3 Important Questions You Have to Ask About AI

If you're just getting started with artificial intelligence and/or machine learning, there are some questions you must ask of your company before you spend money on AI software.

We're as excited as anyone about AI's potential to increase revenue and reduce costs. But, too often, companies spend money on AI technology or pilot projects or consulting without a clear understanding of where they're going or their AI strategy.

At some point, it will likely make a lot of economic sense for your brand to adopt AI tools like a chatbot or a content tool — or hire experts to help you pilot AI.

It might even be time for you to do that now.

But, before you take your next step into the world of AI-powered tools, ask yourself, your colleagues, and your company these questions.

1. Is this really a problem you need AI to solve?

There are literally hundreds of use cases in marketing where AI can help you increase revenue and/or reduce costs.

In fact, we talk to lots of marketers who want to know if AI can solve specific challenges they have.

In some cases, AI can transform their work.

In many other cases, however, their problems aren't caused by a lack of AI. They're caused by a lack of plan for AI.

Imagine you're a B2B marketer who sends emails to a database of 100,000+ contacts. Email is a critical communication and sales channel for you, but your open rates are terrible. So, you want to use AI to write better subject lines for you and boost open rates.

Artificial intelligence might be able to help you solve your open rate problem.

AI might be able to craft subject lines that get your database eager to open every message you send.

Or, your problem might not be subject lines at all.

Maybe you don't segment your audiences, so your content is misaligned with what contacts actually care about.

You could have a poor quality database that hasn't been maintained in some time, so your messages get lost in someone's old or secondary inbox.

You might even have deliverability issues that prevent your carefully crafted emails from ever reaching contacts in the first place.

The point here is:

The problem you think you have may not actually be your problem.

In the case of our beleaguered B2B marketer, maybe the problem is bad subject lines. Or maybe it's database quality or deliverability. It could even be something else entirely.

Smart, accurate identification of your problem is critical before you undertake any AI project. If you try to throw AI at a problem that isn't caused by a lack of AI, you're likely to both fail and spend a lot of money failing.

2. Can AI really do that?

Even when you accurately identify your problems, you need to ask yourself-and others-if AI can really do what you think it might be able to do.

AI is still in its infancy. Tools exist that have formidable capabilities, to be sure. But you need to be careful that you don't oversimplify or overestimate what artificial intelligence can do.

Like any complex technology, AI rarely involves just flicking a switch and watching the results pour in. Many AI tools are highly contextual and require time to work properly.

Let's revisit our B2B marketer in the example above. He decides that, in fact, he needs AI to improve his subject lines at scale. Technology exists to do this, so he starts booking demos and vetting vendors.

But, our B2B marketing friend needs to understand that any solution he selects will likely need a healthy amount of time to learn from his email database and start writing subject lines better than his human team.

Even then, the system will probably need more time to improve, too.

That's why the answer to "Can AI do X?" is usually "It depends." Very few solutions can be implemented completely out of the box on Day 1.

3. Do I have enough of the right data to make this work?

Many companies don't realize the type and amount of data they have can make or break the adoption of an AI tool.

Marketers must ask vendors about the data and the data strategy required to use their solutions.

In some cases, you may need tens or hundreds of thousands of custom data points that you own to benefit from an AI tool.

In other cases, an AI solution may find this data in the public domain (by scraping websites, for instance), so you don't need to possess it yourself.

The difference, however, is crucial.

Some AI tools may be unusable if you don't have enough data. But, if you're a small company, some tools may be very usable if they get their data from outside your organization.