When we sandboxed our first DAM MCP server at Stacks, it took about five minutes to connect. Getting an organization to actually use it takes a lot longer.
That was the heart of the session I gave at Acquia's DAM Workshop in Washington, DC. The room had already heard what MCP makes possible. I wanted to talk about what happens next, because that's where most AI rollouts stall. Ours did the first time around, and we learned a lot from it.
Someone said something at a conference recently that stuck with me: people don't resist change as much as they resist uncertainty. That one line is the whole case for change management. A good plan doesn't just get people to change. It gets the change to stick.
Why this moment matters for DAM teams
After almost 20 years in DAM, I can say this with a straight face: the technology is finally the easy part. DAM platforms are releasing MCP servers that let people reach their content from the AI tools they already use, like Copilot or Claude.
That means natural language search, workflows connected across systems, and content that comes to people where they work. No matter how good a DAM's interface is, it's still one more interface. Now it doesn't have to be.
But excitement isn't a rollout plan. To get going, your organization needs three things:
- Clear use cases, so people know what they can do and what to expect.
- Confidence in the content, so people trust that what comes back is accurate and cleared for use.
- A plan to bring people along, so the change lands instead of fizzling.
The first two are familiar territory for DAM practitioners. We can come up with 500 ideas for how people might use their content before lunch. The third one comes less naturally, which is exactly why we focused on it.
The four parts of a change plan
I'll be honest. The first time a colleague (who is certified in change management) handed me one of these plans, my reaction was, "You've got to be kidding me. This is so hardcore." By message three, I couldn't live without it. Now we use it on every project.
At a high level, the plan breaks the work into four parts.
1. Audiences
Identify who is affected and what they'll need. Go past department labels like "marketing" or "the executive sponsor" and get to your actual distribution list. Who is on those emails, and what do they want to know?
Those questions vary a lot. IT may ask whether the AI will respect rights restrictions. Another group may ask whether you're trying to make their job redundant. Lay out every reaction you can anticipate before you start.
2. Messaging
Decide what you'll say to each audience, when, and who will say it. It can't all come from you. You'll probably draft most of the emails, but some messages only land when they come from someone's direct manager. Others only land when they come from the photographer in the field.
3. Milestones
Map the moments worth communicating. Some are wins, like your first use cases coming back clean. Some aren't, like the first time the AI gets an answer wrong. That's a milestone too, and it deserves its own message so people know where things stand.
4. Feedback loops
This is where you tackle uncertainty head-on. Set up feedback channels, tell people about them, actually check them, and respond. Doing that relieves a surprising amount of stress for the teams you're asking to change.
What it looks like with a real client
We're using this plan right now with a client standing up a new DAM. They have about 3,500 employees spread across the country. A small marketing team runs the platform, but photography comes in from people in the field and in regional offices, and the regional offices often know more about the contracts than the field teams do.
Access to an MCP server was a deciding factor for them. Like a lot of organizations, they've standardized on one AI layer (in their case, Copilot), and IT doesn't want another agent living inside every application. They also work under confidentiality rules at both the contractual and federal level, so governing all of it consistently matters.
Here's the lucky part: demonstrating the MCP server showed that their governance questions aren't a DAM problem. They're organization-wide.
If you've ever been the DAM manager who ends up as "keeper of all the rules," you'll appreciate this. When someone in Copilot asks for photos from a highway project over the last five years, the DAM's results should match what Microsoft Purview returns. If Purview surfaces projects the DAM won't, that's a governance conflict higher up in the organization. MCP gives you a not-so-secret way to put a spotlight on it and get it resolved.
Our plan for this client covers the audiences you'd expect, each with its own messaging:
- Executive sponsor: help them understand what an MCP server is. It's new for most organizations. Anyone who tells you they're an expert is either lying or should become your new best friend.
- Legal and risk: often forgotten until something goes wrong, and fully able to stop a project in its tracks. Bring them in early.
- Information security and IT: tell them what you're trying to do and what it will take from them. Often the answer is good news: it's a light lift.
- Early adopters: the people you most want using the tool. If they stop engaging, the rollout dies a slow, silent death, so give them reasons to keep coming back.
- Everyday DAM users: the fun announcement. You get to tell them there's an even easier way to get to content.
- Agencies, partners, and license holders: anyone outside the walls who needs to know how content is being accessed.
Where to start
Start with what you know. Filling out the plan will surface your known unknowns, and finding those gaps is part of the point.
- Pick three to five use cases tied to business outcomes. Write down what "right" looks like for each one. When someone runs that query through the MCP server, you can check the answer and fix what's under the hood if it's off.
- Name real people, not departments. Put your actual distribution lists in the plan so there's no question who gets which message.
- Map the workflows that will change. Today, when someone asks marketing for project photos, a person asks follow-up questions first. Is it for a presentation? Does the timeframe matter? You can now build skills that make the AI ask those same questions before it returns results.
- Test hard cases and easy ones. Nothing shuts a project down like a copyright infringement. Include the scenarios you think will never fly, like talent with exclusive contracts, alongside a few quick wins. The wins build trust. The hard cases prove the rules hold.
- Set milestones for the messy moments too. "Nobody has used it in a while" is a milestone. It calls for a check-in message, not silence.
- Make feedback impossible to miss. Put a feedback form link on your DAM's home page and in every email you send. Ask people to submit even if they think someone already has. At your quarterly or annual governance review, "10 people asked for this" tells you what to fix first.
- Give training a home. A SharePoint site, a Teams channel, or a skill that points people to resources, so nobody has to ask, "Where do I catch up on this again?"
- Define success, then commit. Decide when you'll say, "From now on, if you need photography like this, use this channel."
Get the worksheet and the prompt
We didn't want this to be one more session you nod along to and then forget. So we're sharing the same tools we use with clients:
- The change management worksheet. The Excel workbook we use to map business objectives and use cases, audiences and stakeholders, messaging, and the rollout timeline and milestones.
- A prompt that helps you fill it out. Run it in your AI tool, and it interviews you about your stakeholders, audiences, goals, and milestones, then helps you review and refine the plan.
The prompt is built around launching an MCP server, but it adapts easily to any DAM change, like rolling out a new search interface. One thing we baked in that you should keep: "Never invent a name or a job title. If I didn't tell you, put it in square brackets." AI likes to tell you what you want to hear. Your change plan needs the truth.
Download the worksheet and prompt
You don't have to do it alone
Change management is a big lift. Any change in DAM is a big lift right now with the technology moving this fast. It's also something we lean into on every Stacks project, because it's the difference between a platform people have and a program people use.
The worksheet will get you started. When you want a hand with the rest, from planning the rollout to training your teams, let's talk.
Where DAM Gets Done.
