Getting Started with Bots:

Your Guide to Creating Virtual Assistants That People Actually Like

10th April 2025

by Jonathan Redsell

Chatbots are transforming how businesses connect with customers and streamline operations. At their best, these digital assistants create seamless experiences, answering questions instantly, processing requests 24/7 and freeing up your human team for more complex interactions. The potential is huge. From boosting customer satisfaction to reducing operational costs.

But, let’s be honest, we’ve all encountered bots that left us feeling frustrated rather than helped. The good news? With the right approach, your bot can be one that users genuinely appreciate and maybe even prefer for certain tasks.

Whether you call them virtual agents, virtual assistants or chatbots, these digital helpers are popping up everywhere. The big question is: How do you create one that delivers real value to both your users and your organisation?

First things first: What’s your bot going to do?

There’s no magic formula for the perfect bot, but to get it right, we recommend asking yourself some basic questions before you dive in:

• What problem is your bot solving?
• How will it make life better for your customers (and your business)?
• What are people already contacting you about that a bot could handle?

If your bot isn’t going to make something easier or better, why build it at all?

The “should we bot it?” matrix

I love the Value-irritant Matrix from Bill Price (Amazon’s former Customer Service VP). It helps you to evaluate use cases for a bot in a simple grid. My take on this is it helps you to put each use case into 4 categories:

• “Please Don’t” – These are things that frustrate both your customers and your business. Don’t automate these headaches – fix the underlying problems first!
• “Make It Simpler” – Your business needs these processes, but customers find them annoying. Focus on making these interactions smoother rather than just automating the pain.
• “Human Touch” – These are the golden interactions that both your organisation and your customers value. Keep these with your human agents when possible.
• “Bot Sweet Spot” – Things customers need but are tedious for your business. This is your automation bullseye! Think balance checks, appointment scheduling, status updates and returns.

Check out the official matrix in our guide: Getting Started With Bots.

Let your data do the talking

Before you build anything, look at what your existing data is telling you, from contact centre stats to website behaviour.

It’s important to consider whether you’re interacting with people who might find technology intimidating. Keeping it simple is generally always good advice.

For voice bots, consider: Will regional accents be an issue? Do you need to offer multiple languages?

Can we actually do this?

This is a big one. Before you promise your bot can change a customer’s address, make sure it can actually:

• Verify who they are
• Access their current details
• Update your systems.

I’ve seen too many bot projects crash and burn because someone promised functionality that the underlying systems couldn’t support. If integration isn’t possible, be honest and direct people to alternatives.

The AI shortcut

“Let’s talk about generative AI for a second. It’s a game-changer for Q&A bots, letting you handle a much wider range of questions without manually writing hundreds of responses. If you don’t have access to this tech, keep your scope tight and focused on your most common questions.”

Jonathan Redsell

Partner Success Manager, Netcall

The secret to great bot scripts: Play pretend

Here’s my favourite trick for writing natural bot dialogue – role-play! Get two people in a room, one plays the customer trying to complete a specific task, the other plays the bot, limited to what the bot would actually know and be able to do.

For chat bots, use Teams or WhatsApp to simulate the experience. For voice bots, do it over the phone or sit back-to-back (no cheating with visual cues!).

Try this with different “types” of customers, including people who don’t know your business well. You’ll be amazed at how many edge cases and natural conversation flows you’ll discover.

Test, fix, repeat

Once your bot is built:

• Get a diverse group of colleagues to try to break it
• Make a clear list of requirements it needs to meet
• Fix what’s not working and enhance what is
• Test again (and again).

Remember, your bot is never really “done.” Keep an eye on what questions trip it up and where people abandon ship. Whether those abandonments are good (they got what they needed quickly) or bad (they gave up in frustration).

Treat your bot like a team member

When done right, your bot becomes an integral part of your customer service team. Think about it like you would a new employee: Give it a clear job description, train it properly and equip it with the right tools. And don’t forget to review its performance regularly and help it to continually grow its knowledge and skills.

By focusing on solving real problems instead of just implementing flashy tech, your bot will become a valuable team member that both your customers and agents appreciate.