Last month, we were excited to announce Talla’s Learning Dashboard, a report-card-like view to show Talla’s learning. Our customers are loving it, and today we’re going to take a look at some of their favorite things that the learning dashboard measures.
Knowledge Base Quality
Customers can see the general quality of their knowledge base on the learning dashboard, which is driven by activity in Talla. The activity is broken up by page and you can easily see what is working and what needs extra attention. One great thing about this view is that it highlights how many pages Talla is learning from in the “Trained Pages” section. This comes from page-level training activities. Some activities include identifying entities, defining glossary terms, accepting suggested keywords, and defining acronyms. All of these activities help train Talla to provide more precise answers to questions it has never seen before.
Knowledge base quality also comes from use of the verification feature. By using the verification feature to authenticate information, you are essentially telling Talla, “Yes, this information is correct. Use it. Learn from it.” Using this feature to train Talla prevents your knowledge base from getting stale. Our best customers assign a subject matter expert to review each page on a regular basis. Depending on how often the information on that page changes, this can vary. Some pages need to be verified daily, others weekly, others annually, and so on. The verification trigger serves as a stop-gap and ensures that the knowledge base stays up-to-date. It’s also a great checkpoint to ensure all of your team’s current tribal knowledge is being documented. Talla is learning each time information is verified, so that in the future it can predict when certain information will need verification, and proactively ask the subject matter expert, “Is this up to date?”.
Customers can easily see how many questions users are asking Talla on the learning dashboard. This is a great way to measure user engagement. The more questions people are asking Talla, the more language it is being exposed to. Feedback usage is another way to gauge user engagement. Users easily provide feedback within their workflow by clicking "Helpful" or "Unhelpful" when Talla delivers an answer. This feedback is also a great way to detect a knowledge gap. If Talla delivers the incorrect answer (one marked as "unhelpful"), many times it is because the correct answer has not been documented. Looking at questions that were marked as unhelpful gives the opportunity to close that knowledge gap.
By measuring training effectiveness, customers can see evidence of Talla learning over time. The dashboard shows a graph reflecting the amount of times that Talla has returned correct answers. This feature encourages users to keep training Talla and serves as a great checkpoint. Just like you, our customer's information is always changing. This view reflects Talla's delivery of precise answers, even through times of change.
Talla is a true lifelong learner, and you can easily track that on the Learning Dashboard. Sound interesting? If you would like to get a demo of Talla to learn more about how you can save support costs with automated responses, actions, and increased rep productivity, let us know!