Advertising and analytics have changed drastically in the past few years, especially in 2020. This series, “The State of Advertising and Analytics,” explores the evolution of media, advertising and privacy, the rapid growth of data analytics in marketing, and the future of marketing and analytics, providing useful tips and value along the way. Read the first and second chapters.
Ten years ago, marketing analytics were a “nice-to-have” that no one understood except tech nerds, and trying to get buy-in from stakeholders was nearly impossible. “We’ve always done it this way,” or “TV/radio/out-of-home has always worked, why change?” were the mantras. Cue the mobile revolution, and suddenly everyone has woken up to a world where if you’re not learning from your web analytics, you’re falling behind.
In the third blog of this series, we’re going to take a look at the three stages of data analytics, how data collection has changed over time, and how companies can win by leveraging marketing analytics.
You’ll shoot your eye out, kid!
Marketing analytics are nothing new. Nielsen, a company hell-bent on going down with the ship, has been analyzing and measuring audiences for over 30 years, providing well-paying marketing teams with insights across demographics, industries and geographies. Although most of us weren’t around for it, who can forget good old pre-television radio, where marketing messages were woven into serial storylines that led people to buy all sorts of products, eventually having them mail-in proof of purchase?
In the cult classic movie A Christmas Story, Ovaltine asked people to buy the product, send back proof, and in exchange, Ovaltine mailed back a prize. The marketing people at Ovaltine would compile these addresses and bombard homes with more marketing, knowing those homes had persuadable kids ready to buy more products.
“A Christmas Story,” or should I say, “A Product Placement Story,” was a prime example of analog marketing analytics.
Smart brands got that kid hooked on everything.
This phase in what we’ll call the “data journey” is called descriptive analytics, which tells someone what happened through the information they’ve gathered. It’s often good enough to help move the sales revenue needle, and it’s pretty much the only form of analytics that existed before computers became popular.
Computers have been around for fifty years, so why now, and why is it important? Well, everything has changed. Data isn’t held by big companies like Nielsen anymore. Facebook, Google, heck, even the average college student is collecting data on spreadsheets and generating insights to help improve their test scores or athletic abilities.
Data has become democratized, and the gatekeepers have been removed. People don’t need a master’s degree in computer science to build charts in order to understand how ads are performing, and people certainly don’t need to shell out cash to big-name media conglomerates for long commitments to place and run an effective marketing campaign.
Time is important. Marketing dollars are important. Efficiency is important, and all three of these concepts can finally converge now that marketing data is free and accessible to everyone.
Marketing data old enough to buy itself a drink
Let’s talk about the rapid maturation of data in the short time it’s been free.
As mentioned previously, measurement and data centered around volume and vanity in the past, and casting a wide net. The “buckshot” approach, hoping one ad would eventually hit, worked for a long time, but it hooked marketers onto vanity metrics to help them pat themselves on the back when things weren’t going well. Reaching a million households means nothing if you can’t drive consistent revenue — just ask pets.com and their darling sock puppet.
Sure, Super Bowl ads reach a lot of people, but can you get those people to buy your stuff?
Enter conversion-based marketing data. Marketers went from hoping people would go to their websites and logging visitor counts at the bottom, to tracking clicks through digital ads and measuring click-through rates, to tracking “thank you” page views and learning exactly how many people purchased goods, all in a matter of 10 years.
The terms “multi-touch attribution” (MTA) and “media mix modeling” (MMM) sprung up seemingly overnight, as marketers began to have enough data to evaluate every touchpoint and channel along the consumer journey. There are now loads of different ways to evaluate the success of a campaign. If a company is only looking at impressions or reach, it’s like they are only eating the icing on a rich, decadent cake filled with their favorite ingredients.
If you like cake, you’ll really like this next part.
Most companies stop here. They’ve set up their audience size, impressions, clicks, “thank you” pages, and mix of channels. They’re seeing page views, getting results, and they can sleep well at night knowing that for every dollar they put in, they’re getting a certain amount back.
However, with the rapid maturation of data also comes the rapid escalation of the amount of data available. Moore’s Law — a 1965 prediction that computing power, information generation, and data would grow exponentially — has led to massive leaps. More data means more analysis, and more analysis leads to whole operations built around that data. Those operations generate more data, and ultimately, sales revenue for companies. The cake we mentioned before? Now it has data layers.
A three-tiered data layer cake
In order to put all that data to good use, marketers can now integrate their favorite marketing platforms with data connectors (like Supermetrics, FiveTran, and Airbyte), store that data in repositories called data warehouses (like BigQuery, RedShift and PostgreSQL) or data lakes (like Azure, Snowflake and Delta Lake), and create machine learning models, training them to help reach goals with tools like TensorFlow, SageMaker and MindsDB.
There’s a platform for every problem, including open-source companies dedicated to helping those with lots of patience build a marketing data stack of their own. Predictive analytics, which makes use of these data pipelines to make predictions based on what’s already happened, is the next progression in a company’s data journey.
Some applications might involve identifying which users prefer to watch videos compared to users who like still images, which can help make creative decisions for campaigns. Models can also be trained to predict behavior, like how many conversions a campaign might see in a given month, or how many it would get with an additional $100.
Your business could be like Paul the Psychic Octopus, hinging on loads of concrete data, which is way more than just tasty mussels.
The final step in the data journey is prescriptive analytics, where your data tells you what you need to do to achieve a certain result, based on previous descriptive data and assumptions about predictive data. At the moment, this is where the industry as a whole has paused.
Humans are still the best prescriptors, which is why we go to doctors and don’t trust WebMD to diagnose our issues. As multitudes of open-source projects move the data analytics industry forward, the next decade is sure to usher in marketers and advertisers leveraging reliable prescriptive analytics in their decision-making processes — and doing so without breaking the bank.
To bring you back from scary, confusing, data-speak land, here’s the final word: Patience.
Rome wasn’t built in a day, and the behemoths we know as FAANG didn’t come to be overnight. Whether you’re just starting out or have been in business for decades, start building a descriptive analytics practice by gathering as much data as possible and applying some math formulas to supercharge your findings.
Predictive analytics will come with time, and your prescription for more conversions isn’t too far off in the future. To fill that gap, agencies like ours are always innovating and iterating to help clients progress in the journey, and we’re always just a few taps or clicks away to get the ball rolling.