#triggered - using intent in marketing campaigns

triggered_cover.jpg

Human beings are predictable. The common initial reaction to this is anger and disgust - how dare someone make such a generalization?! Freedom of choice and speech is the essence of civilized society, and there can't possibly be predictability to that. 

And to some extent, that rebuttal is right. Humans aren't ENTIRELY predictable, but in a lot of ways, we are. Our demographics, geographic location, and lifestyle choices have an incredible influence on how we behave and interact with the world around us. 

The prime example of this is Facebook Advertising. With substantial amounts of data on you, Facebook can sell spaces on your timeline to brands that want you to engage with them. This is already intriguing at a surface level, but where it really gets interesting is with retargeting. 

 My new job has led me to be aggressively retargeted by software companies on my personal Facebook account.

My new job has led me to be aggressively retargeted by software companies on my personal Facebook account.

If you visit a brand's website, they can tag you with a pixel that connects to your Facebook account. Then, when setting up their Facebook Ads, they can retarget you on Facebook with messaging having known that you've already been on their site. I'll dive deeper into ads in another post, but the main concept here is identifying intent

What gets you hooked? 

In digital marketing, there are indicators in every facet of a business that demonstrate the intent of a user. I've gone over this a bit in one of my earlier posts, and how each page on a website has a different purpose for the visitor. The pricing page might be geared to getting an email from or demo scheduled with the visitor, whereas the case study page might be designed to educate the visitor on the product. 

Just as how a visitor that hits your pricing page has a certain type of intent, your customers/users have different levels of intent depending on their interaction with your product. This is where marketing can really shine - instead of guessing at a vague buyer persona, you can use the data you collect to really dig into what your ideal buyer looks like, and how they progress throughout your funnel. 

pizza_guy.jpg

That was a bit of a jump, so I'll walk through a cheesy example (literally) to make it clearer. Imagine you're running a pizza deals + delivery app, and you're trying to figure out what marketing campaigns to run. The goal is essentially to drive more revenue, but where do you start? 

Ask yourself the following questions about your existing users: 

  1. How many days does it take for a user to go from registering for the app to ordering their 1st pizza? What does this order typically include? 
  2. How many days does it take for a user to order a 2nd pizza after their 1st? What do these orders typically include? 
  3. Which users order more than a single pizza? What times are these types of orders most popular? 

Starting with those fairly basic questions, we can start to identify the intent of our users BEYOND just buying a pizza / downloading the app. For example, the following conclusions might be able to be drawn: 

  1. 90% of users order their 1st pizza within 2 hours of downloading the app. The average order is (1) medium pizza. Of those users, 40% used a discount code that gets them free delivery.
  2. Users typically order their 2nd pizza about (2) days after their 1st. Only 15% of users had a 2nd order, but of those users, 60% had used a discount code in the past. The majority of 2nd orders include (1) large pizza and 1lb of wings.  
  3. Users with their credit card attached to the app are 2x more likely to have larger orders. These orders are most popular between 11pm and 2am on Friday and Saturday. 

These insights are all incredibly powerful. Each point above indicates a level of intent from the user, and a hypothesis that can be tested by the marketer in a data-driven way. 

A/B Testing?! 

The gut reaction to hearing the word 'test' in relation to marketing is to suggest an A/B test. Just run two different versions of the campaign, and whichever converts better wins! This is great in theory, but in practice there needs to be more thought put into how the campaigns will work. There should be reasoning behind each test version and a desired outcome + way to track it. 

Given that we're already pretty baked into this pizza example, lets run with it to get some ideas for campaigns to test. See below for a breakdown of the campaign name, goal, and details following the number format above. 

#1: Free Delivery Campaign

  • Purpose: offer free delivery as an in-app push notification while on the checkout page, see if this increases the chance of conversion (completing an order) for users who just signed up.
  • A/B test: can free delivery increase the order value? Try only offering free delivery if they add a side to their order. 
  • Logic: 40% of first users are using a discount code at checkout, meaning that's probably an incentive to download + use the app in the first place, so let's test on everyone else. 

#2: Better with Friends 

  • Methodology: send a push notification with a discount code if user splits with a friend (on the app). See whether this increases order value + conversion in comparison to the core group. 
  • A/B test: split order with or without discount code - see what effect the code has on orders if the user is already prompted to split. 
  • Logic: the order size of pizza and wings is fairly large, so we can assume it's for more than one person. If we test to split the cost, we not only can increase order value but also the amount of people on the app.

#3: Late Night Munchies 

  • Methodology: send a push notification around 10pm to encourage impulse purchases after a night out. Compare campaign conversion rate to typical conversion rate. 
  • A/B test: messaging could be especially interesting here, whether appealing to club/bar go-ers, or people that stayed in for other reasons. 
  • Logic: help suppliers by encouraging purchases before the peak 11pm to 2am times. Delivery usually takes 30-45 minutes, so correct messaging can get users their food on time VS ordering last-minute. 

Conclusion 

It's easy to come up with marketing campaigns based on general assumptions, but it's hard to justify the spend on them or the structure of the test. Looking for user intent through data is the ideal way to setup trigger campaigns that leverage what you already expect your users to do. And while it is a little sad that humans are becoming more and more predictable, at least we will all get discounted pizza! 

Like what you read? Subscribe to my email newsletter below to get the latest articles, top things I'm reading, and more.