“Even after making the most calculated decisions, we have seen that Digital Marketing Strategies may fail due to improper implementation or a possible disruption in the Market. What is the best way to redesign the strategy when your previous strategy has failed?”
We use the hierarchy of analytics to determine the cause of failure. This short 8-minute video explains the process:
Got a question for You Ask, I’ll Answer? Submit it here:
Subscribe to my weekly newsletter:
Please subscribe to my YouTube channel for more marketing and analytics videos!
Transcribed by AI. May contain significant errors.
In this episode of you ask I answer Pradeep asks, even after making the most calculated decisions some digital marketing strategies may fail due to improper implementation or disruption in the market. What’s the best way to redesign your strategy when your previous strategy has failed. Well, that’s a loaded question
in order to understand failure or success for that matter. You need to follow the hierarchy of analytics. So if you remember the hierarchy of analytics is descriptive what happened diagnostic why predictive what’s going to happen next proactive prescriptive What should we do about it. And finally proactive. Can we get it machine to do it for us
if we want to understand a failure, failure analysis is inherently looking backwards. So that’s going to be descriptive and diagnostic what happened and why
when we look at what happened. We have a KPI of some kind. Some outcome that we were.
aiming for. It could be
lead forms filled out. It could be purchases shopping carts completed it could be, it could be awareness. We just want to know that people are finding us but whatever the outcome is that we didn’t get we need to run intensive analysis on the data on the KPIs to identify
what was the KPI
what how far did we not achieve the goal. And then what we’re contributing factors that went into those KPIs.
The there’s two machine learning methods that I really like for understanding that one is called random forest and the other is called a Markov chains either one. If you have the right data that supports them will help us understand. Yep, that
KPI failed because these three variables did not add up these variables did not perform.
In doing that, analysis, we not only understand what happened we start to understand why we start to understand why we didn’t meet our KPI
and then we have to dig in qualitatively into our strategy to figure out why those variables didn’t match up. So let’s let’s use a real simple example I want to drive subscribers to my newsletter and I run a campaign run a bunch of ads do as a bunch of social posts
and I’m running and analysis and say, Okay, what variables feed into email subscriptions and if it turns out it’s a combination of social posts plus
say YouTube ads plus organic search. Let’s say those are the three variables that are machine learning spits out and says, yeah. These are the things that went wrong
with these are the things that lead to email subscriptions. I can then go and look and say okay well let’s take a look at each of these are the channels.
Which one which one did the least. Good job, and find out it was YouTube ads. Why, why did YouTube ads not do as well as they could have what we have done better. Maybe my content was boring. Maybe my lighting was terrible.
Maybe my microphone didn’t work. Something went wrong in that channel or in that method that failed.
So what do we do from a repair perspective, we could try and experiment where we shoot 1235 new YouTube videos as a way to determine did this is is this challenge. Actually the problem and and if we run a whole bunch of new ads and YouTube still doesn’t perform, then we have to say okay let’s elevate our questioning from
the data itself to the strategy.
when you reverse engineering a campaign. It’s kind of like reverse engineering getting lost when you go driving. If you get in your car and you know Google Maps up and running on your phone.
There are wrong directions which could be provided to you by the system that would lead you astray.
But if you fix the directions or the data sources fine on the roads are fine and you’re not achieving your goal at some point you have to question. Did I am I going to the right destination. You ever have that happen we yeah you run into, put your stuff in the GPS real fast and you realize that you’re starting to drive this doesn’t seem right. And then you realize he put it into like completely one place he went to want to go to the fast food place in in your town but he actually put the fast food place in like three towns over like whoops,
that wasn’t so good
strategically you pick the wrong destination.
So the same thing is true of analytics. If you do, you’re just died descriptive analytics and you do diagnostic analytics you figure out what happened you figure out why.
you’re still not achieving an overall outcome, then you know your strategy is wrong when you know that no matter how many different tactics and executioner methods and measurement methods used you just got the wrong destination in your corporate GPS,
then you have to pivot there.
That’s the best way to redesign the sort of thing is to understand at what layer. Did we fail at that we felt execution. We just didn’t do the thing right did we choose the wrong things. And that’s what the machine learning analysis your variables comes into play and
our we’ve to setting the wrong destination in the in the corporate GPS
the in terms of strategic identification.
That should be the last.
resort because you should have done that research up front. Now he didn’t do any research and you just said the strategy and you don’t hit go or a strategy was handed to you and then you probably want to do the research before you launch the campaign so that you prevent the strategy from failing entirely. Before you begin,
it’s the same as yes the GPS might be wonky but if you just set the wrong destination. It doesn’t matter how good the GPS is you’re going the wrong place. So the strategy is something that should be data driven and research driven to begin with. If it’s not, if you’ve been hand the strategy, then to avoid failure to avoid putting your own neck on the line on the chopping block do the research yourself and if the strategy is clearly wrong and you provide the data and you provide the analysis and you provide the insights to your stakeholders and they say, No, no, this is the strategy and you just know what’s wrong
might be time to update your LinkedIn profile because you know when we research. It’s just not going to go well and.
No matter how clever, you are with your marketing tactics and execution. If you’re going to the wrong place. You’re going to the wrong place. So
that’s the answer to redesigning your strategy when previous strategy has failed you go from the bottom up. Okay. What happened, why and then you can figure out what’s likely to happen and what you should do about it
tuned up the execution first understand the execution first was that a bad execution then got his tactics, did we choose the wrong make the wrong choices to implement the strategy and then get to the strategic layer and say, I’m not sure we got the right strategy. I think we went to the wrong place. We had the wrong destination and the GPS. That’s the set of models that I would use that allow you to document your thought process that allow you to prepare intelligent responses to difficult questions, particularly if your strategy was handed to you by another stakeholder and to ultimately protect you against making a career ending the.
stakes so pretty great question probably could dig in for hours on end into this, but that’s the 10 minute or less summer. Thanks for asking if you’ve got questions for you ask I answer, go ahead and put them in on my website Christopher S Pen calm. I’m happy to answer them at any time and please subscribe to the YouTube channel and also do the weekly newsletter which will see at the end of this video. Thanks for watching.
You might also enjoy:
- Is Social Listening Useful?
- Transformer les personnes, les processus et la technologie - Christopher S. Penn - Conférencier principal sur la science des données marketing
- The Biggest Mistake in Marketing Data
- Best Practices for Public Speaking Pages
- Marketing Data Science: Introduction to Data Blending
Want to read more like this from Christopher Penn? Get updates here:
Get your copy of AI For Marketers