The Human Coefficient: How anthropology can solve the MarTech stack
The real MarTech challenge is not what your stack can do. It’s bridging the gap between the operating model you designed and how your people actually get the work done.
Most MarTech, DAM, workflow, and creative automation implementations are built for the organisation on paper: the process maps, governance decks, workflows, and platform logic. But they land inside the organisation as lived: informal approvals, shadow libraries, local workarounds, creative identity, stakeholder politics, and deadline pressure. Anthropology can help read and understand those human variables to move from surface-level adoption to real absorption.
Unfortunately, most of us have endured a disappointing transformation more than once.
A MarTech, DAM, workflow, or creative automation platform is selected through rigorous evaluation. Configuration completes on time. Assets are migrated. Workflows are mapped. Training is delivered. The dashboard turns green: logins, completions, ticket counts, template usage.
By the measures of the project plan, the implementation is working… but then twelve months pass.
Marketers are still briefing outside the system. Approvers are still staying on email. Brand teams are maintaining their own version of the truth. Local markets are keeping shadow asset libraries. Creative teams are quietly avoiding the automation tools. By any technical standard, the implementation succeeded. But by any operational standard, it did not.
Typically, we blame change management, training, leadership for not doing enough to champion the change. The usual suspects. And sometimes that is true. It’s rarely the whole story, though. Because it treats the human side of implementation as a residual solve. You know, the odds and ends to be cleaned up after the real work has been completed. Which is a mistake.
The platform didn’t fail the platform. The failure happened in the gap between the operating model the organisation designed, and the workflow behaviours people actually lived by. The cultural, psychological, and social reality that determines whether technology becomes part of the work, or simply interrupts it.
This is the human coefficient.
One organisation. Two operations.
There are two marketing operations in every business. The designed workflow and the lived workflow.
The designed workflow lives in the implementation deck. The one the platform is built for.
The lived workflow subsists somewhere else. It runs on shortcuts, trusted people, agency-side fixes and undocumented rules. Briefs get written after the conversation that defined the work. Approvals get negotiated in Slack, Teams and WhatsApp, then logged retrospectively in the workflow tool. Senior stakeholders override governance in ways nobody contests but everyone remembers. Local markets keep unofficial libraries because the central one is missing context they need. Production tracking lives partly in the workflow tool and partly in a shared spreadsheet that one person updates and everyone trusts.
Both workflows are real. But MarTech is designed for the first and lands in the second. It’s not a technology deployment problem. It’s a human implementation problem. While the first can be specified, the second can only be read.
Implementation begins to fail not because formal operating models are useless, but because they are incomplete. They describe how the organisation wants work to move. They do not necessarily explain how work really moves. Work that was already in motion, but now integrates the stack in survival mode.
What specifically lives in the gap?
None of it is random. It always shows up in the same places. If you know where to look.
Trust is the first place.
Trust in the asset library. Trust in the metadata. Trust that an approval is final. Trust that creative automation will not damage the brand. Trust that the DAM is better than the folder people already use.
Trust builds slowly through repeated proof that the system behaves as expected. It collapses quickly through visible failure. One bad template experience can become a team-wide story. One missing rights field can send everyone back to email. One failed search can confirm the suspicion that the DAM is just an archive with a login.
Identity is the second.
Creatives, brand teams, and producers do not see themselves only as executors of a brief. They see themselves as the people responsible for whether the brand is well represented.
A platform that asks them to accept auto-generated copy, approve template variations they did not design, or trust an AI-generated image with a brand-coded subject is asking them to take on professional risk on behalf of the system. Whether they will do that depends less on what the platform can technically do, and more on whether the operating model around it makes that risk acceptable.
Authority is the third.
The lived approval process in many marketing organisations is faster, looser, and more political than the documented one. Senior approvers are consulted out of band before formal approval is requested, so formal approval becomes confirmation rather than decision. Brand interpretation is negotiated in conversations the workflow tool never captures. Local-market judgement often sits outside central governance, even when central governance claims otherwise.
This is why workarounds matter. They are human signals telling you the official system does not reflect the production reality. Read them that way.
To read the gap you need anthropological thinking
Social and cultural systems aren’t unreadable. Anthropology has spent more than a century studying exactly the kind of material in the gap between the formal organisation and the lived one. And has been part of how technology gets designed for nearly forty years.
Lucy Suchman’s ethnographic work at Xerox PARC established the foundational point. People don’t follow plans. They orient to plans while acting in response to a continuously shifting situation. Plans are resources for action, not specifications of it. A workflow tool that treats its own diagram as a specification rather than a resource is asking the human to behave like a machine. People do not. The friction shows up as workarounds, shadow systems and selective compliance.
