What You Must Ask to Make Your AI Project A Success

What You Must Ask to Make Your AI Project A Success

SnatchBot TeamSnatchBot Team, 08/12/2019

What You Must Ask to Make Your AI Project A Success

 

Artificial Intelligence tools and services are fast becoming a part of the business landscape. Companies of any size can create a chatbot, use an AI analytics tool or adopt a service that uses AI in some way, but before doing so, you need to ask a few key questions about the benefits and expectations.

When it comes to marketing or customer experience, plenty of businesses are jumping on the AI bandwagon. The benefits seem obvious enough, greater automation but with more detail in messaging and a personalised view of the customer. Customer support teams can use chatbots to automate a high quantity of their engagements. 

But before looking at the gold at the end of the AI rainbow, companies need to figure out what their plan is, as with any type of digital switchover or transformation. Here is a list of questions that you need answering before getting into the exciting world of AI for business.

 

Who is working on the project?

 

Small businesses can point a finger at one or two people and give them the job of delivering a digital AI project, if they are keen and suitable for the task. For larger companies, there needs to be representation from a range of departments. Whatever the size, give the team the resources, support and time to build a successful project. If no one has the expertise then getting outside help in can give a kickstart to any project.

With any AI project there needs to be a voice that represents the end users or customers, otherwise, it can end up lopsided and not benefitting that key audience.

 

Where do we start?

 

With a team in place, they need an objective, and - in today’s fast delivery climate - a way to show progress. With AI most companies need to demonstrate and prove the technology before adopting it full scale. Finding a few quick-win examples or use cases within the company can help.

There are plenty of existing AI examples from other markets or rivals you can use as a template. If there is only one goal “a chatbot for our customer support” or “an AI to crunch down our factory data” then work toward that goal but build it in stages that demonstrate viability and provide benefits in steps, rather than an all-or-nothing launch.

Build a step-by-step plan that gives you time to overcome technology, personnel or other barriers and factor them in.

 

How do we prove success?

 

Many businesses avoid unnecessary data like the plague, there is enough of the vital kind around to keep most people busy. Even so, measuring metrics around how you used to do things is essential to see if the new way is an improvement. Measure the time and revenue spent on existing projects you plan to replace.

Then, you can see the return on investment from the new process. If the return is lagging, the project should be given time to improve and prove its worth.

 

How do we get the business to love AI?

 

Lots of people have a fear of AI, based on tabloid fodder headlines and a lack of detail or knowledge. When introducing AI projects to the company, involve everyone who will be affected, and use demos or in-house tests to show the difference they can make.

If there are business leaders, budget-holders or other blocks to progress, using success stories and the logic of numbers should bring them around to supporting a project. If there is a total Luddite in the business, engage and discuss their issues, rather than trying to work around them.

 

Do I use our customers as guinea pigs?

 

In-house testing is essential for any AI project, and training with chatbot or deep learning data is key for larger projects. But there comes a time when you need to let actual users or customers try it out. The key to success here is to prime them with enough information to understand the benefits of any chatbot or AI tool, and have workarounds for any known weaknesses.

AI chatbots can address a wide range of customer questions, but there are bound to be a few outliers. Here, you need agents on standby to step in and help the customer, and also to provide data for the team to update their bots.

If possible, make the bot available to a selection of customers first to get feedback and provide live reporting of any AI service to see where it struggles with input.

 

What happens after day one?

 

Traditional IT projects were rolled out and if they worked everyone went to the pub, then moved on to the next project. With today’s AI-powered processes, there is always room for improvement and new features.

Once a bot or AI has proven itself, business leaders or workers will want them to do more tasks, while other departments might fancy a bot of their own. Those with the AI experience need to keep control of evolving or future projects and deliver new versions based on the proven route to success.

Large businesses can struggle with departments using their own tools for AI, creating silos or incompatible metrics that can cause chaos within a business. The advantage of a small team is they keep control, stay on top of new developments and have a wider view of what is possible (and what is not) across the business.

There are way more than five questions that need answering to get any bot or AI project off the ground, but starting here will set the basics and provide the realization that AI is a serious business tool, with major benefits and not a novelty project for a company to try.

If the answer to any of these or other questions from a leader is “because a rival is doing it,” let them think long and hard about the real reasons why a project is needed and how it can succeed.