Chatbots 101: a guide for small businesses

Anna Roberts

Jan 2018 ⋅ 7 min read


Depending on who you ask, the word 'chatbot' either encapsulates everything that's wrong about brand communication, or is the future of all our brand interactions.

We wondered who was right, so we did the research so that you don't have to.

Are chatbots a legitimate marketing and customer service tool for small businesses, or the latest marketing fad that isn't worth bothering with?

What is a chatbot?

A chatbot is software designed to 'speak' to humans in order to provide relevant information or assist with a task. Chatbots usually speak to humans online using text (and sometimes GIFs or images).

You've probably used a chatbot before, perhaps without even knowing it. You'll find them on messaging apps like Messenger and Telegram, and some brand websites.

Generally, chatbots fall into these categories:

  • Information providers. These bots send the user updates on a specific topic when requested, such as weather and news updates. (Example: Poncho)
  • Shopping and recommendations. These bots recommend products to a user based on their inputs, and often allow the user to proceed with the purchase from within the messaging interface itself. (Example: DOM the Pizza Bot)
  • Internal project management chatbots. Instead of providing a service to customers, these chatbots can be used by individuals, teams, or departments to streamline internal processes such as arranging meetings, distributing documents, or setting reminders. Most of these bots integrate with Slack, or other chat apps. (Example: Meekan Scheduling for Slack)
  • Customer service bots. These bots attempt to answer customer queries and, if necessary, pass the conversation onto a human customer support rep. Some customer service bots can provide information related to individual accounts (such as order progress or account balances), or general help and support. (Example: Intercom's Operator)

Chatbot IQs

Bots can also be sorted by their 'intelligence'. The simplest bots are rules-based: developers write a set of rules that determine the commands or queries the bot is able to respond to.

In other words, imagine a simple weather chatbot that's designed only to tell the user what the weather will be like tomorrow, at their current location, when requested.

The developer sets rules based on user input.

An input of "tell me the weather for tomorrow" should lead to an output containing tomorrow's weather forecast.

It's easy to develop a chatbot that responds accordingly.

But the developer also needs to think about the various ways users will word this request.

"Will it be sunny tomorrow?"

"What's tomorrow's weather forecast?"

"Give me the weather."

"Do I need to wear a coat tomorrow?"

...and so on.

These rules are usually added to the bot manually, with new additions or adjustments made as more user data starts becomes available.

Generally, you can map out a rules-based chatbot's 'script' with a flow chart, showing the various paths a user might take, and the inputs required to get there.

Natural language processing (NLP) is the next step up in intelligence. NLP bots have some awareness of sentence structure as well as keywords, helping a bot understand and respond to a wider range of inputs.

Bots with NLP functions might be able to notice typos, use of idioms, or a user's 'intent' (what they want out of the chatbot interaction).

These bots may be smart, but they have their limitations. They don't 'learn' as they go along — they can only become smarter when a human adjusts their code.

The most intelligent level of chatbots use artificial intelligence (AI). These bots learn from conversations and adjust their output based on context or the user's mood or preferences. Some chatbots contain elements of AI, but this technology is still in its infancy and it's not accessible to most small businesses.

How can chatbots benefit my business?


Let's be honest here, the vast majority of businesses don't need chatbots — particularly customer-facing chatbots. But there's no doubt that chatbots hold significant benefits for certain companies.

Customer support chatbots can be incredibly helpful for businesses dealing with a high volume of support queries, particularly when the requests are often similar. If these queries are general instead of related to a specific account, the chatbot will be particularly easy to build.

A customer support chatbot can:

  • Process and resolve many support tickets simultaneously
  • Save labour costs
  • Provide 24/7 support
  • Link to further information in your knowledge centre
  • Resolve customer queries with no need for the customer to switch support channels (website, phone call etc)

If you're more interested in using a chatbot for shopping recommendations or even for processing orders, here are the benefits:

  • Increase quantity of mobile orders (messaging app usage has surpassed traditional social media use)
  • Prevent 'choice paralysis' by recommending products based on price, colour, and other preferences
  • Provide a convenient, alternative route for purchasing products, often without needing to leave the messaging app
  • Offer customers a personalised shopping experience

Using chatbots for internal management also hold many benefits — and because many of these bots are designed to fit seamlessly with apps you already use, there's very little hassle involved in using them. There's a Slack bot to help with almost every admin task you can think of, as well as bots that focus on other aspects of life at work.

