How to Craft Your 2021 Conversational AI Roadmap

  1. CX Insights
  2.  » 
  3. Linc Video Library
  4.  » How to Craft Your 2021 Conversational AI Roadmap


The Future of Chatbots and Conversational AI in a Post-Covid World


In today’s webinar session we’re going to be walking you through the steps of how to craft a conversational AI roadmap that is specifically tailored to the unique challenges of 2021 post Covid-19 Pandemic.

Things have changed drastically in all business sectors but especially in retail.

The same rules of retail and automation that applied last year in 2020 are out for this year underway.

New advancements in high fidelity omnichannel digital workers and robust engineering have created new avenues of automation in 2021 that were previously not possible.

The amount of savings to you and the seamless shopping experiences you can create for your customers with these technologies is endless and absolutely vital in the “New Retail  Normal”.

In this webinar Durk Stelter Chief Revenue Officer for Linc Global dives deep into the unique options available to your Enterprise in regards to technological advances in natural language conversational AI and also the pitfalls to avoid while building a successful conversational AI roadmap for your brand.

If you have any questions from the webinar feel free to Schedule a Demo of the Linc platform and we can walk you through all of your automation questions.


About the Webinar Presenter


Durk Stelter Chief Revenue Officer Linc Global

Durk Stelter Chief Revenue Officer | Linc Global

Durk Stelter, CRO | Linc Global

As the Chief Revenue Officer for Linc Global Durk is responsible for everything that faces Linc’s customers.These include sales and marketing, customer success, and delivery. Durk comes from a cross-functional background in product management, product development and engineering operations playing key roles in enterprise brand startups in customer experience automation, including his own. Durk is an industry expert and early pioneer in customer experience automation, working with systems dating back to voicemail and Ivy. Durk has spent the last five years working with enterprise businesses across 17 different circles including finance, insurance, pharma, hospitality, consumer brands and retail.



The Most Important Element of Defining a Conversational AI Road Map is Automation Prioritization


know what to automate first using pareto chart

The First Step to Building a Conversational AI Roadmap is to Prioritize Automation Efforts Using a Pareto Chart.


The image above is from an actual enterprise brands customer contact center. This is a simple bar chart that prioritizes information on scale. If you have ever worked in a contact center this chart will be very familiar with you.

We will be using this as our example on where to start in prioritization that will bring the greatest impact based on your business goals.


actual results conversational ai roadmap

These are the actual results achieved in the first month of implementing a well structured conversational AI roadmap for their Enterprise. Total of $215,000/month in customer support cost savings.


As you can see in the Pareto chart above the results that can be achieved in just one month are pretty incredible. This is just their contact center.

Their customer support center received 75,000 successful automated resolutions with a savings for each resolution is $2.87 bringing us to a total customer support cost savings of $215,000 a month.

When you do the math this cost savings is pretty significant.

During a crisis like the Covid-19 pandemic where retailers and businesses are struggling to keep their doors open in 2021, this kind of savings cannot be ignored any longer.

In this article we are focusing primarily on the external applications of conversational AI in regards to interactions with your customers on the front end because that’s really where the volume of savings can add up fast and the implementation process is 85%+ automation of customer inquiries out of the box.


What Does Your Customer Support Pareto Chart Look Like?


In the case of your customer support Pareto chart you could be looking looking at something that’s very different from the chart above.

Your goals could be to use automation in an internal application, workflow automation, or making a transaction happen for an employee.

The possibilities to automate processes within and without your organization are literally endless with natural language conversational AI technology.

The Pareto chart that you will be producing for your business depends on your goals.

Your unique goals could be;

  • increased engagement
  • more sales
  • lower cost in the contact centre
  • best in Class customer experience

The good news is thanks to high fidelity omnichannel conversational AI digital workers you can lower cost and increase satisfaction at the same time, something in early automation days we simply couldn’t do.

With advanced natural language customer support platforms like what the engineering team at Linc have developed – lowering costs and increasing customer satisfaction are connected through a seamless process, you really can do both.


Expert Tip:

When Building Your Automation Chart Don’t forget about the customer interactions that never make it into the contact center. For example 60% of customer support conversations end up being pre-purchase most of which would never have reached the contact center. By following the link you can see how Lamps Plus maximized on this crucial customer support data.


When you put your Pareto chart together don’t just put a context Pareto chart together also put a contact center Pareto chart together.

You will need a Pareto chart that that covers the broad range of what customers do with your brand, how they are interacting and on what channels.

