the gaps where you need it most. See how we helped a client streamline marketing operations through Cloud optimization. Our team of Salesforce consultants stays current on both the Marketing Cloud Engagement and Marketing Cloud Next ecosystems and knows what it takes to achieve success on either.
Here's a brief overview of the key elements our Marketing Cloud consultants take into account when starting a project today:
Monitor implementation for performance, adoption, and optimization opportunities, revisiting your AI crawl, walk, run plan as governance and confidence mature.
Ready to learn more? Connect with us to learn how Kenway Consulting can help you successfully implement Salesforce Marketing Cloud into your company.