A second strand of practice-based research extended the point. Technology isn’t a stable object that gets installed and then used. It’s recurrently enacted through the practices people develop around it. A DAM is not the platform on the server. It is the platform on the server plus the tagging behaviour of marketers, the trust behaviour of designers, the search behaviour of producers, the override behaviour of brand managers, and the workaround behaviour of agencies. These practices reproduce or modify the platform every time someone interacts with it. The implementation is not finished at go-live. In this sense, it has barely started.
On top of this, marketing itself has used anthropological methods for years to understand customers. We observe behaviour, study context, decode habits, look for meaning, and try to understand why people do what they do rather than relying only on what they say they do.
We study consumers before building campaigns. Yet we rarely apply the same discipline to understanding ourselves as consumers of this technology we’re implementing. For whatever reason, we don’t watch what we’re up to with the same intensity or enthusiasm.
Fully knowing ourselves and our processes tells us whether the operating model we are designing has any chance of surviving the organisation we’re deploying it in.
Change management remains necessary, of course. Communication, training, champions, leadership alignment, reinforcement, and support all matter. But too often, change management arrives after the future state has already been designed. Which is why anthropological wisdom needs to move upstream.
Starting with discovery and requirements, then workflow and operating-model design, then governance and taxonomy and everything else. Into how success gets measured. Because humans will be humans.
The paradigm needs to shift from adoption to absorption
Adoption is the wrong metric once you start reading the lived reality within the organisation.
Adoption is access, logins, training completion, feature usage, ticket counts, template usage, asset uploads, and workflow volume. These measures are useful. But they are weak on their own. They tell you whether the platform is being touched. They do not tell you whether it has changed how work happens.
A dashboard can be green while behaviour stays exactly where it was. Logins can rise while trust stays low. Workflow volume can increase while senior approvals still happen offline. Metadata completion can improve while search relevance stays poor. Template usage can rise while creative teams disengage. Assets can be uploaded while shadow libraries keep growing.
Thinking of it as absorption makes a lot more sense.
Measuring the absorption of technology into how the organisation works is more practical and revealing. Absorbed technology is not an additional layer the everyone has to work around. For example, a DAM can be considered absorbed when teams trust the library and stop creating shadow repositories. Then search behaviour changes and reuse becomes the default. Metadata is inherently maintained because people understand its value. Rights are respected because the system makes them visible and credible. A workflow tool can be considered absorbed when stakeholders stop bypassing steps and approvals. Then briefs improve because the structure of the request forces better thinking. Exceptions are visible because the system makes hiding them difficult.
You need to think of successes as series of behavioural tests rather than platform ones.
Understanding people is anthropological work. Understanding technology and process is operating-model work. And both need to be resolved to deliver an effective ecosystem, where the stack is absorbed into the lived experience on the production line.
Designing for absorption changes both design and implementation
Moving from adoption to absorption is about reconfiguring the question asked.
From: “What is the process supposed to be?”
To: “What does work actually look like as a campaign moves through the organisation?”
Workflow design must start from observed work, not an academic unpacking of process. Approvals must stem from authority structures that already exist, not a textbook. That means surfacing them, interrogating them, and looking at working with them rather than wholesale replacement. Taxonomy and metadata must be designed around how people search and tag, not only how an information architect would prefer the library to behave.
Governance must become an evolving structure. Training must become continuous learning. Absorption is always ongoing.
This is what it means to solve the stack anthropologically. Not by making technology more human in a vague sense. By designing technology-enabled operations around the human system that already exists, then shaping that system deliberately towards a better one.
AI has become the clearest way to see whether a marketing organisation has truly absorbed its technology. Not because it created the problem, but because it exposes it when the lived organisation is invisible. BCG reports on AI in marketing and commercial functions (2024–2025) typically find that AI in marketing requires operating-model reinvention, before productivity gains are realised.
Organisations handling AI well are not the ones with the most advanced models. They are the ones whose operating model could already absorb new capability - whose DAM was actually used, whose workflow was actually trusted, whose creative teams had a clear view of where judgement was theirs to exercise and where it wasn’t.
The choice is in front of us
And, for MarTech, CreativeOps and Marketing Operations leaders, it’s not the one the platform conversation usually suggests.
Before investing in the next tool, the next AI agent, the next workflow upgrade, or the next automation layer, we need to start relooking the gap with the eyes of the anthropologist. See the people. Because adding another capability on top of a capability that has not been fully absorbed only introduces more complication. This is the lived reality.
We must start diagnosing the production ecosystem we actually have, before we invest in the ones we wish we had.
Move from adoption to absorption with ManMachine
We operate in the gap between operating-model design and lived behaviour: the place where workflows, platforms, partners, and people either become a functioning system or drift into workaround culture.
Across People, Process, Platforms, and Partners, we help organisations understand how work actually moves, where the formal model breaks, and what has to change before MarTech, DAM, workflow, automation, or AI can be properly absorbed.
We do not start with the promise of the stack. We start with the conditions that make the stack usable. Because implementation does not end when the platform goes live. It starts when people change how the work gets done.
Thanks for reading Interface!
Subscribe for free below to receive future releases.