We Have Automation. But Which Level Are You Actually At?

A marketing team has had Dynamics 365 Customer Insights for somewhere between six months and two years. They’ve built journeys. They’re sending emails. And when you ask how automation is going, someone says: “It’s going well — we have automation.”

And then you dig a little deeper.

Lists are still being pulled manually. The same welcome journey has been running untouched since go-live. Segments get rebuilt from scratch every campaign because nobody documented the logic the first time. Consent data lives in at least two places and neither of them fully matches.

They have automation. They’re just not getting anywhere close to what Customer Insights is actually capable of delivering.

That gap between “having the tool” and “using it well” is what this post is about. I’ve mapped it into five levels, each one grounded in how Customer Insights actually works, and each one with a concrete move to get you to the next stage.

Read through and ask yourself honestly: where do we sit?

A Quick Word on What Customer Insights Is Built to Do

Before we get into the levels, it helps to understand the architecture you’re working with.

Dynamics 365 Customer Insights is two applications that are designed to work together. Customer Insights – Data (CI-D) is your customer data platform: it ingests data from multiple sources, runs deduplication and identity resolution, and produces unified customer profiles. Those profiles can then be enriched, segmented, and measured. Customer Insights – Journeys (CI-J) is your orchestration engine: it takes signals from the real world (form submissions, event registrations, purchases, behavioural triggers) and uses them to move customers through multi-step, multi-channel journeys.

The real power of the platform lives at the intersection of these two. By connecting both Customer Insights applications, you can target unified customer profiles and segments, engage every customer regardless of where their data lives, and base dynamic content in emails, SMS, and push notifications on measures such as loyalty status, subscription renewal date, or any other measure captured in the unified profile.

Most teams are using one half of this equation. That’s why they’re stuck.

With that context, here are the five levels.

Level 1: Broadcast Mode

You run campaigns like tasks: build a list, send a message, move on.

This is where most teams start, and where too many stay. At Level 1, the tool is essentially being used as a more powerful Mailchimp. Campaigns are one-off events, not repeatable systems.

You’re probably here if:

• Lists come from Excel exports or static views someone built in Dataverse
• Every send requires a human to kick it off manually
• The same message goes to very different people because segmentation is too broad
• Your reporting is mostly open rates and gut feeling

There’s nothing shameful about being here, especially in the first months post-implementation. The problem is when it becomes the default operating mode eighteen months in.

What to do next: Pick one real business moment, a demo request, a brochure download, an event registration, and convert it from a one-time send into a trigger-based journey. Trigger-based journeys react to customer action, such as filling out a form or registering for an event, meaning once you build it once, it runs automatically every time that action happens. That’s your first genuine piece of automation.

One practical tip: Use a single reusable marketing form with a hidden field that captures the campaign source from the URL. This lets you route people into different journey branches without duplicating forms every time a new campaign goes live.

Level 2: Starter Automation

Some journeys run automatically — but they’re still one size fits many.

At Level 2 you’ve got journeys. You probably have a welcome flow, maybe a post-event follow-up. But the journeys are mostly email-only, the segments get rebuilt from scratch each campaign, and if you looked across all your sent emails from the last six months, they’d look inconsistent, different fonts, different tones, different levels of personalisation depending on who built what.

You’re probably here if:

• You have a welcome journey and one or two other automations
• Every new journey starts from a blank canvas
• Emails look different depending on which team member built them
• You’re rebuilding the same audience segments repeatedly

The main problem at this level isn’t a data problem, it’s a governance problem. The tools are capable of far more than you’re asking them to do, but there’s no shared system for building things in a repeatable way.

What to do next: Standardise before you scale. Create a small set of reusable email templates (three is enough to start), document your core segments so they don’t have to be recreated, and agree on a naming convention for everything. A pattern like Purpose + Audience + Trigger + Version (e.g. Welcome_NewContacts_FormSubmit_v2) sounds small but it’s the difference between a manageable environment and one that nobody wants to touch.

One practical tip: Journey templates are underused by almost everyone. You can create journey templates by building a journey, saving it, and then selecting “Save as template” which means the next person starting a similar journey doesn’t start from zero. Use them.

Level 3: The Frankenstack Problem

You have the tools. But the data underneath them is a mess — so results feel random even when you did everything right.

This is the most frustrating level to be at, because you’ve put in the work and it still doesn’t quite function. Journeys feel unpredictable. Segments return different numbers depending on where you build them. Someone discovers that the same contact exists in three different forms in Dataverse and nobody is sure which one is “real.”

The culprit is almost always the customer record itself.

You’re probably here if:

• You have duplicate contacts with no clear winner
• Consent data exists in multiple places and doesn’t always agree
• Journeys feel random because key data fields arrive late or are inconsistent
• Reporting requires someone to manually stitch together multiple exports

This is where CI-D or a Customer Data Management becomes critical. Data unification combines your various customer data sources and creates a single customer profile record per customer, eliminating duplicate data and combining all the important fields from your various data sources into a single record, eliminating data silos. Until that process is set up and trusted, everything built on top of it, segments, journeys, personalisation, is built on an unstable foundation.

