This is the third in a series of blog posts that will focus on answering the question “what is the long-term future of Jive?”
In this blog post, I’m going to start digging a bit deeper into the use case impacts of PeopleGraph. As a quick refresher, PeopleGraph is the technology on which we are betting the future of Jive. We began working on this roughly six months after acquisition, and in the last post in this series I provided a bit more detail on what PeopleGraph is and why we believe it is both important and transformative.
The key capabilities that PeopleGraph is designed to enable – Connection, Discovery, and Collaboration – are the topics of this and my next two blog posts. As a finale, I’ll also be describing the powerful new “organizational insights” that Community Managers will be able to glean when our new reporting dashboard begins inquiring and inspecting PeopleGraph.
For this post, I’ll focus on Discovery. Where possible, I’ll try to give guidance for things that are explicitly “on the roadmap” vs. those that are concepts still in the investigation phase.
The Most Basic Facet of Discovery: Search
Search has always been a key feature of Jive and of most enterprise social networking, content management, and interactive intranet solutions. Irrespective of how they use Jive, I have yet to talk to customers that have not cited search as a critical capability – and an area where they would like to see significant innovation and product improvement.
Search in Jive today is adequate – better than most of the competitive offerings, but materially weaker than the consumer equivalents that are the basis for how most users will judge enterprise software today (comparing it to Google, for example). This can generate a great deal of user frustration, and inadequate search in a content-rich enterprise portal can be one of the earliest and most important signals of the “digital crowding” problem I addressed in earlier posts.
Search in the enterprise is uniquely hard – which is why Google abandoned its search appliance and why search within Google docs is so much worse than Google web search. Techniques that are so successful on the internet, such as Google’s PageRank, are significantly less effective in the enterprise because the things those algorithms depend on, such as backlinks, don’t exist in the enterprise content context.
This is where PeopleGraph comes in. PeopleGraph enables us to replicate, in many ways, most of the relevancy and intent advantages that Google’s PageRank and successor algorithms have applied so successfully to the web. At its core, PeopleGraph is a series of links; this is precisely why the same principles that Google uses to make decisions regarding user intent and search result relevancy can be applied by Jive search using PeopleGraph. Jive will make decisions using the volume and strength of various connections between people and the content associated with those people.
The advantages of this go beyond just intent and relevancy. Much as Google cannot manage the content of the Internet, the new PeopleGraph powered Jive search will not depend on a content managed enterprise ecosystem. In theory, as the enterprise evolves and PeopleGraph reflects that evolution, the links and strengths of links between people and the content they are associated with will change. Those changes will alter search results, such that “old” content becomes less relevant as the links to it weaken both in number and in strength.
Let’s dig into a specific example. In this scenario, let’s assume that Jan is looking for information on Amazon Web Services, and specifically the Amazon Web Services migration plan. If you were to run such a search in Jive today, you might type “Amazon Web Services migration plan” yielding (in our own Aurea51 instance) the following results:
Two problems are immediately apparent. First, people routinely refer to “Amazon Web Services” as “AWS,” and because Jive search does not understand that these are the same thing, those entries are all missing from the search results. This is a problem of intent – the old Jive search has no understanding of your intent here.
The second problem is that the old Jive search (and most enterprise search) will bias to older documents precisely because they are old. The newer content with the most relevant information on the migration schedule is lower on the list.
With PeopleGraph powered search things will be different. Let’s take a look at the results of the same search when run on PeopleGraph.
You’ll notice a few things almost immediately. First, the bulk of the results reference “AWS” – Jive search now understands this to be the same as “Amazon Web Services” and, it turns out, most people refer to it that way. This “intent” engine will also help with common situations such as name misspellings or common words that can be spelled differently (i.e. organization vs. organisation). With PeopleGraph, intent can be inferred.
Second, you’ll notice that the nature of the results are different with a significant emphasis on more recent content. This is almost certainly because the link strength to this content is very strong, despite its recency, because of the people who created it or are consuming it. Jive search can understand the strength and breadth of these links as a good indication that this is a substantive, definitive document.
Finally, you’ll notice the search now even includes people, despite the fact that we are searching for what is obviously not a person. PeopleGraph enables Jive search to identify experts on the particular topic – in this specific case the person who is [TS4] accountable for the AWS migration plan. The searcher can use this additional information to go “right to the source” – either directly or by starting a group that includes that person. This is obviously a use case that makes little sense in the internet context but can be extraordinarily powerful for the enterprise.
Passive Discovery: Contextualized Suggestions
A completely new element of discovery that PeopleGraph will enable is something we are calling “contextualized suggestions.” The general concept is not terribly different from how consumer browsing or shopping applications provide suggestions for other content or products that you may be interested in based on the content (or product) you are currently engaging with. The difference here, though, is that in addition to recommending content, PeopleGraph will suggest people in context whom it would be valuable to engage with around the content in question.
Let’s look at an illustration on how this will work. In the example below, you will see that the individual user is involved in a group discussion on a document about a supply chain proposal written in the programming language Python that is for a French speaking customer called Roederer. In this example leveraging PeopleGraph, you will notice now that the document being viewed is now making several “suggestions” – both people suggestions and content suggestions.
The people suggestions are folks within the organization that PeopleGraph has identified as experts on the discussion topic in question. The content suggestions, similarly, are related documents whose content might inform the discussion.
In this example, let’s assume that Marcia (the user) is interested in possibly involving some of the identified experts to further inform the discussion. She clicks on Jimmy to understand who he is and the nature of his inferred expertise.
We notice that Jimmy is among the highest rated resources in the company on python, and furthermore he has created several pieces of content that are highly related to the document being discussed in the group. Marcia can invite him to participate in the group, and Jive can immediately provide the context as part of the invitation.
A Long-Term Future of PeopleGraph-Powered Discovery
Better search and contextual suggestions are straightforward applications of PeopleGraph for discovery, and ones we expect to deliver early in our roadmap once PeopleGraph is deployed. Over time, though, one can imagine additional possibilities.
PeopleGraph is being architected as a service external to Jive, and will include a robust API (integration) layer to make it easy to connect with Jive and other critical data sources – HR systems, Active Directory, Office 365, Slack, Box, Salesforce, etc. Every source that PeopleGraph connects to will enrich PeopleGraph’s understanding of the organization. This means that, long-term, PeopleGraph can understand:
- The content associated with Slack or other transient messaging applications, enabling that to inform search as well as PeopleGraph’s understanding of organizational relationships
- Connections to people outside the walls of the enterprise, such as customer relationships as mastered in Salesforce and how those relationship strengths manifest themselves in the company context (how strong is our company’s relationship with Apple, and what are the specific relationships we have and who has them?)
- Content associations for documents stored outside of Jive, ultimately enabling search to find people or content that is not exclusively informed by what is resident within the walls of Jive
We believe PeopleGraph will drive a profound change in people and content discovery – initially within Jive but over time, increasingly drawing from the ecosystem that surrounds it. Over the weeks and months ahead, we will be releasing demo videos of the prototypes in action to give you a real sense of how this will work and the progress we are making on it.
As always, I invite your feedback and questions. Thanks for taking the time to engage in this discussion with us.