Before You Optimize Salesforce Marketing Cloud, You Need an Audit

Most teams inherit a broken Marketing Cloud instance and immediately start building new journeys on top of it. Here's why that's the wrong move — and what a proper audit actually looks like.

The Instinct That Makes It Worse

When a client hands you access to their Salesforce Marketing Cloud instance and says "fix it," the instinct is to open Journey Builder and start improving. New journeys, better segmentation, behavioral triggers instead of static sends.

That instinct is almost always wrong.

Building new journeys on top of a degraded Marketing Cloud environment doesn't fix the problems underneath — it layers complexity on top of them. Campaigns fire on stale data extensions. New journeys overlap with old ones that nobody turned off. Email volume spikes because frequency caps were never configured across the full journey landscape. Deliverability suffers. Open rates fall. The customer experience gets noisier, not better.

The correct starting point is an audit. Not a light review. A systematic examination of what is actually running, why it's running that way, what's broken, and what the data underneath it looks like. Everything else follows from that.

What We Found Before Touching a Single Journey

During the Marketing Cloud discovery workshop with a global sports and lifestyle brand, the pre-audit picture was stark. The instance was active — campaigns were running, emails were going out, journeys were live. From a system health perspective it looked operational.

It wasn't.

Here's what the current-state assessment revealed across five dimensions:

1. Manual Campaign Workflows Were the Integration Layer

Campaign setup and updates were entirely manual. The workflow was: Excel sheets → manual data uploads → Marketing Cloud journey updates. There was no automation connecting source systems to SFMC. Every campaign required human intervention to execute.

The consequence: campaign turnaround times were measured in days, not hours. Translations and asset updates alone created one-week delays per email localization cycle. The marketing team was spending the majority of its capacity on execution — manually fulfilling requests — rather than strategy or optimization.

Manual workflows are not just slow. They are inherently unreliable. Manual uploads miss records. Manual updates to journey logic introduce errors. And they create a dependency between the quality of the marketing output and the availability of the people doing the manual work — which means any team capacity issue directly becomes a campaign failure.

2. Automation Decay Had Spread Across the Journey Landscape

The instance had approximately 35 active B2C journey categories: new subscriber onboarding, club membership stages, post-certification nurture, nudge series during courses, abandoned cart, travel post-booking, birthday journeys, and seasonal certification campaigns.

Most were not functioning at full capacity. The audit revealed a pattern called automation decay — journeys that had been built correctly at one point and then never updated as the underlying data, the product catalog, or the customer lifecycle model changed. Post-certification journeys were recommending courses that no longer existed. Club member journeys were firing on segment definitions that hadn't been refreshed in over a year. Travel journeys were using destination data that predated a product restructuring.

Automation decay is invisible from the outside. The journeys are "live" — they appear active in the interface, emails are going out, no errors are surfacing. But they're operating on stale logic and stale data, which means every send is either irrelevant, inaccurate, or both. The customer experience degrades silently.

3. Over-Communication Was Measurable

The most concrete signal of what happens when journey orchestration breaks down: one user had received 27 emails in 60 days.

This wasn't the result of a single misconfigured journey. It was the cumulative effect of running multiple journeys against multiple partially-overlapping segment definitions, with no cross-journey frequency cap and no audience exclusion logic. Every journey team thought they were targeting a defined segment. None of them knew that their segment overlapped significantly with seven other journeys also running against that population.

27 emails in 60 days is not an edge case. It's a measurement of what happens at scale when Marketing Cloud is treated as a collection of independent campaign tools rather than a unified orchestration layer. The same customer — someone who should be receiving a coherent, personalized experience — is receiving noise. And noise drives unsubscribes, deliverability degradation, and brand fatigue in ways that take months to recover from.

4. Segmentation Was Static and Manual

Segmentation relied on manually maintained data extensions. The process: export from the CRM or data warehouse, clean, upload to SFMC, build the journey against that upload. This produced segments that were accurate at the time of the upload and increasingly inaccurate every day afterward.

A club member who had just upgraded their tier was still in the standard-tier segment for 30 days after the fact. A customer who had completed an advanced course was still receiving beginner nurture emails because their certification status hadn't been refreshed in the data extension. A lapsed member who had reactivated was still in the lapsed segment.

Static segments don't just cause irrelevant messaging. They create a specific kind of brand damage: the customer knows you know about them — they've interacted with you, they're in your system, they receive emails from you — but the emails reflect a version of them that is weeks or months out of date. That disconnect is more damaging than not messaging them at all.

5. Content and Localization Were Bottlenecks

The instance supported over 90 renewal segment variations across 13 languages. This was being managed manually — translation, creative versioning, approval workflows, scheduling — by a small team.

The consequence was twofold. First, content updates were slow: changes to a single email could take a week to propagate across all language variants. Second, consistency was poor: different language versions of the same campaign were often at different versions of the content, meaning customers in different regions were receiving materially different communications.

