What's The Big Deal With AI Anyways?

Data Strategy

May 18

A Pragmatic Approach To AI

Artificial intelligence as a topic has been steadily climbing in popularity on Google Search Trends since November 2022, and recently has exploded with the success of ChatGPT.

Google Search Trends on AI for the last five years

At Shoptalk 2023, AI was one of the buzziest terms - with new AI startups in every corner and every big name brand talking up their AI strategy on the big stages. It’s one of the hottest recurring topics on Linkedin too. Every major company is clamoring to tell the world what their buzzy AI strategy is.

So what’s the big deal anyways?

Should we all be bowing down to our AI overlords? Will marketers be out of a job?

This article will explore Insight Lime’s opinion on what AI means for marketing and analytics - and what you should consider as a business leader in the next few months and for the years to come.

Some Truths To Consider About AI

As anyone with a nose knows, trends come and go. There is often a new topic that we are all excited (or worried) about that fills up our days (Y2K anyone?). You also may know as a seasoned practitioner that it can take some time to slice through the noise and come up with your own informed opinion about a topic…and that exercise is incredibly worthwhile.

Without taking some time to understand a new trend or issue, it will be harder for you to advocate in your organization when the CEO asks you “What are we doing with AI that we can tell the board of directors about?”

So before we get into whether AI is something your company should pursue, we should address some truths about data and AI that are worth noting (and not ignoring!) before you start your decision-making process.

Truth #1: You still can’t create good machine learning without good quality data

(Crap in, crap out) If you are struggling with data quality and observability, you’re not going to be able to build an AI model that blows your current activities out of the water. Why? Because you will be training a model on bad information.

In addition, if you take a look at your current in-house data science and analytics resources and find them woefully lacking - you should be focusing on the basics. You’ll get so much more out of investing in smart people that are looking to solve your company’s biggest problems than you would out of an AI chat bot for your company or exploring replacing copywriters with AI.

The 10/90 rule can guide you here - in short terms you should be spending 10% of your budget on tools and 90% on people. If you are planning on buying tools for AI, check this number. If you’re planning on using your current people resources to develop AI of your own, check to make sure you have at least close to this balance.

Truth #2: Some tools and applications of AI are better than others’

Artificial intelligence is only as smart (and as worldly) as the people who created it. And, there are some applications that are more useful than others. Here are some applications that I believe could be incredibly useful:

Automating data cleansing - data cleansing takes up a huge amount of data scientist’s time and it isn’t where their brilliance shines through. Finding ways to automate and outsource this to AI could be a brilliant way to improve data quality and save time for your most valuable resources - people.
Anomaly detection - While we might be slapping the fancy term “AI” onto this, anomaly detection in digital analytics has been around for years. Adobe Analytics has had very strong features around anomaly detection in data for a long time. This is another one where it can save analyst time and help you catch problems faster.
Improving Product Recommendations - Like analytics anomaly detection, product recommendation engines have been around a long time. But if you think about brands like Stitch Fix, they have built “AI” or machine learning into their own products to the core of what they do, even though they are a styling & eCommerce brand at heart. This is an incredible way to use data to improve the final output of your company.

If we were exploring non-digital business applications there are many that are incredible examples of AI that could truly transform the world for the better - think climate change, hospital operations, and more.

However, with every good application there are quite a few that are not as helpful and could be a waste of investment for your business. Being discerning about this is going to help you keep a hold of your budgets while you explore (or choose not to explore) AI. Every time there is a shift in the industry there will be companies looking to profit from it. There’s nothing inherently wrong with that, but at times this means there are solutions being created for problems that aren’t that important - or aren’t the right AI to fix the problem.

