The pandemic has exacerbated many challenges that (thankfully) Shelf helps contact centers overcome, like getting new employees up-to-speed and ensuring distributed workers have access to oft-changing organizational knowledge.

This AI-led vision is proven to help contact center operations control call volumes, improve handle times, reduce the training burden, and achieve operational efficiencies across the different channels where businesses meet their customers.

Let’s dive into just how Shelf’s knowledge automation platform drives valuable outcomes and tangible ROI that today’s successful contact centers require.

Clear ROI

Knowledge Automation brings value to many contact center interactions, and it all boils down to one simple formula—eliminating the unnecessary steps to find answers, saving agents and customers a huge amount of time, translating into considerable ROI for every business.

Our formula accounts for a few factors:

Number of interactions per month

Across the entire contact center

Average time saved per interaction (by type)

Based on contact center averages

Average training time for new agents

Based on contact center average

Hourly rate of an agent

Average rate of fully trained agent

% of interactions per KA component

Based on rate of interactions per channel

*Assumptions included in the following calculations are based on contact center averages and listed at the end of this article.

# of Interactions per Month

Avg. Fully Loaded Agent Hourly Rate
Avg. cost per Interaction
% of interactions Agent Assist Relevant
Time saved per Agent Assist
3 minutes
% of interactions require Guided Workflow
Time saved per Guided Workflow
7 minutes
% of interactions aided by IVR Knowledge
Time saved per IVR Knowledge
3 minutes
% of interactions reduced by Self Service
Average Training Time
26 days
Avg. Cost per Agent Training Hour
Avg. Agents Trained per Year
Avg. Training Time Saved with Shelf

Total ROI from Knowledge Automation


 Let’s break down how we arrive at this total. All scenarios below assume the contact center conducts 1M interactions per month.


Our assumptions estimate an all-in average agent salary of $20 per hour, with approximately 20% of calls requiring an agent to access knowledge to provide an answer.

In a scenario without Knowledge Automation, the agent must perform the following steps:

  1. IVR routes the call to an agent
  2. Agent opens a new window or tab to access their content system
  3. Agent conducts a search
  4. Agent scrolls through results to find the document in need
  5. Agent scans document to find solution
  6. If solution not found, agent must repeat the process

In this scenario, an agent spends an average of 3 minutes to find the information they need. With Knowledge Automation tied to the IVR, we’ve saved 3 minutes per interaction.

If 20% of conversations are aided by IVR Knowledge, we can calculate the following ROI:

$20 per hour

1M interactions per month x 20%


120,000h saved per year

1M interactions x 20% x 3; converted to hours per year


$2.4M saved per year

hours saved x hourly rate of agent

Agent Assist ROI

In this example, we make the same assumption that without Knowledge Automation, an agent conducts a 3-minute search process to find an answer.

Based on our direct experience approximately 25% of contact center conversations are aided by Agent Assist in some way. This is supported by the fact that customer interactions usually include multiple questions and answers and Agent Assist is designed to assist on all of them. Thus we conservatively calculate the ROI as follows:

$20 per hour

1M interactions per month x 20%


150,000h saved per year

1M interactions x 25% x 3; converted to hours per year


$3M saved per year

hours saved x hourly rate of agent

Guided Answers ROI

For this scenario, we are calculating ROI based on the amount of time an agent spends searching for a document, then executing a step-by-step procedure manually.

We estimate that an agent will take 14 minutes to complete the following process:

  1. Begin scanning document to initiate procedure
  2. Find initial step and share with customer
  3. Scan document for next step
  4. Find second step and share with customer
  5. Continue scanning page after page to until complete

Guided answers can cut this time by half. Thus each interaction that requires an agent to walk a customer through a process or procedure would save the company 7 minutes. If we assume 20% of interactions require an agent to use a process or procedure, the ROI calculated is:

$20 per hour

1M interactions per month x 20%


280,000h saved per year

1M interactions x 20% x 7; converted to hours per year


$5.6M saved per year

hours saved x hourly rate of agent

Guided Training ROI

To calculate the ROI for training, we’ve taken some contact center industry benchmarks to create our formula:

  • Cost of $20 per average hour of training per agent
  • Approximately 250 agents trained per year
  • 26-week average training time for new agents

Through using the combination of guided answers, IVR knowledge and Agent Assist companies can shorted training time by approximately 25%

Using these criteria and assuming that a company might onboard 250 new agents each year, we formulate the following ROI:

$20 per training hour


65,000h saved per year

26 weeks per agent x 250 x 25%; converted to hours per year


$1.3M saved per year

hours saved x hourly rate of agent

Self Service ROI

In a Self Service ROI scenario, we’re again using contact center industry standards to create our formula.

Companies report deflecting anywhere between 30-50% of interactions with AI. For the purposes of this calculation, we’re assuming a conservative 10% of interactions will be deflected.

