Denis asks, “How will AI help marketers with retargeting and sales conversion?”
The greatest challenge most marketers face is too much data. At any given time, we could be looking at hundreds, thousands, or even millions of data points, from retweets to bounce rates. No human can compute these, and even if we could, it wouldn’t be in a timely enough fashion to be usable. Learn about the 3 machine learning techniques that make our marketing data manageable.
Can’t see anything? Watch it on YouTube here.
Listen to the audio here:
- Got a question for You Ask, I’ll Answer? Submit it here!
- Subscribe to my weekly newsletter for more useful marketing tips.
- Find older episodes of You Ask, I Answer on my YouTube channel.
- Need help with your company’s data and analytics? Let me know!
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for watching the video.
In today’s episode, Dennis asks,
how will AI help marketers with retargeting and sales conversion? This is a really interesting question because one of the things that marketers struggle with is what causes a conversion? Or what factors? What measures what metrics learners what dimensions lead to conversion, or contribute to conversion. A big part of this is a foundation of attribution analysis. What things what pieces of data lead to have driven conversions on the past. And then obviously with things like retargeting. You’re trying to focus on what things are likely predicting what things are likely to cause conversions in the future and the way AI and machine learning help with this is dealing with what are called week learners.
A week learner is any dimension or metric who’s predictive power is just barely above chance. Just slightly more than that random chance. It’s, it’s a call a week learner because it’s a weak signal. It’s not a signal. That is that by itself is a very strong signal. So for example, number of times someone has retweeted your tweets, right? That’s for a fair number of visitors, that’s going to be a really weak learner. In fact, it may or may not even be statistically relevant. But the very least is probably a week learner
that are things like
how many times someone has opened an email the number of social channels, someone follows you on the pages they visited on a website, the amount of time they spend on a page when you think about all the data that we have access to as marketers and then we consider that most of these metrics are pretty weak. You get to sit start getting a sense of the scope of the problem like we have all this data
None of it is the one answer that we’re looking for. The answer says, this is the thing, this is the thing we need to do more of it would be nice. If it didn’t work that way, we nice to know say you should always send email on Tuesdays, that’s going to cause all your conversions doesn’t happen. So how does AI help with this through techniques that aggregate week learners together and and make them function as a stronger learner, we can get a sense of what combinations of dimensions and metrics matter most like give an example let’s say that
each one of these playing cards, what the thickness of it indicated the strength of that learning, right? So you have
tweets and emails and
Taiwan page and bounce rate and stuff and you see all these cards over and over again. Now by themselves. Each one of these cards is a signal and these signals are real thin right now if we just start doing what’s called bagging, we start putting similar cards together similar cards together.
Similar cards together,
we end up with a stronger signal do this enough and you end up with a very strong signal.
And there are techniques like bagging and boosting that allow you to do this that are machine learning technologies that used to physics and use math to figure out how you go from pile random cards to block that indicates this combination of things is the thing that works
and that’s really where AI will be used to help marketers with with the retargeting the sales conversion, by identifying through techniques like gradient boosting through techniques like random forests and stuff.
This combination of variables, whether it’s time or attributes of the person or number of times they’ve done something or percentage of things that they’ve done something this combination is the combination that will lead the outcome now if you want to kick it up a notch.
There is then a machine learning technique called multi objective optimization that says, hey, what do you care about? If you’ve done an attribution analysis or you’ve done the process of attribution analysis? You know that
many times there isn’t that one answer. There could be a bunch of things that perform well or in your company. different teams may have different KPIs, but they’re all important. So the events team has registrations as as a KPI, the direct response team has opens, right, the sales team has Mark has has deals closed.
One of the challenges that that folks working with machine learning and marketing data is, is not optimizing just for one metric, but optimizing for multiple metrics. And that’s where something like multi objective optimization will come into place to take
all these signals that you’ve aggregated through bagging and boosting entry and forests and stuff and then use those
blend them together and figure out okay, now how do we allocate both credit and resources across these different objectives to help everyone meet their objectives without causing significant harm to another objective. And that may mean that by the way that
you may not get everything you want for your objective, you may have to compromise with other departments of the teams with other pools of resources. But the net effect the net computation will be that you’ll have the greatest overall benefit your company and that’s another that’s another part of machine learning. That’s super important. That is very difficult to do. Because the math behind it is crazy. It’s like huge, but if you’ve got it working, then you can understand better what you need to do, what combination things you need to do in order to get the outcome the overall outcome you want.
This is all predicated on having a strategy with clearly defined goals because you
Can’t do multi objective optimization if you have no objectives, so you have no clearly defined objectives. So that’s an important part is a prerequisite.
But that’s how AI is going to help marketers with retargeting and sales conversion the most is by taking all these data points that market is simply are not as much as marketers, by the way, its people we are not able to do computations have 100 or 200,000 a million different variables at the same time, we physically cannot do that appear our software hardware just can’t do that. So we need our machines to do that for us. But that’s how marketers will get benefit out of these technologies that and relevant benefit to the performance. So this is all part of the fuse digital conference which I’ll be speaking at in November in Philadelphia. So make sure that if you’re interested in exhibiting or you’re interested in attending that you reach out to the folks there
but good question. It’s a fun question and the and the technology behind it if you get it working is a serious competitive advantage. It is a competitive
advantage like no other because you will know what works. For the most part you will know what works obviously still based on the quality of your data safe data if your data is terrible is not going to help you, but assuming good quality data, it is an advantage that few others can can put together to compete against you. So as always, please subscribe to the YouTube channel the newsletter and I’ll talk to you soon take care want help solving your company’s data analytics and digital marketing problems. This is trust insights.ai and listen to how we can help you
Want to read more like this from Christopher Penn? Get updates here:
Get your copy of AI For Marketers (2019 Edition)