5 ways chatbots are revolutionizing knowledge management

The fields of knowledge management, information management, and content management have become critical to a modern workplace. Finding, documenting, and knowing things in an environment where data is dispersed, employees are always on the fly, and career paths change fast must be intuitive, simple, and seamless.

Since the early 2000s, employing sound file management practices, mixed with optimized search, have been the rules of thumb when it comes to KM and ECM systems. But no longer will that be enough. From this nascent point in the era of AI and chatbots, you will be left behind if you’re not putting a good bot to work. Now’s the time to put chatbots in your strategic plan.

Finding content in a site structure requires a mental roadmap of where things live. Search may provide good results, but not direct answers; the answer is usually in the file it returns, meaning more time digesting to understand. Bots let you skip straight to the answer while pointing you to the source for reference, saving everyone time and bridging what is now becoming a major gap in IM strategies.

1. Bots organize information better

How your information is organized affects how people find and use it.

In a standard site and library hierarchy, your files can be well organized using a strong folder or metadata structure. Of course, the effectiveness of the hierarchy depends on 1) the strategy used for organizing the content from the start and 2) how well the owner of the hierarchy has maintained the structure and the content over time (including removing ROT when necessary). Generally speaking, a well-organized hierarchy that is intuitive, structured, and timely can work well for finding information.

With search, the content isn’t organized, which is kind of the point. The search engine will provide multiple results in an organic fashion based on keyword matches, any metadata refiners, and, of course, past popularity of the files. Generally speaking, this can work well when the user has no idea where to find the destined information (or cares not to spend the time going through a file structure).

With a bot, the bot content owner(s) predict what users want to see and provide direct responses (especially direct answers), with links back to the source content. From the users’ perspective, the information isn’t organized (even though on the back end it is) nor does it provide organic options like search; the bot gives the best answer it has (assuming it has one) and provides it in a conversational way. This direct method of providing information means the user does less work for the same information and can perform the task over and over as necessary.

2. Bots provide what’s needed when it’s needed, nothing more

What information is available to the user matters because the more information there is, the more overwhelming it can be to sift through a site structure or scroll through search results.

With a site hierarchy, it’s assumed that a user knows to, for example, browse to the human resources site to find information on their employee benefits. And while this may be assumed, it’s not guaranteed. The site structure may be more confusing than intended or, frankly, the user may suffer a bout of laziness and give up.

But even if they do know where to go for the information, clicking through folders, views, and filters is still a chore and can dissuade someone from looking further for a file they need. Ultimately, they tolerate not finding the information (possibly affecting the quality of their work) or contact someone else for help (using that person’s valuable time on a task that adds minimal overall value). In general, this impact on finding information is okay, but not great.

With search, you’re stuck with results that just recently combed everything that you have access to. Even a user who knows searching best practices paired with a system with a smart search setup (e.g., promoted results, customer refiners), the user still has to deal with extraneous results that just aren’t relevant. From keywords that overlap (e.g., “office” for facilities info or “Office” for IT application use) to outdated information, one must sift through plenty of data in search thanks to the nature of its organic results. It can lead to an overall negative impact on the experience of getting to your information.

With bots, the information available is completely dictated by the bot owner(s), those people who organize the information the bot has and how it directs users to the source information they seek. A good bot has answers to most common questions for each group or department in an organization, actually answers the question being asked (rather than solely providing a source for the answer), and links back to the source as a reference for further information.

The answer is valuable because, well, it’s what the user was actually looking for. The reference is also powerful because it immediately points the information seeker to the source should they need it. When it comes to the available information, bots maximize what’s available and provide high returns on the investment on responses.

3. Bots make the high-value information predictable to find

When it comes to finding what you expect, site hierarchies, search, and bots score very differently.

With your folder structures, the content is generally well organized. Finding what you want to find is predictable because if a file was there last week, it’s likely still there this week, probably in the same site, library or folder. The hierarchies are dependable old friends that, once we know our way around, can be used over and over to find information. It may not be easy to set up and maintain for the owner, but a well-organized information structure will be easy to use and appreciated by users.

Search, on the other hand, is unpredictable. The entire point of a search algorithm is to give various attributes higher and lower values, which pushes them up or down on the search results. Knowing that it’s almost a law of nature that nobody goes beyond the first page of search results, it’s critical the right information gets where it needs to be so users can actually find it. Pinned search results (e.g., best bets/promoted results in SharePoint) give administrators a little wiggle room to force a result, but usually only one item per search term can be pinned. Generally, the predictability and dependability of search in this situation is bad because what shows up in today’s may not be what will be there next week.

