The CEO asks the company’s data scientist if the marketing strategy should incorporate generative AI and large language models.
The data scientist responds:
“The epiphany of large language models, distinguished by their profound deep neural architecture and expansive parameterization, constitutes a quantum leap in the domain of marketing. Their extraordinary capacity to unveil intricately woven, nonlinear structures within vast and voluminous datasets begets unparalleled precision in the domain of hyper personalized marketing campaigns, transcending the realms of conventional methodologies …”
When the data scientist finally stops, the CEO asks, “Can you dumb it down for me?”
To which the data scientist replies:
“Oh dear, it seems like you’re having trouble grasping the basics. Let me break it down for you like you’re a little kid. Large learning models are super-duper smart tools that help us know what our customers like.
“They’re way better than the old stuff we used. With these fancy models, we can make ads that really speak to people and make them buy our stuff. And guess what? They do it super-fast, too! So, we gotta use these cool models to beat the other companies and be the best in marketing. Got it now? Good!”
That unhelpful interaction underscores the challenges arising when people have different levels of understanding of technology. The data scientist is deep in the trenches, and the CEO has a cursory knowledge of how generative AI works.
Putting together these people of varying knowledge can lead to frustration, or worse, butting heads.
Katie Robbert, CEO of Trust Insights, provided solutions to this challenge in her Marketing Analytics & Data Science (MADS) conference presentation, Managing the People That Manage the Machines.
How communications go wrong
Katie asked managers and technologists how interactions, like the one between the CEO and data scientist, go wrong. Interestingly, both roles identified the same issues with the other party, including:
- Being too bossy
- Assuming the other party isn’t smart enough
- Asking the wrong questions (managers) and poor communication (technologists)
- Overcompensating
- Bad attitude
- Blaming each other
As organizations get deeper into generative AI, machine learning, natural language processing, and deep neural networks, expect an increase in communication breakdowns between managers and technologists, Katie says.
“You need to have a framework for how to communicate with the people who are doing work, the people who are managing the machines,” she says.
To make that possible, Katie shares a framework to help managers and technologists work better together.
Use the 5P Framework to improve tech communication
Katie developed the 5P Framework to help people with different backgrounds and levels of expertise work better together. The five elements include:
- Purpose: What is the problem you’re trying to solve? What is the question you’re trying to answer? Why are you having this conversation?
- People: Who is involved in this conversation, and what other stakeholders need to know the result of it?
- Process: How are you doing this? When do you need to do it? What tools do you have to use?
- Platform: What tools are you using to converse? Slack, phone, in-person?
- Performance: Did you adequately address the problem?
Katie explains the first three — purpose, people, and process — involve expectations. Process and platform are about execution, and performance is about measurement.
Craft requests with user stories
The 5P Framework works well in crafting user stories. Taking a page from Agile software development teams, the story structure is simple: As a [persona], I [want to], so [that].
Katie says the framework maps perfectly with the user story format. The persona is the people. The “want to” is the process and platform. The “that” comes from purpose and performance.
Let’s consider an example.
Katie’s past employer developed products to help substance abuse clinicians do patient intake. The primary stakeholder was an academic who ran clinical research. He came to development meetings with an attitude that he knew what was best for the product. He asked for a laundry list of product features.
The developers felt micromanaged, while the stakeholder felt like he wasn’t getting what he needed. Poor communication, friction, anxiety, and frustration arose. “Anytime anyone saw him even approaching a meeting room, the anxiety would just swell, and people would start to shut down,” Katie says.
The stress from those conflicting perspectives could have been reduced if they created user stories, such as:
- As a clinician [persona], I will use this software during the intake process of new patients [want to], so I can expedite the process and have more meaningful conversations during our appointment [that].
- As a patient [persona], I will use this software during the intake process as a new patient [want to], so I can spend more time talking to my clinician about my issues and not be answering standard questions [that].
User stories like these are powerful because they clearly state the purpose and goals. “Every single thing that we should be making decisions on should map back to these two user stories because this is the whole point of the product that we were launching,” Katie says.
User stories can ease turf wars
Now, let’s see how user stories can also ease friction between teams. Conflict often arises when one group says, “I own this. Why are you getting into my business?” User stories can answer or even thwart that question between groups in your organization.
Front-end development vs. back-end development
In Katie’s prior job, front-end development worked with design, UX, and product teams on layouts and images. When the front-end team turned over the design, the back-end development said it wasn’t technically possible based on the system’s constraints. The front-end team then complained to its stakeholders that the back-end team was not cooperating.
What if they had created a user story like this earlier?
As the back-end dev, I want to be brought into the design process earlier, so that I can weigh in on what is possible.
It expresses the team, its goal, and the benefit to all.
Marketing vs. development
Katie’s former employer had a pie-in-the-sky-thinking vice president who had big ideas about what to add to the website’s homepage. A marketing team member would make the request of the development team only to be told no because it had already committed to working on a different feature in the two-week sprint.
What if they had created this user story?
As the dev person, I want to understand the marketing strategy, so that I can include website features into sprint planning.
This story would have helped the vice president understand that the development team operated in sprints and couldn’t slot in any request. It also works well because the development team can better understand the marketing strategy and proactively schedule time for marketing features in its sprint planning.
The benefits of user stories and the 5P Framework
User stories like these can remove or lessen the emotional impact of making decisions and receiving feedback. “We may not express it. We may not know it immediately, but [emotion is] where all of that friction starts to come from,” Katie says.
For instance, if a manager tells you, “That’s the wrong decision, and I don’t want you to do that,” you would likely feel hurt. But, if the manager shares a user story with less emotionally charged wording, you’ll understand and react better.
That’s the power of user stories. They reframe the conversation around the user and align all parties’ goals and motivations around that user.
In addition, the 5P Framework and user stories work well to organize your thoughts. “It helps you set expectations for yourself and other people. It helps you verify that the outcome aligns with the questions being asked, and it helps you cut down on distractions,” Katie says.
Reframing the LLM request as a user story
Remember the CEO and data scientist interaction about generative AI and LLMs at the beginning of this article? Here’s how Katie would reframe it from the data scientist’s perspective, with a little humor mixed in:
As an underling doing the bidding of the CMO, I want to understand the use cases of generative AI so that I can tell my overlord if there are any applications to consider.
With this understanding, the data scientist will now have generative AI use cases that will help the CEO understand:
- Generation
- Extraction
- Summarization
- Rewriting
- Classification
- Question answering
This user-story attempt will result in a more productive and useful conversation.
Try user stories in your work
The next time you don’t get what you requested from another team, don’t get frustrated and conclude they can’t help.
“The way to fix that is to say, ‘This is exactly what I need from you. This is why. This is how,’” Katie says.
HANDPICKED RELATED CONTENT:
Cover image by Joseph Kalinowski/Content Marketing Institute