You may not have all this information until you actually get your customer into a conversation (this is a feature of the Linc Platform that takes care of this data collection for you) but you need to take your best guess at those numbers.

This is going to greatly impact the direction your conversational AI roadmap is going to take.


How to Prioritize Your Conversational AI Roadmap By Focusing on the Value Quadrant


how to prioritize conversational ai roadmap

Mapping out automation priorities when building a conversational AI roadmap doesn’t have to involve complicated spreadsheets and non relatable data. Focus on simple value as this 4 quadrant chart outlines.


When you start to figure out what should goes on your Pareto chart consider something simple just about value like the example above.

You don’t need complicated spreadsheets, just simple prioritization based on how much automation saves you multiply that times the volume of it along with cost savings per transaction as the chart below describes.


conversational ai roadmap high transaction savings



In the graph above we can see how to break priorities down into four quadrants.The first one is an obvious and definite yes! Right?

What we are seeing is a high transaction savings and a high-volume, definitely going to go onto your Pareto chart.


high support volume low transaction value

High support volume but low transaction value is still in our experience a good idea to add to the Pareto chart when building your conversational AI roadmap.


In the example above we see High-volume but low transaction value.

In our experience high volume usually trumps low transaction value. The benefits to your customer support staff far outweigh not including it as part of your roadmap. If you have high volume even if you have lower transaction savings it is more than likely going to end up going on your chart.


focus on the value of conversational ai in customer support

Don’t hastily count out low volume but high transaction cost savings. Depending on your industry automating these process with conversational AI can still bring incredible customer support savings over time.


Don’t count out low transactions volume with high transaction savings.

We’ve seen a lot of this in sectors like the financial services where the cost of an interaction is otherwise quite high and so even if the volume is low you can still do well in customer support savings and take a huge burden off of your customer support staff.

Even a  $30 or $100 savings per customer support interaction can really make a difference.


is customer value worth investing in conversational ai

Is the value worth investing time and resources in automating that area of your Enterprise?


If you have low volume and low transaction savings it is highly unlikely or it’s going to be a very low priority that you will need to add automation to those aspects of your business.

In this case we would recommend you fill your chart with the other three items and scale from there.

Spend more time there; if you’re going to build a roadmap for automatic conversations you need to make it matter.

The next steps are important in beginning the implementation through to the launch of your conversational AI roadmap.


Does Your Current Live Chat Support Redirect or Resolve Customer Inquiries?


do your customer conversations redirect or resolve inquiries

Do your current customer conversations actually resolve your customers inquiries? Or Just redirect them to find their own answers?



The biggest impact we’ve seen in customer experience automation results is the difference between a redirection and a resolution.

Your live chat conversations shouldn’t tell customers where to go or how to resolve the inquiry.

Your digital workers whether human or AI need to be able to actually resolve the inquiry immediately in real time while in the chat window.

This is crucial, especially in 2021, you need to get the customer requests done.


Evaluate Your Current Conversational AI Customer Support Technology


Make sure your conversational AI chatbots and customer support technology is up to date. If your current platform is not set up to resolve customer inquiries directly from the live chat window this is a sure sign to evaluate the cx technology stack your enterprise is using.

You really need to do this evaluation process and here’s why;


evaluate your conversational ai tech stack

Evaluate your conversational AI tech stack. New technologies have emerged in 2021 that are vital to Enterprise retailers post-Covid-19 pandemic.


Things have changed in regards to AI technological processes and as a result we have more powerful high fidelity digital workers that can improve both consumer-driven conversations AND consumer-directed conversations. In a truly holistic and omnichannel experience.

Instead of fixed dialogue flows that you used to have to build yourself, now the consumer can come in and start the conversation and the dialogue is adaptively created based on where the customer comes in and what the system knows.

You don’t need to anticipate every permutation of your customers interactions anymore the software can do that for you.

In today’s technology, context is not just understood but it can be inferred and conversation memory can be maintained over entire interactions or even between interactions, so when you look at inference think of things like time, or the name of a product, or a combination of the two, inferring an order number, or the use of a term.

A good example is in an apparel company using this technology and a customer referring to a specific gender so you can show that particular type of clothing adaptively.

There is hundreds of different ways that you can infer things and you get to make those choices.

It also makes the conversation for the consumer much more seamless and all of this is done above the dialogue in the AI level in software so you don’t need to build the dialog flow to get it.

With conversational AI technology where it is today your customers get an adaptable and individualized conversation that resolves their inquiry inside the live chat conversation and in real time.


