The solution to workplace information overload is limiting data flow to the most relevant and urgent emails, documents, events and posts. One possible way to do so is to organize and present content by business topics, such as customer, project or product name. This all sounds simple, but is really hard for at least the following reasons:
- What is an important topic?
- How do you (automatically) classify content by topics?
- How do you decide what is most relevant and urgent for each person?
These are complex problems because people interpret information differently. Topics differ by individuals and groups, so trying to define them in a meaningful way is impractical for large groups of people like departments and entire organizations. Even in the rare case where organizations agree on a common vocabulary, classifying emails or documents by topics accurately and uniformly is virtually impossible because of the unstructured nature of documents, and because of the ambiguity built into everyday language. Even assuming we were able to solve these issues, we still have to figure out what information is most relevant, interesting and urgent for each individual.
It would be a big deal if someone was able to solve these problems.
Any solution to this problem is not strictly a technological one; it requires translating how people think into a set of topics. Current taxonomy projects, where organizations define a common language, are typically huge consulting projects. Experts interview organizational stakeholders, then build a taxonomy, and finally classify each piece of content according to the agreed upon terms. More often than not, these projects are political nightmares as different stakeholders fight over points like "what is a topic." The problem is not trivial, because defining a business vocabulary often determines which department garners more clout. Case in point, the CTO of a large federal agency told me about her organization’s taxonomy project. After two years of haggling, the organization was able to define just four terms that everyone agreed on, and two of those were official department names and a document’s creation date.
That’s why only the most knowledge-intense organizations have had the motivation, the time and the budgets to invest in such a process. The rest of us are left to our own devices to store and organize our information as best as we can. Can Viva Topics break the logjam?
Is Viva Topics the First Topic Computing Platform?
I’m a long-time believer in topic computing as a practical way to deal with information overload. Topic computing is defined as:
“a new way of working with existing devices and cloud apps that provides a streamlined user experience to connect all information in a meaningful way. With Topic Computing, information is organized by topic, such as customers, products, services, or projects, so workers can focus on what matters, rather than fend off annoying and stressful notifications, or endlessly toggling between apps looking for information.”
Topic computing has proven elusive, which is why I am excited about Viva Topics. Viva Topics is part of the new Microsoft Viva Employee Experience Platform (EXP), announced earlier this month. Viva Topics purports to “automatically organize content and expertise across your organization, making it easy for people to find information and put knowledge to work.” Sounds like topic computing, no?
Let’s take a closer look.
Like other Viva modules, Topics is not a product, but rather a new user experience built on a collection of existing Microsoft technologies. The Topics module sits on top of Azure AI and topic technology originally developed for Project Cortex, which is a Microsoft service that “uses advanced AI to deliver insights and expertise in [popular] apps, to harness collective knowledge and to empower people and teams to learn, upskill and innovate faster.”
The first product out of the Cortex project was SharePoint Syntex, which uses AI to automatically classify documents with SharePoint managed metadata. Metadata values represent the documents’ topics. Viva Topics takes Syntex one step further by automatically applying SharePoint metadata to other types of Microsoft artifacts like emails, events and Teams conversations.
Related Article: Vannevar Bush Points the Way to a New Era of Computing
The Viva Topics User Experience
The picture below demonstrates the Viva Topics experience. In this case, an employee encounters an unknown term in a Teams conversation. The term has been flagged as a topic, so when the user clicks on it, they are presented with a Topic Card or a Topic Page, which displays details about the term, including what it means, pointers to documents that contain the topic, and a list of colleagues who have created or shared content related to that topic. The user simply clicks through to find related content or people who have expertise with that topic. By focusing on the topic at hand, the worker stays focused on their given task instead of having to toggle between apps looking for related content.
In general, users can add and edit topics, to enrich the overall taxonomy. Organizations are free to determine how much autonomy they grant users to alter topics. As we shall see, selecting the appropriate degree of topic freedom will have a lot to do with the success (or failure) of an implementation.
Related Article: Microsoft's Project Cortex Wants You to Hire a Knowledge Manager
How Does Topics Work?
We have already seen that SharePoint metadata is the underlying construct for mapping content to topics. The second important Viva Topics construct is the Microsoft Graph, which is “the gateway to data and intelligence in Microsoft 365.” The Microsoft Graph is essentially a graph database that maintains the relationships between topics, content, and people in Microsoft 365. When content is created or shared in a Microsoft 365 app, or people interact using a Microsoft app, the Graph records the activities. These signals can then be mined by AI/machine learning models to predict which content someone would most likely want to see in the future. As people engage with (or ignore) recommended content, the machine learning models improve their predictions. Because the models see everything a person does in Microsoft 365, they are well-positioned to predict the content that is most relevant and urgent for each individual.
But we are not out of woods yet. Two thorny problems remain to be solved.
Can AI Solve the Taxonomy Problem?
The first problem is overcoming the political issues associated with defining organizational topics, as described above. As a technology platform, Viva Topics assumes that an organization (or at least groups within the organization) can agree upon a common language for topics. While automating the topic extraction process will certainly help by suggesting topics from existing content, it remains to be seen how organizations will cope with the flexibility to override automated topics with others. As good as Viva Topics is, it won’t solve political issues associated with agreeing on a common language.
The First Step Is the Hardest
The second problem is how to bootstrap the system so that users see value from day one. This is one of the biggest challenges with any AI-based solution. Classically, there are two solutions: one is to pre-populate the system with a set of topics as defined by content subject experts and let people edit those; the other is to start with a blank slate and let early adopters populate the system with topics as they go along. Of course, a combination of the two models can be used as well, by starting with an incomplete set of topics and building up the topic store over time.
How does Viva Topics work best? I put this question to Naomi Moneypenny, Microsoft director of content services and insights. According to Moneypenny,
“Companies have a choice of how they implement. They can do something that's a bit more formal with knowledge managers and then invoke people when they feel it necessary, or they can crowdsource this across the company. [Since] every company has a little bit of a different culture, how they run their employee experience platform is a function of that culture … we give them options on how to implement that.
"In some organizations a Knowledge Manager might be a great role and other places might prefer an open, crowdsourced approach… We're really trying to make it easy for people to adopt Microsoft Viva in the way that makes sense.”
Related Article: Microsoft Viva – Who Is It For?
Will It Play in Peoria?
How many organizations will be able to overcome these two hurdles to successfully implement Viva Topics and gain the benefits of Topic Computing? That’s the $100 million question. Of course, companies don’t have to go it alone. An army of service provider partners will surely be ready to help organizations devise best practices for topics and to implement Viva Topics.
I, for one, can’t wait to give it a whirl.