Getting started with AI in marketing

Posted by Chris on 21st October 2019 (about a 2 minute read).

For a few years now we’ve been hearing all about how AI will transform marketing. But so far, not so much.

It’s not that it won’t happen, or can’t happen. It can happen right now.

People who can design and run AI systems are still rare commodities. Given the choice (and pay packet) between creating self-driving cars, and optimising your marketing, which will they choose?

So if you want AI for your marketing, you’ll have to start yourself.
You need two things: data, and questions.

Don’t think of AI as some mythical beast that works magic. It is simply the ability to perform lots of mathematical calculations on lots of data, very quickly.

When we think of self driving cars, it sounds magical, but turns out to be lots of data and questions. Can you identify a pedestrian in the images of the video stream? Can you accurately measure the distance to the person? What angle should the wheels turn to avoid a collision with the person.

Step 1: Get data permalink

Most AI needs lots of data (and if your data is small enough that you can analyse it yourself, then you don’t need an AI tool). Can you gather your sales records, customer service records, call centre logs, website traffic and analytics, social media historical data, email marketing data, direct mail, media spends?

Step 2: Ask Questions permalink

So in your data, what are the questions that you want to answer? Do you want to understand relationships between spends and sales? campaigns and website traffic? Forecast sales? Explain customer buying journeys. Identify trends in your social media?

Step 3: Implement permalink

The first 2 steps will give you a brief that lets you plan the implementation. You can go to an agency, hire a freelancer, or build an in-house team, or a combination of roles. Getting the work done is a whole different topic, but you are now in a position to get started, which is usually the hardest step.