Set Bold Automation Objectives and Expect More From Your Conversations


As we have shown above, the technological stack available today at companies like Linc supports these bold automation endeavors and can deliver the kinds of 1-1 and 2-way conversations your customers have come to expect.

The biggest automation implementation failures that we’ve seen are from not thinking big enough, because the Enterprise thought that what they were doing is impossible or it was too high-risk, so they picked something of little value on their first try and it didn’t succeed or didn’t create enough value.

Do something that’s going to create great results right out the chute!


When Setting Up Your Conversational AI Roadmap Avoid Poor Results By Skipping These Mistakes


conversational ai roadmap avoid these mistakes

Expect more from your customer conversations. In this example we see a typical pre-programmed FAQ type answer. With no conversational flow at all. This does not answer the customers question in a meaningful way.


A mistake we see often in setting up a conversational AI roadmap is treating the website like it is an FAQ.

True automation and customer resolution is not about using FAQ’s or knowledgebases and re-directions.

it’s not that customers don’t ask broad questions requiring a knowledgebase and/or FAQ, these are standard and included in the most basic cx automation platforms.

What we’re addressing is that in a chat your answer can’t be too long, especially on channels like voice assistance or serving multiple channels. An answer like this example above absolutely wouldn’t work on voice assistance.

It’s also already too long for standard chat and customers rarely ask questions for the fun of it, they’re actually trying to do something, so it’s best to lead them to a transaction that they likely want to do.

Customers literally want you to just answer the question they asked. Sometimes even reading their minds and offering the information they’re looking for.

in this example the customer asked a very specific question about returning something they bought yesterday and instead they get three generic questions about return policy.

Not going to lead to a satisfied customer.


The Pitfalls of Using Redirection vs. Customer Inquiry Resolution


dont use live chat as redirect

Redirecting a customer might seem like the simple, fast and easy way to solve their problem. The risk lies in not knowing if their problem was actually resolved, because they have left the chat window and the support resolution process. This can have damaging consequences in brand loyalty if there was an issue with an order that was ‘redirected’ but was never actually ‘resolved’.


Redirects may be simpler, we talked about redirects vs resolution earlier. In the example above you can see a redirection happening. You may even think it’s quicker for the consumer, but you can, should, and the technology exists, to deliver more and your customers are expecting it in 2021.

In the example above the system doesn’t even understand the inquiry with confidence.

It offers multiple links which either give general answers or redirect the customer to a web page instead of resolving the customers inquiry in the chat in real time.

You can lose pretty big on a redirection because you don’t actually know if they went to where they redirected and completed the task, and it’s difficult to correlate the data, it costs significant time and money to add a resource to be able to gather the kind of information and data that you would have otherwise gotten if you kept them inside the structured environment of the chat window.

By keeping them in the chat window using a high fidelity omnichannel digital worker you would get structured data that you can use to analyze what’s going on.

In the case of a redirect answer what if they want to do something else?

it’s higher friction for your customers to get back and continue the conversation and it’s certainly not really possible for them to make a u-turn into a different subject.

Another common mistake we see: Conversations in Live Chat That Do Not Know the Customer


live chat conversation should be natural

The live chat process should feel natural, it should know the person its speaking with through inference and sentiment, which we do not see in this example.


Your Conversational AI Chatbot or digital worker conversation should always know the customer.

The technology exists and is being used by our customers at Linc to create an individualized pre-purchase to post purchase omnichannel conversation.

This chat in the example above should know the customer’s orders and should be able to infer the order number and the role of the digital worker is to reduce customer friction not create more potentially through not understanding the nuances of conversation.

We’re going to show you some examples of how a seamless interaction with a high fidelity omnichannel chatbot, live chat or digital worker can go.


Real Examples of Intelligent Omnichannel High Fidelity Digital Workers Solving Customer Inquiries Seamlessly


This customer support example is recorded from a publicly live website. The customer has asked a very specific question in our example;

“I placed an order earlier today, has it shipped?”

The answer is; “No it hasn’t shipped.”


example high fidelity digital worker resolving problem

Digital worker has ability to send generic answers to questions while getting back to a customer in a meaningful way.


You can get back to the customer and tell them that they can also ask general questions just like any other system and get a general answer back.


example of high fidelity digital worker solving customer inquiries

High fidelity digital workers can give general answers but can also switch topics on the fly at the discretion of the customer.


