Here’s the deal. Marketers spend a ton of money and effort to get customers to buy. Their audiences are usually a mix of people who have and haven’t purchased from the company before. Countless hours go into sending emails and direct mailers, plus creating ads on platforms ranging from social media channels to radio.
A problem with this blanket approach is you don’t know who’s most likely to convert. Essentially, you’re wasting resources on audience segments with little chance of making a purchase. The effectiveness of your efforts is determined after the fact, with the team going back to the drawing board each time.
While the group debates how to make the messaging or ad placement better, there’s a key component missing. You may not be targeting the people most likely to buy. Custom audience models help solve the problem by examining variables such as website visits and purchase frequency. By doing so, custom audience models can predict who you should target and why. Here are some different ways to create these models so you can boost the team’s results.
Table of Contents
Use Predefined Metrics
Once you start using custom audience models, you may hear others reference them by various names. Propensity models, custom intent audiences, and predictive audiences are the main terms you’ll find. What all these names have in common is an underlying formula that relies on predefined metrics to predict a likely outcome.
Examples include churn and purchase probability, in addition to predicted revenue. Churn probability foretells which members of your audience are most likely to leave the brand. Purchase probability says who’s going to convert. Predicted revenue is kind of like customer lifetime value. The metric estimates how much revenue an active customer will generate.
Each of these metrics can be further refined. You can use specific time windows, such as seven or twenty-eight days. Say you want to know who’s most likely to leave the brand within the next seven days. Perhaps you’d like to predict revenue for the next month from your most active list of subscribers. Predictive audience models allow you to set these parameters so you can target segments with the messages they need to hear.
Focus on Interests and Behaviors
You might think you’re reaching a good portion of your targeted audience when you launch a digital campaign. After all, you’ve worked hard on every aspect, from the research to the design. However, statistics say the odds of reaching your audience are against you. When social media ads were analyzed, over 80% of the ad sets had saturation rates lower than 5%.
Yes, that’s right. The vast majority of social media ads didn’t even reach 5% of the intended targets. Plus, only 10% of social media ad sets reached over 12% of targeted audiences. If this doesn’t make you rethink your strategy, nothing else will. These stats show a more refined approach is needed if you want to reach most of your audience.
Creating custom models that take into account interests and behaviors helps identify who’s most likely to engage with your brand. For instance, your model could use keywords related to interests matching your products. Your model would capture audience members who search for and browse online content with those keywords. To put it more precisely, an athletics apparel company would build a custom model using keywords like physical fitness and outdoor exercise.
Separate Repeat Buyers From First-Time Customers
You may already distinguish between prospects and existing customers. Naturally, your prospect list contains leads you’re trying to convert into first-time customers. These leads have expressed interest in your company by filling out a form, attending an event, or contacting a sales rep. Your customer list includes people who have purchased from you before. Some of those customers may be active, and others might not be.
Say you decide to offer all these audience members the same discount on their next purchase. Your team sends out an email blast advertising the discount, hoping to increase sales for the month. Will this email produce the results the team wants? What’s likely to happen is varied outcomes between different segments.
At best, existing customers who aren’t likely to purchase will wonder why they’re getting the email. At worst, they’ll get upset over a discount they weren’t offered before on the product or service they already have. Leads who are not ready to purchase won’t suddenly become motivated by the discount and possibly get annoyed. And those most likely to buy may not need the discount as a reason to convert.
Instead, you could build separate custom audience models based on existing customers and leads most likely to purchase within seven days. Your messaging and incentives would differ based on each audience’s triggers and needs. Plus, you’d be targeting leads and current clients who you stand a higher chance of winning over.
Predicting Purchase Intent With Custom Audience Models
Reaching target markets in crowded online content spaces is a tough enough job. Why make it more difficult by using broad-based techniques? While you might see some glimmers of hope, you’re more likely to set yourself up for failure.
Your brand is better off building custom audience models that will increase the effectiveness of your marketing efforts. You’ll be reaching out to people who intend to buy from your brand soon. Once you fine-tune your metrics, audience behaviors, and segments, you’ll have predictive models for your team to leverage.