Data and robots to make enterprise social networks successful


One day, annoyed by an unexpected problem, an employee figures out that the best may to find a solution is to use the social network his company deployed months ago. He logs in, throw a message in a bottle and….will never get any answer.

The “message in a bottle” case is the typical use case of an enterprise social network and most business cases rely on it. It’s also the case that’s often the least successful.

Why ?

1°)Because the employee had never used the network before and decided to do so in this case, out of frustration. He doesn’t know the tool, how it works, neither its codes and good practices, how to identify the right person, communities, how to take a stand. Since he had never participated or contributed, it’s also possible that people won’t be likely to answer quickly compared with an active contributor. Since he has no contact, no one sees his message in a bottle that eventually gets lost in the Bermuda Triangle.

2°) The platform is underutilized or even not utilized at all. He won’t find anything there because…there’s nothing at all. Poorly identified business needs, designed to be nothing more than a water cooler….many factors leads there.

Enterprise social networks replace inadequate search engines

3°) Information exists but somewhere else. Organizations didn’t wait for social networks to produce value-added information the employee could use. The CRM could help him discover who worked on such issues for a customer, project management tools, shared disks are full of insights and procedures, as well as the employee directory if rich enough, fed by external data and updated by employees. Unfortunately the “global search” is a wild dream in many companies, useful data locked into inaccessible directories etc. In the end, what’s the purpose of an enterprise social network ? Replace the search engine by people and ask them to find what one can’t find by himself. Of course the network can deliver more value because people can add context to raw information to make it relevant and actionable but in most cases the purpose of the network is to find people who know information because the information itself is unreachable.

4°) The right people are not available. Enterprise social networks are asynchronous while this employee needed a quick answer. Those who can help are certainly busy working on something else and don’t have time to check every single update or answer him despite of their willingness to help. The less the social network in integrated with “usual” tools like email le less it’s anchored into people’s day to day lives the more this scenario is likely to happen.

At least 4 good reasons for the “mother of all use cases” to be rather deceptive. The life of a social platform relies on the active participation of a critical mass of people, their availability, attention..and therein lies the rub

But solutions exist and coming ones are even more promising.

Some businesses already found a response by being a little bit more directive. The idea is to smartly counterbalance the lack of practice, contacts and availability. It’s about pushing the request to the right person, no matter they are in contact with one who asks, follow him or are members of the communities where he shared his problem. Provided he used the right hashtag, is request can be pushed to the right people, based on their previous activities, their profile etc through the use of semantic and analytics. At the very beginning it was a custom developement but I bet such things will become a basic functionality in a near future.

But what’s coming – and I’m excited to see coming – is much more promising and is a very good operational use case of cognitive computing as a collaborative agent.

Imagine a system that indexes all the valuable data through the company. Structured and unstructured. It’s possible and I already mentioned a case based on implicit social networks. Then imagine that, when a message in a bottle is being thrown in the social network, via the microblogging function for example, a “Watson” or one if his likes processes the request and answers the user.

Agents will replace human intermediaries for the better

Pros :

– users ask questions in natural language. “is it possible to…”, “how to…”

– he gets an instant answer

– the result is not or not only a link to relevant information but an understandable and actionable answer in plain english (or french :)). It’s important to understand the difference between links leading to sources allowing to find a solution and a solution in plain text. More, the system can tell how confident he is with the delivered answer.

– no matter the user is a neophyte or has a developed network : he will get the best possible answer.

– everyone saves time because solicitations will decrease

By the way, I’m talking about embedding such a system into an enterprise social network because its usually associated with the idea of messages in a bottle but that’s nothing more that the future of the enterprise search engine. It will ideally find its place on the homepage of the intranet.

Employees are not the “poor man’s search engine”.

Asynchronous participation will always remain an issue for enterprise social platforms. But when people are looked for only to serve as an advanced search engine, there are alternative solutions that can improve things. Participation becomes passive, more valuable and people will always get what they need.

Tomorrow we’ll ask question to smart agents and “Hey Watson” will replace “does anyone know…?”. A revolution in enterprise social networks use cases but a necessary one.

Employees have an incredible added value to deliver to their colleagues. But not when they’re just asked to replace a search engine.