Here are some things we think AI won’t be good for:

Creating SEO content - Any time a “hack” for SEO content comes up, we almost always recommend steering clear. Why? The only way to win at SEO is to create genuine, valuable content that speaks to your target customer. Having AI crawl your competitors' site and puke up piles of similar content to rank higher won’t provide value to your clients. And Google spends much of their time adjusting their own algorithms to reward high quality content and punish those trying to game the system. Essentially, the takeaway is if you aren’t creating the best piece of content on a topic, don’t bother creating it.
Writing marketing copy - This is an example that keeps coming up and you may have seen this on Linkedin. “Create your emails with AI and save hours of copywriter work!” While this can sound absolutely fanciful, and many of these tools tout that they are able to intelligently predict what your ideal customers want to hear in their copy, it’s another one where you can likely get AI to write in your tone of voice about a topic, but it will be missing something - real human opinion. And, it’s not a job that needs to be automated. There is an art to good copywriting and AI is never going to be as creative and original as an expert copywriter.
Being an analyst - While anomaly detection is great, and there are many other enhancements that can be made to how we analyze data, you will be hard pressed to replace your analysts with AI. Why? Being a good analyst isn’t about crunching numbers, it’s about bringing humanity to them. A brilliant analyst can connect the dots between what the data is saying, the context of the product, and the nuances of business to provide information that can be a catalyst for change. AI will not be able to negotiate all the complexities of organizational politics to make a difference.

Truth #3: If you don’t already have a data leader, don’t go looking for a Chief AI Officer

Some large companies (and small ones too) are hiring CAIOs. And data careers are especially prone to trends - as soon as a new one pops up like the rise of data science, everyone has shifted to wanting to hire someone with a “data scientist” title vs an analyst. However, if you don’t have the structures within your organization to support these very senior titles you will be setting this executive up to fail.

Before the CAIO, consider elevating a Chief Data Officer or a Chief Analytics Officer to head up the data efforts across the company while sitting on the leadership team. If the closest you have to a data person touching leadership is a director of analytics, you are likely not at a maturity level where you can get the most out of an AI specific role.

Think about it this way - you need the building blocks of data before you can stand on that foundation with advanced analytics & AI.

While the above visual won’t always be a linear path, it’s best to consider that if you’re missing that data foundation or data leadership, it won’t matter how fancy the AI tool is that you want to use - you won’t have the people to tell it what to do. Things like Chat GPT are essentially advanced search engines that do some of the work for you. The better you are at using those tools (and understanding them) the more you’ll get out of the tool.

Truth #4: AI will be as biased as the people who built it

This is a truth that can be scarier than some of the others. Especially when you think about applications like health care, where bias in care can impact someone’s longevity and well-being.

AI isn’t smarter than us. It isn’t more pragmatic. It’s what we make of it, and much of the time there are only a few demographics contributing to the world of AI. If we’re trying to market to women of color, for example, we could be risking getting it wrong (all the time, regardless of AI) if we don’t have people who look like our target audience on our marketing and data teams.

Conclusion

So what’s the big deal with AI? It’s the next evolution in how we can use and train data to help us do a variety of things as humans.

While it isn’t as sexy, being laser-focused on your business goals & focusing your analysis on answering your most pressing business questions will likely have a higher ROI than AI. If you’ve exhausted all other possibilities of ways to use your data, then you’re ready to push the envelope with AI.

If you’re curious about how to improve your data maturity as a company or want advice on how to build a successful analytics organization, Let’s Talk.

Resources you might like:

Let’s get squeezing!

Join our founder on Linkedin and hear thought-leadership on marketing analytics and analytics consulting.

Featured

Sep 18, 2023

Data Analytics Consulting: How Much Does It Cost To Hire An Analytics Consultant

Sep 18, 2023

Many major organizations rely on consultants to fill a gap - and you could be wondering if it’s worth it. The hourly price tag on analytics consultants can vary anywhere from $120-$1000 an hour depending on the seniority and specific expertise. So when should you consider bringing in outside help, and when does it make sense to begin building out your own team?

Sep 18, 2023

Jul 19, 2023

Your GA4 Migration is Finished - Now What?

Jul 19, 2023

The cutoff for migrating to GA4 has come and passed. We’ve outlined how to get more value from your data and what you should focus on now.