Today, we’re turning our attention to Paid Social.
- Good for something Paid Social
- Now with added smarts!
- Performance. Liftoff! We have liftoff!
- Finding Incrementality
- Predictive Acquisition using CLV
- How all this differs (+ bonus analogy)
Before we get going, a bit of housekeeping on this series:
Any activation recipe we cover isn’t limited to the channel being covered in the forthcoming chapters, but we’ll attempt to use the most common recipes per channel for a punchier read.
In the previous chapter we discussed the difference between a marketing recipe and a customer recipe, we’ll stop using these distinctions and simply trust that you see the potential to deliver on both.
What is Paid Social good for?
Umm, can o’ worms, Dean!
Disregarding the moral or political debate for a second, let’s simply agree that what Facebook & Instagram is accomplished at is allowing you to place an ad in front of a consumer mid-scroll, to influence desired behaviour or outcome.
It is therefore so, that we’re able to deploy a wide range of goal-oriented strategies; from acquiring new customers, to stimulating and converting existing ones.
Yeah, how do we do that with a bit of smarts?
More often than not we’re tasked from the off with sending performance metrics sky-high. This demonstrates that we’re good at making customer predictions (spoiler: we are) and they in turn carry a superior conversion probability.
The following visual represents a Programmai recipe that selects for consumers with a high Purchase Propensity (e.g. high conversion probability in the next 14 days) and syncs them to Facebook for targeting in order to achieve maximum ROAS and CPA efficiency.
What’s clear is we’re ignoring consumers outside of this selection, isolating those with the highest conversion probability to maximise the efficiency of ad spend. Bang for buck.
This produces an increase in return on ad spend, while sending the cost of acquisition six feet under — cutting out the perceived fat in our recipe for this goal.
What about Incrementality?
…said the sophisticated marketer!
Right, so you’re bored of showing off how powerful your new predictions are — I get it. Let’s start to work our way down the conversion probability ranking to find greater incrementality in your target audience.
Programmai is both well-equipped to manage a control group for you, as well as make calculated budget recommendations based on the lower conversion probability and the investment it’ll take to nudge these consumers into the higher predictive lists.
We’ve got you. This way we can report on something other than last-click, given our desire to nudge propensity, while still using the predictive model to help keep our ROAS and CPA optimal for this type of goal.
What do we want, New Customers. When do we want them, ALWAYS!
Loud and clear, Mr. Foghorn.
What I’m about to show you next requires you to take a mini leap of faith with me. You in?
By now you’re starting to believe Programmai is capable of predicting future conversions, customer by customer. Cool, let’s extend our look-window forward 12 months.
Recall our Customer Lifetime Value model can predict sales & revenue by customer, annually. When we covered this in detail, we revealed that what’s unique to Programmai is our ability to leverage the complex interactions found using deep learning techniques across your session analytics and transactional data sets.
Simply put, if you have a consumer on your website behaving in a way that a high CLV customer does then our model is going to start scoring and describing them as such.
This next visual shows that we’re able to isolate those with a predicted high CLV over 12 months and regardless of their immediate conversion probability, we can use this predictive audience to create a “lookalike” that instructs Facebook to go find us more of the same.
Awesome, but how is this different to what I’m doing?
Good question you at the back there!
There’s one key difference here, this invests in future behaviour to yield superior return.
These high predicted CLV consumers haven’t all realised their potential yet, that’s for you to nurture, but they’re already behaving like your best customers — so let’s use that as a means to recruit others to the website who’ll behave similarly.
It might’ve taken you years to develop your best customers, so that limited list will quickly fatigue and saturate — you’re restricting Facebook to find people with those characteristics, but things move and change at a rapid rate.
Using our predictive approach, you’re giving Facebook a constant, dynamic, fresh list of high predicted CLV customers for them to prospect with every day. This optimises for the right behaviours, not merely purchase frequency like rearward CLV models.
It’s one helluva dreamy flywheel.
Bonus analogy for those who’ve made it down here *hi-five*
My Dad loves analogies, so I’d like to close with one that depicts our view of activating CLV in the paid channels and the greater good this brings about.
Farmers selectively breed the best crops and grains each year that withstand the winters. They put their faith in the strong genes of their best crops for next years harvest.
We do something similar when helping to find new customers, using the behavioural traits of those showing signs of longevity with the brand, to find new customers who are just like them.
This is not simply about finding extra return from each customer, it’s about the experience they’re having and the knock on effect that has on your brand and how it’s perceived.
Cultivating and selecting for super fans, is like having an external sales force out there saying great things about what you stand for, and your product or service, who recruit others through the passion they exude when referencing you.
Today’s fickle, race-to-the-bottom culture is akin to the fast food industry. It’s made cheaper, leaner and is ultimately not good for health. Sooner or later we’re going to realise this shrinks the whole and makes us suffer as a consequence.
We’ll eventually crave quality again, but quality is available today — let’s nurture it.
Next up, Paid Search.