With a bot, you have a happy medium: you dictate what the answers are to the most-sought-after queries and provide resources as a quick way back to the source information (via link). It can be overwhelming at first to decide what to include. An easy way to start is to combine the top, say, 50-most-common search queries from your intranet’s search analytics with a known list of FAQs per department or group in your organization. If you have even three-quarters of those topics covered, you’ll see plenty of use of the bot. Capture any unanswered comments from users to identify what else people want to know. A bot bridges the gap between predictable and unpredictable information management.

4. Bots force you to curate only the high-value content

Curation of content is important. Your intranet home page may provide dynamic content, but ultimately someone with a plan has organized how that content will display and has chosen what to feature and what not to. The same goes for your overall content management.

In your sites and libraries, you host everything you have. And to make the content easiest to find, content owners truly have to put in effort to curate the content. Without it, you’re left with a mess of files strewn about in an unpredictable and ad hoc setup. And it’s not uncommon for this to be the case no matter how strong a strategy you have. Regardless of the quality of your curation, any curation takes time and effort to start and maintain. And if you do it to one area, you’re pretty much doing it everywhere in that area and in others as well. It can be a lot of work.

Search is the opposite. You don’t have to really curate anything. The algorithm provides results that are organic. Any curation that is applied is generally done using search refiners and pinned results. Search comes with minimal effort required in the realm of curation, but it also means your results are highly subjective and unpredictable.

Bots bring you a happy middle ground where you can curate only the content that’s valuable. Sure, it’s important to retain records of things that happened seven years ago, but it’s unlikely you’ll need to see that often. That kind of file is curated in your site. Search provides organic responses and its analytics can provide insights into what’s popular. But search can only give you the source of the information.

If you want to know about the vacation policy, search will likely return the employee handbook; but you’ll have to sift through that file to find the section on time off. A curated bot can answer the question about time off and link to the employee handbook for reference. But the curated response is the answer the user was looking for, rather than the source. A curated bot skips the annoying step of having to read, digest, or further search for information after you found the source you wanted.

The bot curation process is perfect for high-value information that’s requested frequently. For minimal back-end effort, maximum front-end successes can be garnered, making bots an excellent supplement for information management.

5. Bots require minimal user training

Unfortunately, in today’s information management sphere, good IM can only come with a well-trained user base. But everybody has their day jobs and nobody really enjoys having to take a course (required or not) to learn something as basic as folder structures and search. (Even though, yes, they should.)

A good site and library structure takes a while to understand and explore on the user side. If you’re skeptical of that assessment, ask the newest member of your team how long it took them to understand the information architecture in your team. Even the best structures take time to learn, and that’s time taken away from other work that could be getting done.

Search, one could argue, comes with no learning curve. Everyone knows what to do with an empty box with a magnifying glass next to it. But it’s not that simple. Sure, you can use out-of-the-box search that way if you’d like, but any smart search setup with use configured search refiners, pinned results, and more to get the most out of search. Then there are the user-level best practices that many people just aren’t aware of. (But they would be if the information was made available.) And that training and setup process takes time.

A bot, on the other hand, is trained to answer the common questions directly and if it doesn’t have an answer, it can provide a contact person as an alternative. On the user side, the bot truly requires no training. You simply ask a question, get a response, and hope it’s what you want. If it’s not, you provide feedback indicating so.

Wrap up

The KM/IM/ECM field(s) is in for a shake-up with real, practical AI solutions that can get us over the limitations of site structures and search but connecting users directly to the information they want. Point A to Point Z without the stops along the way. A smart implementation process can buy huge, immediate wins. As you start playing with bots in your organization, keep these steps in mind:

1. You need a bot. You can spend many thousands of dollars and months of development time building one, or you can get up and running in a matter of hours with a bot like AtBot. There’s even a 30-day free trial to see if he meets your needs.
2. Document the top x-most common search queries (I suggest starting at 50) in your search analytics and revisit this list every month or so.
3. Collect and document each team’s list of frequently asked questions, any knowledge bases they have, or any cheat sheets they use to find commonly needed content.
4. Set up your bots using a tool like QnA Maker to answer these questions. In your responses, link to the sources.
5. Collect user messages that garner no good response. Apply a feedback mechanism to understand user needs.
6. Review your answers on a timely basis to ensure they’re accurate.
7. Remove outdated responses. Document changes. Follow best practices.

AtBot brings AI within reach

AtBot is the premiere bot-as-a-service solution for the Microsoft cloud. Built completely within Azure, AtBot is your out-of-the-box, easy-to-configure bot for Teams, SharePoint, or the web. Teach AtBot tasks using Power Automate, make him your corporate source of knowledge with QnA Maker, help him understand almost anything your colleagues could ask thanks to LUIS, and manage his features with the AtBot Admin Portal. Get going with AtBot Free or start your free trial of AtBot Premium today.