It seems like every marketer out there is trying to get ahead of the pack in the AI space. Most start by reading articles by industry luminaries, paying big bucks for AI training systems, and watching data science YouTube videos for inspiration. This is extremely lucrative for many industry experts, but also expensive, and potentially risky. So should I, and other marketers, get ahead of the game, and start building AI solutions for my business?
Ultimately, this is a question for your business. When should you hire a data scientist for your marketing operations? Consider a few factors:
What does your marketing technology look like?
In my experience, most brands begin by creating content using traditional data sources, such as a blog, e-mail campaigns, and small collections of analytical dashboards. However, companies that are investing a significant amount of money in their marketing tech—whether that’s search analytics or data analytics for a real-time marketing platform—are more likely to start using AI in a big way. If you see AI as an opportunity to replace employees, consider this: artificial intelligence currently costs between $80,000 to $100,000 for a data scientist, depending on how much research and development is needed.
How much is your data or analytics team already costing you?
It might sound like a difficult question, but keep in mind that some companies spend $300,000 per person on their marketing technology. Cost effective solutions exist in the form of the likes of Grid Dynamic’s Digital Transformation Solutions. Otherwise if this is your team, that’s potentially $3 million per year, or around 20 percent of your business budget. In other words, your data team might already be costing you a substantial amount of money, so it might be worth breaking it up to pay for one person’s salary, rather than an entire team.
Can you afford to spend an additional $3 million per year on something you don’t understand?
Perhaps you’re a small company. In this case, you’ll have to try, test, and adjust your analytics methods until you get your AI to work. Start small, and do it in small batches. Start by using your AI to compare one product over another. Once you’ve proved your solution works, your AI should also be trained to make predictions for new products in the market. You can then start to train your machine with a large amount of data, using a small batch size. In the end, this process will be a learning process for your AI, and it will probably break down at some point. In this case, your team will need to keep training your AI over and over again, creating a path for the machine to learn, and determine which products are most profitable for your business.
Building AI isn’t as hard as you might think. But you need to start with simple analytics. If you build it, you’ll be able to reap the benefits and build more sophisticated machine learning. And with AI, the most important part is understanding what you’re building, and using it to reach your business goals. So, if you’re already developing marketing automation platforms, and you’re spending lots of money to build advanced machine learning solutions, go ahead and take the plunge, and do it now. Just make sure you understand your business first and find the technology to support your specific goals.