They can even switch topic;


high fidelity digital worker can switch topics seamlessly

High fidelity digital worker can keep up with customers switching topics and going into a completely different question with ease.


They can turn right around and ask a completely different question about a different thing. “Can I still return the leggings I bought back in April?”


high fidelity digital workers can handle multiple things

High fidelity digital workers can make the right inference and detect sentiment and intent.


This is where the system actually makes the right inference. It takes the combination of leggings and April figures out there was an order number for leggings at that time. It automatically checks for eligibility before actually offering a return.


high fidelity digital workers infer customer sentiment

Linc Conversational AI Powered High fidelity Digital Workers Easily and Accurately Gave Results to the Customer Intuitively and Accurately


Ultimately this entire process is accomplished inside the chat window in real-time, seamlessly and intuitively.

This creates an opportunity for the customer to direct the conversation, this system understands and infers the context to resolve the customer problem AND the whole conversation is adaptable based on where the customer came in and what they said.

You saw in this example above a particular version of a conversational Ai powered digital worker in action, this could be different depending on your unique use case.

This could be used to automate anything available on any channel and the resolution is in real time.


High Fidelity Digital Workers Use Cases in Retail


use cases conversational ai in retail

Conversational AI powered digital workers have endless use cases especially in retail.


in the retailing space you would be amazed what true natural language conversational AI, high fidelity, pre-built, packaged, configurable and extensible digital workers are capable of offering you and your customers.

The key benefit of this method is that you don’t spend months building fixed dialogue flows.

instead of trying to anticipate just some of your customers approaches you get an AI driven dialogue that handles all the permutations, so it’s way lower lift on your part for a much higher fidelity much more intuitive conversation.

Digital workers get turned on with your specific business information not built and as a result you get high fidelity adaptable conversations that are much more pleasing to your customers.


Optimize and Extend Your Automation Capabilities With Structured Data


how to optimize conversational ai structured data

How to optimize conversational ai roadmap with structured data.


We all know that kpi’s and measures of success are important and you have to set these to know you’re on track and when you’re successful. But the most important thing to keep in mind about building a conversational AI roadmap is that it’s a long-term commitment.

Until you actually engage your users in real conversations you don’t know what they want to talk about.

You need analytics to be listening to your customers and reacting to their shifting needs fluidly. We’re not talking about about analytics on intents or utterances.

Consider that things happen and sometimes in a very unanticipated manner.

No one properly anticipated the Covid-19 Pandemic, yet here we are in 2021 venerable brands claiming bankruptcy and many struggling to survive the year. Cutting costs and increasing revenue wherever possible is crucial and automation achieves this out of the box in the first month.

In retail for example;

  • they suddenly needed to come up with new pickup and delivery methods,
  • they had static store hours that suddenly became dynamic,
  • research online suddenly became Uber important and
  • digital overall across virtually every vertical has become much more important.


By most accounts this digital first trend is not going to go backwards.

Just like your company’s website, your conversations are going to change and you need to be listening with analytics.

The automation systems we have available today allow you to simplify this process, they speak as we mentioned earlier in ‘customer intents’ not just in utterances that you need to manually convert to an intent.

Because these systems know hundreds of intent they can automatically do that for you and you can also drill it down to sub-categories, you can adjust dates, and of course you can do all of this to be laid out as trends.

Digital workers can also sense sentiment. 

With these technologies you know your customers experience in the conversation and you can analyze the sentiment and intent, it’s a very powerful way to understand how customers feel about that conversation – good, bad, or indifferent.


‘Think People’ Into Your Conversational AI Roadmap


conversational ai roadmap think people when planning

Crafting your conversational AI roadmap requires deep insight into the people mindset. In all areas of the Enterprise.


For the best conversational AI roadmap you have to think people.

We have never seen a successful automation launch that was complete without them.

The most strategic thinkers in automation will think way less about the simplicity of reducing headcount and way more about how they help their people provide best-in-class customer experience and how automation does the same.

it’s really the interaction between systems and people that is the key.

A few points to consider;

First, you have to know where do your customers or better said where should your customers be able to converse with you?

There are many channels, many touch points and you need to figure out which ones are important to your brand and your customer.


conversational ai roadmap know when to transfer to agent

Unlike the robust high fidelity digital workers at Linc most competing automation platforms are not capable of making these kinds of decisions in real-time, in the live chat window, while your customer is present and asking the questions.


Second, when and how to hand off to a person?

  • Is it immediate?
  • Is it after some automation?
  • When the customer asks?
  • Based on sentiment?
  • All of the above?