Here's what you could gain:

  • Significantly reduce miscommunication
  • Save time on paperwork and business processes
  • Increase team morale and engagement
  • Build a stronger team

Of course, there's such variety in Slack chatbots that we can't list every benefit here.


How do I get started with chatbots?

map showing journey plan

Here, we're going to focus on customer-facing bots, because if you're looking to use internal bots, your best bet is to make use of existing chatbots instead of making your own. Head to your chat software's apps directory to get started — here's Slack's.

Your first decision: What do you want your chatbot to actually do?

Think about the current issues you have with your brand's buying experience or customer support.

Is there a role that a bot could play?

Start simple. Think of one thing you'd like your chatbot to be able to achieve — for example, being able to recommend dresses based on a colour and price input from a customer.

Next, is there a link between this chatbot's function and a business goal?

Will your chatbot boost revenue, loyalty or another metric you're looking to target?

Your second decision is to choose how much coding you'd like to do.

If your first thought was 'none', that's fine. There are plenty of tools to help you put together your own chatbot without needing to be a programmer.

And there's one more choice to make before you start: which platform will your bot operate on? Facebook's Messenger app will almost certainly be the best choice, but you might want to consider Skype, Telegram, Kik, Viber or your own website.

If you're a developer, you'll probably want to check out bot development frameworks like (for Messenger apps) and the Microsoft Bot Framework to start off.

If you'd rather not code, you need to choose a chatbot builder to plan your bot. Consider the following tools for Messenger bots:

These tools will help you create rules-based chatbots that may have some natural language processing capabilities — but you won't be able to build a more intelligent bot without getting stuck into the code yourself (or paying someone else to).

If you have some programming knowledge, look out for chatbot templates to edit to meet your needs, or dig into Facebook's Messenger documentation.

We can't advise you how to use a particular chatbot building tool, but there are some best practices that apply to most chatbots:

  • Always make the next step clear. From the very first 'Get Started' command to the closing stages of the interaction, ensure it's completely clear to the user how they're meant to proceed. For example, if you're providing two options for users to choose between, clarify whether they need to tap their preferred option, or type it in. Chatbot users aren't going to be 100% engaged with your bot — they'll probably flick back and forth between your message and other conversations and apps.
  • When mapping your bot, think about user intent to help guide the flow of your chatbot. Once you've figured out intent, don't confuse users by backtracking or asking questions that are irrelevant to their intent.
  • Plan for failure as much as success. If your bot doesn't understand an input, be sure to add some responses that either politely ask the user to rephrase their query, revert back to a previous state, or suggest an alternative means of contacting your business.
  • Be concise. Remember that most chatbot users will be using Messenger on their phones, so keep questions and responses short so that they're easy to view and scroll through on a smartphone.
  • Keep your tone 'on brand'. Most chatbots will have a conversational tone, but make sure your bot sounds like the extension of your brand that it is, rather than anything insincere.

Once your chatbot is live and happily answering questions and providing information on behalf of your business, don't abandon it.

Non-AI chatbots are only as smart as the information you provide and the rules you set, so it's always worth looking through transcripts and analytics to see where conversations start to go awry.

In particular, look for instances of bots failing to understand users, as well as any query types (intents) that your bot isn't very good at dealing with. Make adjustments to improve the user experience on a regular basis, pulling the bot offline if necessary while you fix any malfunctions.

Final thoughts

Chatbots are still an emerging technology — and the novelty factor means that your customers are very likely to give your chatbot a try.

Of course, you'll need to provide your users with a satisfying chatbot experience if they're to use it again in the future (and tell their friends about it), but so long as you understand your customers' needs and how to fulfill them, you stand a good chance of building an effective chatbot.

You don't need to be a developer or work in a big business to create a chatbot to help customers or manage projects. Sure, chatbots won't be necessary for every brand, but don't just dismiss them because you're a small business!

Have any favourite examples of business chatbots? Give them a shout out in the comments section below!