How to use Trends to find Hidden Stars and Work on a Perfect Project? People Analytics will make you a star!
What is a trend? Oxford English Dictionary defines it as "A general direction in which something is developing or changing". It is a general definition; more commonly used nowadays is "trending", which means what is "currently popular or widely discussed online, especially on social media websites."
In a large multinational organization it would be very useful to understand what topics are internally trending. Knowing what is trending would help employees navigate through vast complexity of such large multinational organization.
How to bring these benefits to every employee in an organization - these are the questions we've set ourselves to find out in our latest innovation project.
Imagine the following scenario: every employee can see the list of trending topics in its own organization. Let's imagine a large automotive company, which has hundreds of thousands of employees. How would an employee find what is going on around her? It is a daunting task to find out about existing projects; let alone the latest movements and trends in the manufacturing, software, marketing or sales.
Why is that important for an employee?
To navigate internal complexity. To learn something new and up-skill. To find new, exciting project to work on. To find who is working on it, so that they can work together, to perhaps create something new and exciting.
How could it look like?
Note: screens below are one of our concepts, the final version may vary
Here is the deep dive into one of the trending topics "Future of Mobility". An employee can see that the topic is having an upwards trend lately. Moreover, an employee can see the most influential people involved in the topic. (internal trend setters, or internal influencers)
If an employee likes this topic, how about connecting to key influencers that are working on (or dealing with) the topic? Algorithm finds it for you - making it easier to connect and share knowledge.
What else could be important for an employee? For example, what are related projects from the internal marketplace - so that he can take part in those projects, if he wants to.
How does it work?
The concept of topics extraction is based on Latent Dirichlet Allocation. It is an unsupervised machine learning algorithm, and it clusterizes topics from set of documents (or emails). Note there is a difference between LDA algorithm and k-means algorithm. Namely, k-means clusterizes documents in set of K disjoint (exclusive) clusters. LDA is more fuzzy, because a document can belong to more than one topic (e.g. 60% to topic A and 40% to topic B). For that reason LDA makes more sense when understanding topics in the real world.
We have used emails from Exchange server to understand and clusterize topics. In our dataset every email represents one document. Firstly, we extract topics from the whole collection of documents (resulting in N clusters of topics, where N is predefined), and then assign each person (email sender = document owner) to given clusters.
For much more elaborate explanation of the algorithm, see here (Part 5)
Privacy Considerations?
Touching employee emails, even for automated processing, can be sensitive. Therefore we have implemented list of "no-no words" that are not associated with anyone in particular. The reason for that is simple, illustrated in the following example: if there are discussions about layoffs - noone wants to be seen as a key influencer in topics of layoffs. List of "no-no words" is company-specific, so it needs to be curated carefully.
Privacy considerations are always balanced with genuine need to share own work inside the company - after all, wouldn't it be great to share what you are working on, and find colleagues that are working on similar (or complementary) topics.
Instead of emails - we can utilize contributions (OKRs) - to find exactly the same information (who is working on which topics). Any channel of communication inside the corporate, including documents stored on Intranet or internal document management system can be used.
Therefore, conclusion is clear - even though this approach can be extremely beneficial for an employee, a special care is needed to prevent any (however small) possibility of misuse.
How is user experience?
User experience is very elegant, and the whole complexity of the back-end calculation of ONA network (influencers, information brokers, etc), as well as topics extraction are hidden from the end-user. Here is how it could look like for an end-user:
And here is how an employee can understand his network in greater detail:
Simple and elegant - while creating contributions, employee can find immediately people working on similar topics (company-wide or from own immediate network), or find projects on the marketplace that are close to what he is working on.
Conclusion
Benefits for an employee can be enormous. In a large corporate - it will help an employee navigate internal complexity, find people working on similar projects, and collaborate. We believe that alignment and collaboration will improve manyfold - all thanks to complex technology in the back-end that makes life simpler for employees!
This article is part of the series on ONA@Haufe
- How to use corporate e-mail analysis to reveal hidden stars and ensure equal opportunities (Part 1)
- Technical overview of e-mail network-based Insights (Part 2)
- Deep dive on e-mail network-based Recommendations (Part 3)
- How to use trends to find hidden stars and work on a perfect project? People analytics will make you a star (Part 4)
- How to implement e-mail content-based analysis (Part 5)