What to do next: Don’t try to solve everything at once. Define one authoritative source of truth for consent and preferences, document the rule in writing, and then run a profile unification pilot on a single, well-defined use case. One use case, done properly, teaches you more than trying to unify everything at once.

One practical tip: Keep your segmentation logic inside CI-D as measures and segments not buried in journey branch conditions. If you find yourself rebuilding the same attribute check in multiple journey branches, that’s a signal: turn it into a proper segment or calculated measure that can be reused.

Level 4: Orchestrated Experiences

Your foundation is solid enough to trust. Journeys run, repeat, and don’t break — across more than just email.

Level 4 is when things start to feel like they’re actually working. You can run a journey, stop it, update it, republish it, and trust that it will behave the way you expect. Segments come from CI-D and don’t need to be rebuilt. You’re monitoring journey entry rates and drop-off points, not just whether the email delivered.

Critically, you’ve moved beyond email-only. SMS, push notifications, event triggers, sales handoffs, at least one additional channel is part of your journey logic.

You’re probably here if:

• Segments are built once in CI-D and reused across journeys
• You have suppression rules and frequency caps in place
• You’re asking “why did people drop off at step 3?” rather than just “what was the open rate?”
• Journeys include at least one channel beyond email

The next unlock at this level is measurement. Not just journey analytics, but connecting your automation activity to actual business outcomes.

What to do next: Define two or three measures in CI-D that map to something your business actually cares about, not vanity metrics. You can build measures to gauge your business goals and KPIs, and once those measures exist in CI-D, you can use them directly inside CI-J: as segment filters, as personalisation tokens in email content, or as branching conditions in journey logic.

One practical tip: Keep your measures simple and clearly named. A measure called Days_Since_Last_Purchase is immediately usable by anyone building a journey. A measure called Calc_CustomerBehaviourIndex_v3_FINAL is not.

Level 5: Predictive + Omnichannel

The system decides the next move — not you.

At Level 5, your unified profiles and measures are good enough that journeys can adapt based on customer behaviour in real time. A customer who opens an email but doesn’t click gets a different next step than one who clicks but doesn’t convert. A customer whose churn risk score crosses a threshold automatically enters a retention flow. Sales gets a task created when a contact reaches a certain lead score.

Personalisation at this level extends well beyond email, to web content, events, advertising audiences, and sales handoff moments.

You’re probably here if:

• Unified profiles and behavioural signals drive next-best-action decisions automatically
• Measures and segments from CI-D influence entry, timing, and branching in live journeys
• Personalisation is consistent across email, SMS, events, web, and sales
• Governance exists: clear naming, documented ownership, a safe process for publishing changes

Using behavioural data from Customer Insights – Journeys to create segments, measures, and predictions in Customer Insights – Data increases targeting precision and personalisation throughout the buyer journey. That loop, journey behaviour feeding back into the data layer, which then improves future journeys, is what Level 5 actually looks like in practice.

What to do next: Document three core customer paths, new customer, returning customer, churn risk, and audit whether your data model and your journey logic are genuinely aligned around them. Most teams at Level 4 discover they have journeys that assume data quality they don’t yet have. Level 5 is about closing that gap completely.

One practical tip: Governance matters more at this level than at any other. Before adding new journeys or measures, make sure you have clear ownership: who can publish, who reviews, what the naming standard is. A sophisticated automation environment that nobody fully understands is a liability, not an asset.

So — Which Level Are You At?

Here’s a quick way to check. Count how many of these are genuinely true for your team right now:

1. We use reusable segments from CI-D, not one-off lists
2. We trust our unified profile and identity resolution
3. Our journeys have suppression rules and frequency caps
4. KPIs are measurable without manual reporting
5. We run structured experiments — A/B tests, holdout groups, iterative improvements
6. Data and consent updates flow reliably end-to-end

0–1: You’re at Level 1. Start with one triggered journey. 2: Level 2. Focus on standardisation before you add more journeys. 3: Level 3. The data layer needs attention before anything else. 4: Level 4. Build your measurement layer, it’s what unlocks the next stage. 5–6: Level 5. Focus on governance and closing any remaining data quality gaps.

A Note on Getting from Here to There

The levels above aren’t just a maturity model for the sake of it — they’re the actual sequence in which things need to be solved. You cannot reliably personalise at scale (Levels 4–5) if your customer records are inconsistent (Level 3). You cannot build reusable, scalable journeys (Level 2–3) if you’re still treating every campaign as a one-off task (Level 1).

The good news is that you don’t need to overhaul everything at once. The next level is always just one focused improvement away, a triggered journey, a naming convention, a profile unification pilot, a measure that maps to a real business outcome.

If you’ve read this post and you’re not sure where you actually sit or you know where you are but you’re not sure what the right next move looks like for your specific setup, feel free to reach out. I’m happy to take a look.

Do you have questions, ideas or remarks? Feel free to get in touch.