Jan 31
Reducing Cognitive Load and Optimizing Marketing Campaigns
How can one think in terms of systems and reduce cognitive load and also apply it to the world of Performance Marketing?
This is an edited version of a post I wrote on Linkedin earlier.
As a Computer Engineer, my education (especially my algorithms class) has distilled a core philosophy around “reducing compute” if and wherever possible, not just in the software I architect and build and deploy but also in the compute I make my conscious mind goes through. Over the year, basis the experiences I have had, I have built several systems that help me reduce cognitive load for various simple activities like driving on the road and investing in indexes of public markets to running performance marketing campaigns. Through this article, let me give you two examples of systems I use often and how they have helped me reduce cognitive load by keeping the answers ready for me even before a scenario/question plays out.
System 1 — Reducing Compute while Driving on an Indian Road
I live in India, and driving here can be stressful for your cognitive function (some of you will agree with me); it definitely is stressful for me. So I prefer to drive on the extreme right of the road, allowing me to skip one side compute. I do not have to think about the traffic coming from one side because there is no space for traffic to overtake/cross me from that side. For those of you who have never driven / experienced driving in India, unfortunately, our sense of lane management is still underdeveloped.
While some might say that why not pick the extreme left, well you can do that, but I prefer not to drive that slow ;) (and not too rash to alert speed cameras)
System 2 — Index Investing on 2% Down Days
On days like today (Nifty ended down by around 2.5%, you can pick S&P or any other popular stock index of your country here), questions like what should one do are avoided by me through a system I call “2% down day System”. Rules of the system are simple: If the Nifty 50 Index is down by more than 2% in a day, I prefer to invest 4 SIP’s (Systematic investment plan, a way to invest money every month instead of going for lumpsum payments) worth of money in a lump sum on that day (so let us say if I had SIPs of 10k, then I will invest 40k on that day through the lump sum route. Why? Because I am in it for the long term, and markets generally have gone up over the longer term. Why bother with short term gyrations?
There have been a total of 419 days out of 6500 odd number of days when Nifty50 ended down more than or equal to 2%. If you had bought equal units of index every time it ended down 2%, you would have made an XIRR of 12.49% since 11th December 1995. If we take this data from 2009 (your first investment would have happened on 7th January 2009), this XIRR stands at 13.47%
In case you are interested, I have compiled Nifty 50 Data since 1995 December at this link (click here).
But how does System Driven Thinking show up when optimizing a Performance Marketing campaign?
If you look at the series of steps (or what I call the path of performance marketing) one must take to optimize any performance marketing campaign, you will realize that there are several opportunities to find performance leaks and plug those through a systems approach.
Performance Marketing (just like investing) is all about finding edges systematically, to find and exploit a certain inefficiency and make gains basis that.
A question I often ask newer campaign managers to think about when looking at data is to find opportunities to come up with rules and then automate those rules using products like Vidtech.ai. I have seen several such rules and been part of discussions around ideas, and these are three large buckets they fall in
a. Static rules — Rules like if eCPM goes above $2, then reduce the budget by 30%. These rules have their history in some past experiences that campaign manager has had.
b. Dynamic rules — Rules like if CTR goes below 5 day moving average, pause the creative and load up a new one.
c. Machine Learning-Driven Models — These are an extension of dynamic rules, but these are difficult to describe in words. For example, a certain ML model might increase the budget even if eCPM is above $2, because that model has seen something in the data that suggests the win rate will go up if one takes such an action.
As a marketer, think if there are opportunities to plug leaks using either of the above three rule sets(static, dynamic, machine learning-driven models) for each of the items in your performance path. Ask yourself, am I spending too much time thinking about things that happen frequently and do not gain significant yield at an individual level? What auto action can I perform in the future without the risk of losing a lot?