These are choices that you need to make.


conversational ai roadmap where to send inquiries

Knowing where to send inquiries is a crucial part of the automation process.


Third, you have to understand ‘where?’ What platforms or group should take the live conversation?

  • Where does this inquiry belong?
  • is it service reps?
  • Social community manager?
  • Sales reps advisor?
  • Designers?
  • Consultants?

Or they could be in a store, office, a home, in a contact centre, you could have a community of online experts or advocates for your brand.

The good news today is you can escalate conversations to any or multiple of these choices based on the intent of the customer and what context they need the request in.

Systems like ours at Linc are designed to pass context in pertinent packages for fast human ingestion instead of a lengthy unread transcript.

This means when the right person or department actually receives the contact from customer support they can immediately be empathetic and have full awareness of the situation which is highly desired by customers.


Think of How Your Going to Implement Conversational AI and Where


implementing your conversational ai roadmap where to send inquiries

implementing your conversational ai roadmap. The use cases are endless.


Implementing a conversational AI roadmap is an area where things have really changed.

We used to talk about minimum viable product because you have to make sure that you give your customers enough to make them interested in conversing with you and it took a lot of significant effort to build even a small number of use cases.

But now all systems can produce large numbers of use cases with much less effort on your part so you can move up the maturity curve very rapidly.

In less than a year you can move from stage 1 all the way into advanced service digital workers in stage 4 where you never could have done that before.


Don’t Wait to Automate With Conversational AI – Time is Money


get started on conversational ai implementation

Don’t wait to automate your customer support. Time really is money in this case.


No roadmap is worth anything if you don’t get started and time really is money in this business.

The financial people know this, but just a reminder you aren’t losing the first month savings you’re losing the last month savings which is much larger so the sooner you start the sooner the cost savings grow.

As we mentioned earlier you get better customer satisfaction not worse through automation.

The automation of customer support has proven time and again to actually deliver better results.

We have data from a recent Linc customer launch that showed a 30% increase in CSAT with conversational automation, why?

Because customers love to resolve things quickly and without having to wait for people, the more you can do that the more they like your brand.

Don’t wait until your systems are interconnected.

CX automation systems can ingest and integrate at specific points and bring the experience together at the customer experience automation level in the back end, so your systems don’t need to be all connected and aligned before you can get started.

Ask the questions and look for those opportunities because they’re there.

Borrow instead of build.

To the technical people out there we know you can build a platform like this, we have seen it, but in our experience it is too hard to keep up and your company will fall behind and pretty quickly.

We strongly recommend that you rely on players that live and die by staying ahead in the conversational AI Innovation game.


Webinar Bonus Material

BONUS: Digital Worker Interview Checklist


conversational ai digital worker interview checklists

Just like you interview human digital workers, AI powered digital workers also need to be interviewed as each brings a wide array of skill sets and abilities to the table. You need to know you are getting the digital worker that suits your needs.


The final step in launching your conversational AI roadmap is to interview your potential digital workers.

We want to leave you with this checklist for consideration of automated conversations.

At Linc our system is built on the notion of pre-built digital workers which are ‘turned on’ and ‘configured’, not built from scratch.

The digital workers do very specific tasks very well in a high fidelity nuanced way and the AI orchestrates the movement between digital workers and manages the common conversational elements like chit chat, restarts, pauses and waits.

Just like “people workers”, “digital workers” should be interviewed too.

This list should help you come up with the right questions to ask your digital workers in the interview process.


Our Shameless Plug of Linc Digital Workers


linc is the trusted conversational ai platform of leading brands retailers

Linc Omnichannel Conversational AI High Fidelity Digital Workers Are Trusted By Over 100+ Leading Enterprise Brands and Retailers.


Linc is the only CX automation solution that is built for Modern Enterprises, Retailers, and Disruptor Brand’s.

The platform engineered by the technology team at Linc rapidly resolves (not just redirects) complex high fidelity use cases through pre-built and extensible digital workers.

Meaning you can extend them without having to build from scratch and the results are outstanding, just ask our customers.

Linc resolves 85% of inquiries across the customer journey out of the box.

We’ve been doing this for awhile, we know how to automate customer experience and how to sell and service anything on the web.

If you have any questions on today’s webinar “crafting your 2021 conversational AI roadmap” or would like to try out Linc CX automation solutions hit the button below and schedule a one on one demo of Linc and audit your conversational AI roadmap today.


Learn More