Raise your hand if you’ve already heard about predictive marketing. While it’s certainly not a new concept, predictive marketing is experiencing a boost in prominence thanks to updated technology which has made it more effective—and popular—than ever before.
Just check out the term’s popularity growth from December 2008 through December 2017:
And it’s popularity is expected to go even higher in 2018:
With that in mind, let’s kick things off with a simple, straightforward definition.
What is Predictive Marketing?
The use of data analysis to predict which marketing actions are most likely to succeed based on historical patterns, trends, and outcomes.
You’ve probably used predictive marketing in the past—or some form of it—without labeling it as such. Anytime you use data-driven research to make sales projections, design advertisements, or write a blog post, you’re using predictive marketing.
Here are three pieces of predictive marketing advice you can start using today.
1. Use Predictive Marketing to Improve Reliability
The idea with predictive marketing is to choose future outcomes with a high potential for profit based on the understanding that historical data confirms a reliable degree of performance results.
Basically, marketing actions with the highest chance of success are chosen by researching historical behavior patterns.
- Historical behavior pattern: People leave church at 12:30 pm
- Marketing action: Restaurants promote a lunch special on Facebook starting at 12 pm
To choose reliable marketing actions that will produce successful results you need to understand the 3 aspects of predictive marketing:
- Identify a pattern, trend, or outcome
- Determine the level of reliability
- Make an actionable marketing prediction
Together, these three parts can be used to create accurate forecasts that improve the effectiveness of your marketing activities.
Check out this example to see how all three parts work together:
- Identify a pattern, trend, or outcome: You have a segment of 1,500 customers that visit your website each year during the 3rd week of November to search for a special turkey seasoning.
- Determine the level of reliability: Based on data from the previous 7 years, you find that 17% of customers purchase every year, 71% never purchase, and 12% add turkey seasoning to their shopping cart before abandoning the page.
- Make an actionable marketing prediction: You make two predictions based on the reliability of your customer’s behavioral data. First, you predict that approximately 71% of your 1,500 customers—1,065—will have enough interest in your turkey seasoning to visit your product page next November. Second, you predict that approximately 12% of customers—180—can be classified as highly interested and likely to purchase with the right incentive.
If you don’t have the luxury of reviewing several years worth of historical data, the next best thing to do is look at industry benchmarks, trends, and reports. The more relevant data you have to fall back on, the more trustworthy and accurate your marketing predictions will be.
The accuracy of your predictions depends on the information you use. David Ogilvy once said:
“The more informative your advertising, the more persuasive it will be.”
While Ogilvy was referring to making ads that were highly informative to your audience, the reverse is also true when applied to your marketing process. The more informed you are, the more accurate your marketing predictions will be.
In this case, you’re using the data to inform actions that will dictate success or failure. Which brings us to the next piece of predictive marketing advice.
2. Use Predictive Elements to Boost Your Marketing Results
In 2015, Peep Laja—founder of CXL—was hired as CMO for then-failing apparel retailer Karmaloop.com. The brand had filed for bankruptcy that year and was losing hundreds of thousands of dollars each month. Incredibly, their marketing KPIs were green across the board just 3 months after Laja joined the team.
How did Laja’s team accomplish this? By using data-driven tripwire marketing. There are 3 steps in tripwire marketing:
- Modeling desired customer behavior
- Flagging deviations from that behavior (i.e., the “tripwires”), and then
- Focusing your marketing time and energy on correcting those deviations
In the case of Karmaloop.com, Laja created two segments to help identify the most profitable, high-lifetime value (LTV) customer behavior.
- The first segment was made up of customers who purchased multiple times with low item totals, but almost never returned products.
- The second segment included buyers who purchased once, had low-value items and returned them often.
Laja identified the first segment as profitable with high-LTV, while the second segment was unprofitable with negative-LTV. Further research showed that the first segment made up just 1% of website sessions but accounted for a whopping 43% of revenue.
Long story short? Laja used segment one as Karmaloop.com’s desired customer model and created a variety of marketing actions based on their predictive behavior.
3. Use a Predictive Marketing Model to Create Actionable Items
This is a simple process. Just apply the 3 predictive marketing steps to your current scenario:
- Identify a pattern, trend, or outcome
- Assign a level of reliability
- Make an actionable marketing prediction.
Your actionable marketing prediction is the key to creating a successful model. If it’s off even a little you could wind up wasting a significant amount of time and money on marketing actions that never pay off.
Going back to the example of the 1,500 customers that historically visit your product page during the 3rd week of November, we can see that two predictive marketing models were created:
- 12% of people will abandon their carts
- 71% of people won’t add products to their carts
So, how do you use these models to create actionable items?
Start by clarifying the purpose of your predictions for each model:
- The first model predicts that people who abandoned their carts will be interested enough in the product to purchase with some encouragement.
- The second model predicts that a percentage of people who never added products to their cart could be incentivized to consider purchasing.
Now, you can create a corresponding actionable marketing item for each model.
Let’s say you decide to create two promotions for the third week of November:
- The first promotion incentivizes abandoned cart customers to complete a purchase using a free shipping offer.
- The second retargets people who visited and left your site without adding products to their cart with a 10% discount.
It’s important to note that these promotions can’t be randomly chosen if they’re going to be effective. They must be chosen based on research and historical data.
For example, perhaps you create the first promotion because you know that 61% of people abandon online shopping carts due to extra fees—such as shipping and taxes.
Creating a predictive marketing model will help you uncover your most valuable customer segments and develop actionable items that produce reliable, successful results.
Final Word on Predictive Marketing
Predictive marketing is becoming more prominent thanks to the popularity of marketing automation and its ability to predict and capitalize on market trends. Many marketing tools—including HubSpot, Buffer, and Hootsuite—already use predictive technology to estimate things like the best social media posting times, leads that are most likely to become customers, and which keywords will have the greatest success as a blog topic.
Moving forward, AI will no doubt play a larger role in predictive marketing, especially as it relates to analytics because data as a marketing tool is becoming more important than ever before.
Check out what Growth Tribe has to say about predictive analytics in the video below:
Here’s the main takeaway—use data-driven analytics to discover which marketing actions have the highest probability of succeeding by looking for patterns, trends, and outcomes that are predictable and reliable.
This article was originally published here.
VISIT THE SOURCE ARTICLE
Author: Brian Appleton