Just Product Management Things - Issue #2
Developing a customer centric mindset; A radical new technique lets AI learn with practically no data; Sneak peek into Twitter’s feed algorithm
Big Skills - Compounding the smaller ones
Morgan Housel sets a very interesting premise - ‘Ordinary things that aren’t noteworthy in isolation can create something spectacular when they mix together’. I believe this premise would especially resonate with Product Managers given the rather generalist nature of their jobs that even over indexes on non-big skills. Instead of looking for that one big skill that will set you apart, focus on smaller things that will compound and help you become extremely valuable (and wise). The article shares a rather long list of often ignored things that if combined together can reap large benefits. I will only share my favorite 5 -
Curiosity across disciplines, most of which are outside your profession.
Outperforming by merely “doing the average thing when everyone else around you is losing their mind.”
Identifying what game you’re playing and not being persuaded by people playing different games.
Accepting some inefficiency and hassle without losing your cool. (My last manager literally hammered this into my head and I can’t thank him enough for this)
Getting along with people you disagree with.
Read more on Collaborative Fund
Developing a customer centric mindset
Customer centricity is about working backwards from the customers in everything the customer experiences and is to be employed by everyone in the organization. Every decision regarding every touchpoint is thought from the customer’s point of view and not from the organization / product’s point of view. Some of the suggested ways of improving customer centricity are:
Mindset: Aim to create customers into advocates of your product / company and in doing so, prioritize customer motivations, needs, reactions, emotions, etc. at each step of the product building process.
Emotion: Humans are irrational beings and extremely driven by emotions. However, emotional sense is least used to drive loyalty and advocacy from customers. Think of your customers’ emotions / intended emotions when they experience your product and tweak the product to trigger positive and strong emotions that result in delight.
Team: It is the natural tendency for people to think of themselves as customers. You can tap into this tendency by having a diverse team. Product teams are always cross-functional (PM, Design, Engg, QA, Marketing, etc.), but that alone doesn’t make them diverse. Instead, focus on driving diversity by actively seeking insights and opinions from everyone on the team
Process: Ensure that you have a process in place to talk to customers throughout your development process and let the development be agile enough to accommodate insights at a later state / during iterations.
Customer Insight: While behavioral data and the VoC (Voice of Customer) will give you some insights into the customer, it won’t provide the complete picture and hence it is important that the team develops an intuition for interpreting these better through leaders.
Growth is about systems, not hacks or ideas
Growth ideas often get spoken about, but they often result from running 10s of such experiments. It doesn’t mean that the remaining experiments were a waste of time. Instead, it means that there are no silver bullets that will lead to growth and that it is the system of experimentation that is more important. The cost of the failed experiments is usually recovered by the winning ones and the cost of not experimenting is usually higher (leading to not growing / irrelevance). A system for growth should answer some of the following questions across the three pillars - strategy, process and culture.
[strategy] Does your growth team have a clear objective to be achieved?
[strategy] Do you have a measurable metric to measure if you’ve gotten there?
[process] Does everyone in the company know how to collaborate?
[process] Are you running your growth meetings religiously?
[process] Do you have consistency in terms of test volume?
[culture] Can your superiors clearly keep track of your work?
[culture] Is there a way to promote wins and deliveries?
A radical new technique lets AI learn with practically no data
The technique is called “Less than one” shot learning. It aims to mimic the human learning pattern where kids can learn about something without looking at 1000s of images of it (unlike machines) and can also imagine an in-between thing (if you tell a kid that a unicorn is in between a horse and a rhino and they only see image of the latter 2, they can imagine and detect when they see an image of a unicorn).
Scientists have achieved this using engineered training dataset that contain all the information from 1000s of data. They experimented this on the MNIST data set (using 10 images instead of the usual 60K). Further, it seems like even two samples could theoretically encode any number of categories.
While these are early days, this research is likely to help push new boundaries in the world of AI.
Reverse-engineered sneak peek into Twitter’s feed algorithm
Michael Girdley reverse engineered some aspects of Twitter's feed algorithm from his experience tweeting. Here’s a summary -
External links punished: Twitter wants users to stay on the platform and hence usually reduce impressions for tweets with links (that takes users outside Twitter)
Competitor mentions punished: Any tweets that mention Twitter’s social media competitors usually get punished with lesser impressions
Bias to positive texts: There is some sort of sentiment analysis running on tweets and the ones with positive sentiments are preferred over negative / neutral sentiments
Controversy punished: Certain words around diseases, controversial topics, etc. are demoted
Punishes flame wars & low engagement: Your next tweet’s impression is likely to be decided with how many low engagement tweets you recently posted
Twitter has ~4 "rings" of followers: Ring 1: See every post by someone you follow. Ring 2: Only higher liked posts by someone you follow. Ring 3: Only extremely outperformed by someone you follow. Ring 4: Extremely outperform tweets + don't follow.
Who engages w/ you matters: If bigger accounts engage with you, it matters to your impressions and engagement overall
Threads rewarded more now: Long-form content on twitter (in the form of threads and not links) are promoted
Read full thread
Other interesting reads:
Interested in learning more about marketing?
We also do a similar newsletter on everything Marketing called - Marketer’s BAE. Do give it a read / share with someone who might find it relevant.
 | A guest post by
|