This is not a staffing problem. 90 segments across 13 languages cannot be managed reliably by any team using manual processes. It's an automation architecture problem, and it requires an automation architecture solution.

The Audit Framework: Five Dimensions Before Any Journey Work

Based on this discovery, the audit framework we use for Marketing Cloud engagements examines five dimensions systematically before any optimization work begins.

Dimension 1: Journey Inventory and Health

Enumerate every live journey. For each: when was it last modified, what data source is it running against, what are the entry criteria, what is the exit criteria, and is there evidence it's producing the intended outcome. Flag journeys that haven't been updated in over six months — these are decay candidates. Flag journeys with no exit criteria — these are over-communication risks.

Dimension 2: Data Extension Quality and Currency

For each data extension used in a live journey: when was it last refreshed, how is it populated, and how much drift has occurred since the last refresh. A segment that was accurate 30 days ago and hasn't been updated since is a reliability risk, not a marketing asset. The audit maps which extensions are real-time (fed by automation), which are batch (fed by scheduled processes), and which are manual (fed by human uploads).

Dimension 3: Frequency and Overlap Analysis

Run a cross-journey analysis to identify customers who are enrolled in multiple journeys simultaneously. Calculate the maximum number of emails a single customer can receive per 30-day period across all active journeys with no frequency controls. If that number is in double digits, you have an over-communication problem. If it's above 10, you have a deliverability risk. If it's above 20, you have active brand damage occurring right now.

Dimension 4: Integration Coverage

Map every data source that should be feeding Marketing Cloud — the CRM, the LMS, the commerce platform, the membership system — and assess the integration status of each. Is the data flowing automatically? How frequently? What happens when a customer's status changes in the source system — how long before that change is reflected in SFMC segmentation? Any integration that requires manual intervention is a point of failure.

Dimension 5: Content Architecture and Localization Infrastructure

Inventory the template structure: how many base templates exist, how many variants per template, and how many language versions per variant. Assess whether the localization workflow is automated or manual. Identify which content elements are shared across variants and which are unique — this drives the efficiency opportunity. An instance with 90 segment variants across 13 languages that are all managed independently is a re-architecture problem, not an optimization problem.

What the Audit Changes About the Optimization Plan

Running a proper audit before touching any journeys produces a categorically different optimization plan than diving in without one.

Without an audit, the typical approach is to add: add better triggers, add more personalization, add new journey stages. The result is more complexity on top of the existing problems.

With an audit, the optimization plan starts with removal and remediation: decommission decayed journeys, fix the data extension refresh architecture, implement cross-journey frequency caps, rebuild the localization infrastructure. Only after that foundation is solid does adding complexity become safe.

For the client in this engagement, the recommended optimization sequence was:

First: Implement frequency governance — a global cap that limits total sends per customer per 30-day period, enforced at the journey level before any email fires. This alone would have prevented the 27-email situation and reduced deliverability risk immediately.

Second: Rebuild data extension refresh pipelines — replace manual uploads with event-driven writes from source systems. A customer whose certification status changes in the LMS should have that change reflected in SFMC within hours, not weeks.

Third: Rationalize and sunset decayed journeys — run each journey in the inventory against its original intent. Journeys that can't be validated against current product and business logic should be decommissioned, not patched.

Fourth: Only then — add the optimization layer. Behavioral triggers, predictive segments, CDP-powered personalization, cross-sell recommendations. These are genuinely powerful capabilities. But they require clean data, coherent orchestration, and functioning automation infrastructure to deliver value. Built on a broken foundation, they just make the noise louder.

The Principle: Optimization Multiplies What Already Exists

The core reason an audit has to precede optimization is architectural, not procedural.

Marketing Cloud is an amplification engine. When the underlying data and journey architecture is functioning correctly, optimization amplifies relevance and drives meaningful engagement. When the underlying architecture is broken — stale segments, decayed journeys, manual workflows, no frequency governance — optimization amplifies the brokenness. More sends on stale data is more noise. Better targeting logic on a corrupt data extension is more precisely wrong.

You cannot improve your way out of a structural problem. You can only fix it and then improve.

The teams that get the most value from Salesforce Marketing Cloud are not the ones who add the most journeys or use the most advanced features. They're the ones who maintain the discipline to audit before they optimize, and to clean before they build.

Read Next

Marketing Cloud optimization becomes genuinely powerful once the data layer underneath it — unified profiles, real-time behavioral signals, cross-system identity resolution — is in place. That's the CDP layer.

→ The Identity Problem Your CRM Cannot Solve (And Why a CDP Is the Only Answer)

Axelerant works with enterprises on Salesforce Marketing Cloud audit, optimization, and multi-cloud transformation. If your team is navigating a degraded or over-complicated SFMC instance, we'd be glad to talk through how we approach it.