‹ back to all posts
If Content Creation Strategies Were People

If Content Creation Strategies Were People

Deciding on the best content creation strategy for your business is difficult.  There are several methodologies meant to tackle the challenge of B2C content, but the high level strategy behind each has unique benefits and drawbacks.  We wanted to put these various methods into a more relatable context - so lets look at B2C Content Creation Strategies as though they were people in your life.

 

One-Size-Fits-All Content - The Telemarketer

You can never control when a telemarketer contacts you, except it's always right as you sit down for dinner.  The telemarketer can't think of you as an individual, because the only information they have is your name, which they often butcher before jumping straight into their scripted pitch.  It's immediately obvious that to them, you're just a number on a list.  However, for every 1000 people they call, they'll find at least one person who's actually unhappy with their long distance phone service.

This is a "one-size-fits-all" experience and B2C enterprises use this approach often because it's the easiest content to produce at scale, despite low conversions and small returns.  Marketing teams create broad, vague content they send to all subscribers or customers at the same time.  Consumers end up glossing over the content in the same way they dismiss the telemarketer's pitch: "No, I'm not interested in a 5% discount site wide, but thank you for calling."

 

Content Swapping & A/B Testing - Your Restaurant Server

Along with bringing you the food and drinks you order, your server's job is to understand their restaurant's menu to suggest items and up-sells that drive a higher tab (and a larger tip).  There are a few things that limit your server's recommendations and the experience they can give you.  First, you're not their only table, and servers can't spend their entire shift understanding exactly what your preferences are.  To recommend items and help customers decide, they'll ask basic questions and return with a pre-prepared statement: "Do you like chicken or fish? Ok, great! My favorite chicken dish is ____."  

Content swapping and A/B testing strategies work in a very similar way, by assigning each user to a segment and swapping out content based on a business rule - your answer to the "chicken" question places you in a segment, and the server provides the corresponding recommendation.  If you had answered differently, they would have had a pre-prepared statement for "fish" ready to go.  

Content swapping works well if you have detailed business rules set but is only effective if there are a few options to recommend.  Imagine if the "chicken or fish" question offered thousands of options instead of just two - you'd never eat!

 

Collaborative Filtering - The Well-Meaning Out of State Relative

Your out of state relatives mean well, and always make sure they send you a present on your birthday -  but the thing is, they're too far removed to keep up with your changing preferences or understand what you (really) like.  So, they take the last thing they remember about you - that you asked for an iTunes gift-card last year - and decide that your next gift will be "music" related.  To avoid repeating the same gift, they recruit the help of your music loving cousins...who recently discovered their passion for "Scottish Pirate Metal" (apparently a real thing).  Now, you have an obscure album full of incredibly loud sea shanties that you don't know what to do with.

Collaborative filtering takes your most recent action and places you into a segment - in this case, "music."  From there, content is automatically generated for the entire segment based on the information received from the segment's majority, no matter how little you have in common other than your most recent action.  While collaborative filtering works for contextualization, generating specific content that's consistently relevant isn't possible.  It's also difficult to jump segments once you're placed in one - so be prepared, you'll be listening to Scottish Pirate Metal with your cousins for, possibly, forever.

 

AI-Powered Predictive Content - Your BFF

You tell your best friend everything, and they understand what you like and dislike across the board: your favorite foods, your aesthetic preferences, the car you drive, your favorite books, where you work, your fitness goals - they understand who you are as an individual.  Because of that deep understanding, they're able to dramatically improve your experience in most things.  When you go out to dinner, they make sure the place is vegan friendly first (you're vegan in this analogy).  They see on Facebook that your car is in the shop, so they offer to drive you to work.  When they suggest a movie they think you'll like or music they think you'll enjoy, you're excited to watch/listen (unlike Scottish Pirate Metal).

AI-Powered Predictive Content utilizes artificial intelligence to fully understand a customer's preference in semantic content attributes, based on every user impression.  This creates a deep relationship with each individual consumer, and gives B2C enterprises the ability to predict and automate the creation of the best content for each customer.  Content created with this method isn't constrained to specific content types either; AI-powered predictive content can surface the best content across differing categories and creative types - just like your bestie can suggest your new favorite music or restaurant.

 

While each of these methods have their benefits and drawbacks, understanding how they work is important as the need for scalable B2C content grows.  Out of these examples, who would you trust to run your business's content creation?

Go Beyond Personalization To Predictions