$4.50 per interaction

avg cost per interaction


1.2M interactions saved per year

1M interactions x 20%; coverted to yearly total


$2.7M saved per year

Time is money, and Knowledge Automation is primed to save contact centers millions per year, just by streamlining access to information and integrating knowledge into all your channels of communication.

From ROI Numbers to Addressing The Problem of Inconsistent, Inaccurate Information

The value proposition of any contact center knowledge technology must begin with providing agents with the means to share complete, correct information on each and every interaction that takes place.

Why? Because when the information provided by agents is inconsistent — or worse, incorrect — you have a First Contact Resolution problem on your hands. To successfully resolve issues when incomplete/inaccurate/information is provided, an interaction will need to be “re-opened”.

Not only is this inefficient, depending on the severity of the misinformation, the interaction has a high likelihood of needing to be escalated. Multiple people get involved, including senior, experienced team members, and you expose the business to the #1 problem every company must avoid — customer churn.


Introducing the Ebbinghaus Forgetting Curve

This all sounds very straightforward, and in concept, it is. So why is this a problem that plagues so many customer service operations? It all comes down to neuroscience and how humans process and retain information. The Ebbinghaus Forgetting Curve states that humans forget 90% of what they learn in one week and 50% of what they learn in one hour. This is incredibly problematic for today’s contact centers, where information is constantly changing.

And there is only one viable solution to address our brains’ inherent limitations – documented knowledge that agents can rely on. But this comes with a major caveat— the technology that houses documented knowledge needs to make it extremely easy for agents to get to the right information quickly.

Introduce friction into the information retrieval process and agents will create workarounds in an attempt to keep pace with their expected KPIs. They’ll ask a coworker, create a personal (and unsanctioned) cheat sheet or they’ll just “go off of memory”.

Additionally, if the documented knowledge forces agents to jump through too many hoops to get the right information, they’ll find workarounds and other ways of avoiding these inconvenient steps. And the more cumbersome the information retrieval process, the less likely agents will be to use the knowledge bases that house all of your policies and procedures.

What does this look like in practice?

  • Key pieces of a policy are not communicated to a customer over the course of an interaction
  • An essential step in a procedure is missed and a transaction is processed incorrectly
  • Outdated information is shared, creating the wrong expectation in the mind of the customer

Each of these scenarios creates a ripple effect characterized by multiple callbacks, upset customers, long interactions with senior contact center resources, accommodations made by the business to “make things right”, and a contact center that suffers under the weight of more interactions than it can reasonably handle in a timely manner.


The Value of Easy to Follow, Guided Answers

The best solution to this problem is to use AI (Agent Assist, IVR Knowledge) to present the agent with the right information, and then utilize decision trees to present this information in an easy-to-digest format. While administrators are making use of front-line feedback through the Feedback Manager and the insights produced by the content health monitor, you have a recipe for success. By combining these technologies contact centers ensure that the right steps are taken in the right order and the right information is presented at each step.

Quantifying the ROI of fixing this dynamic is not easy, but one thing can be said for certain. Ask any seasoned contact center manager, and they’ll tell you that this is where the seeds of customer and employee churn are sown.

More Value Out-of-the-Box

Top Rated Platform for Usability, Ease of Use

While many vendors provide storage, structure, and search to knowledge management, we believe that user experience, ease of onboarding and the ability to seamlessly integrate into your business tool suite are all equally, if not more, important. We’ve spent years performing in-depth data-driven user testing, optimizing our UI and reducing the friction of onboarding new people so you and your admins don’t have to worry about it.

This customer-centric approach has been highly recognized and awarded. Shelf is the #1 rated knowledge management platform on Gartner Digital Markets and recently won top awards from G2 Crowd for Ease of Use, Highest User Adoption, Easiest Setup, and Easiest to Admin across all contact center knowledge technology.

Proprietary, Award Winning MerlinAI

While Knowledge Automation has proven to deliver outstanding results starting on Day 1 of use – that’s just the beginning. Thanks to our award-winning, proprietary MerlinAI technology, Shelf improves its performance and gets smarter over time. By tracking all user behavior and feedback it gradually learns to improve:

  • IVR Knowledge presents agents with more specific knowledge before taking the call.
  • Agent Assist recognizes intent indicators that lead to specific content, presenting agents with answers at the speed of conversation.
  • Content Maintenance Assistance suggests more accurate metadata to improve findability.
  • Training and onboarding new team members is much easier when they aren’t required to retain vast amounts of knowledge to support customers.

Shorter interactions and accurate resolutions require consistent, reliable information. And when it’s predicted and automatically delivered—agents can skip the arduous search step completely.

MerlinAI powers answers across all avenues customers contact support teams, at the pace they’ve come to anticipate. And as customer service orgs continue growing to meet their increasing interaction volumes, MerlinAI will help alleviate training and onboarding of remote agents so they’re up and running even sooner.