In 1997, I was an apprentice engineer in a Tire company in India after finishing my degree in mechanical engineering . One evening, a machine broke down in the line and I quickly figured out that it’s just a broken spindle that needs to be replaced . I did some quick calculations and figured a 10.2 mm diameter is what the replacement should have . I could see the confusion in the eyes of everyone around me . Someone quietly went to the store and got the replacement and work progressed . The next day – my boss took me back to the machine , and showed me there was a panel with clear instructions there on parts – and the standard size replacement was 10mm . There is no such thing as a 10.2mm . He was sympathetic – he coached me that 90% of the time , you don’t need to worry about actual calculations and have to just follow the manual . He never gave me an example of the 10% when I will need to know the calculations 🙂
The next episode happened in Colorado in 2000 . I was a young programmer struggling with a massive old C program that started misbehaving after I added some functionality needed for my project . I didn’t change any existing code – and my code would compile without error and execute when I did it as a stand alone program. I went to the team leader – a long time veteran of HP-UX and probably the best programmer I have seen in my life . He casually asked me “Anything odd with the assembler code?” . I am not a CS major – and while I thought I was a really good programmer in C and a few higher level languages , I didn’t have the faintest idea on how a compiler actually worked or even how to read assembler code . Well, I was given a 30 mins tutorial and a manual for instruction set architecture . I struggled for weeks and eventually figured out what was wrong . I will spare you the details – but I walked away thinking that all mission critical code should be compiled without Optimisations turned on . I also learned to my horror that compilers can actually have bugs . Till today I don’t know if the compiler I used had an issue – but to be honest , I have never felt confident enough to blame a compiler even once when my code fails .
I wrote my first BASIC program in 1986 and first C program in 1989 . Till this episode in Colorado in 2000, I had never thought about the need for understanding what happens at a level below (N-1) what I needed to learn for everyday use . And in general I would say I had spent more thought on higher level abstractions (N+1) from where I am operating from .
My father was a very talented mechanical engineer . He used to tell me when I was in college that an engineer’s job is to make sure that whoever used the output of an engineer’s creation should be able to take it for granted – a lamp should switch on , a car should run when ignition is turned on and so on – without the operator knowing how it happens . And when it doesn’t work – most of the time the operator should know what’s wrong , and quickly decide if it needs expert help . By his definition – I wonder if he would have agreed that software is a real engineering discipline 🙂
If the episode in 2000 with Assembler had not happened – I doubt I would have developed an interest in N-1 thinking as my learning philosophy at all . It did help me quite a bit as moved into more business leadership roles later in my career . As I wrote recently about scaling a business , the ability to go to N-1 is critical when rethinking the building blocks . Otherwise we routinely get stuck in status quo and at best some incremental progress . Equally important is the fact that the moment you have solved things at N-1 , you need to zoom out to N+1 to pick up speed .
2021 has been quite an interesting year and I have alternated between “will this year ever end?” and “Whoa – are we in December already?” . Both from a business perspective as well as from a personal perspective – I had to learn new things and act differently . I thought I will share what I learned , with the hopes that perhaps some of it will be useful to others
1. Every step-change will break things where you least expect it
I was very proud that we were able to shift thousands of people in the team to work from home last year with zero difficulties because we have a strong business continuity plan that we trained for and implemented efficiently . So I had a false sense of confidence that it will be equally smooth when adding more people to the team . I was wrong – everything from courier service to background checks to laptop availability failed to scale at a certain threshold . These are all things I took for granted all my career . Thankfully we have such a great team that they sorted it out extremely fast !
2. Over communication is mostly a bad idea
When we started remote work, the instinct was to checkin with all the teams frequently . But very soon – the teams adapted to the new norms of working , and we didn’t tweak the “checking in” frequency. It became a diminishing returns investment of effort and leaders started burning out faster with the extra time spent on an activity that could have used a different cadence . Same with mass emails , all hands calls etc . Less is definitely more !
3. But you do have to over communicate some times
What works with people who have been in the team for a long time doesn’t work for people who are new to the team . That was true in the past too – but scale puts a spotlight on it quickly ! Questions that would get asked to someone sitting next to you in office would now often need a manager to explain the answer . Mentoring younger colleagues coming from university online is not the same as mentoring an experienced hire online . We had to learn to segment and tune our approach every time we detected a pattern . Again , we also need to learn when to ease off with the new team . I do wonder if these problems will get addressed by HR Tech at some point
4. Free form feedback is way more useful in uncertain times
As an analytics guy by training, I measure everything . That didn’t change during the pandemic times either . But I did learn after a couple of quarters that standardized questions are very limited in these times to address issues and opportunities with the client or my team . Free form feedback is where the useful information was mostly available . I read every comment that my team and client make in the surveys – and we talk about addressing them in our leadership meetings . I also use sentiment and tone analysis with ML to get a gauge of the aggregate as well
5. Invest in leadership ranks ahead of scale
I am a firm believer in leaders at every level making fewer but higher impact decisions compared to their team if they have to be effective . In uncertain times , there are hundreds of more decisions to be made even if the business is steady . There are thousands of more decisions to be made if the business is growing . If you don’t have enough good leaders – you will sink faster than you can imagine . Good teams grow because of strong culture . It’s very easy for the culture to go south if scale happens in an unmanaged fashion . That’s another reason why having good leaders are vital .
6. Invest in operations
A highly efficient operations team ( finance , HR, bizOps ….) is the reason why most business leaders don’t die of panic attacks . When they are very good – leaders occasionally take it for granted that they have infinite capacity . Operations have people and processes . Both parts will stretch only to a limit and they they will break . Relook at literally everything that is needed to keep the business growing and invest in operations and redesign workflows .
7. Relook at all communication channels
I hope there is a massive series of studies done on this topic . Slack has been a life saver for me. I over estimated the effectiveness of video . And I rediscovered how effective good old phone calls are . A great example of the change in effectiveness are the quarterly all hands calls . I don’t see a tenth of interaction in those massive webex events that I get on a slack based ask me anything session with my global team .
8. Business Relationship building has evolved
A pleasant surprise for me this year was that unlike 2020 – it is now totally effective to build new business relationships online via webex and email and calls , without face to face meetings . It’s incredible how long established norms of shaking hands and breaking bread as first steps in a new relationship got replaced by talking about children and pets on webex ! No business scales without scaling relationships vertically and horizontally – so this is a very good change in my view
9. Take good care of people – that is one thing that has NOT changed
All business is ultimately about people on all sides . That’s the one constant that did not change in pandemic times . The great resignation is something we need to learn from and act on quickly . Money , flexibility , interesting work – there are lots of reasons why people quit their jobs . You can’t fight the forces of market – you have to adapt quickly and find your own equilibrium . My fundamental view has not changed in pandemic times – I think the key to attracting and retaining good people is to make sure that leaders and their teams feel comfortable in discussing everything openly and being fair to each other . If I look at where I spend most of my time – I think it’s probably 50% on helping my team , 30% with assisting my clients , and 20% on all other things taken together .
10. Increase the focus on learning
Pandemic has caused a lot of grief in the world . I lost friends and family – and I don’t deny I have an amount of fear in my mind at all times . But for business – it has largely created more opportunities . But to tackle these opportunities effectively – you have to be an aggressive learner , and encourage everyone around to do that . On the technology side – I spent my time learning more on Redhat openshift , Ansible and GCP . I also have been reading up a lot about the tech behind crypto currencies . On the non tech side – I have been reading more about WW2 and life during Great Recession
11. Take some time off – don’t make the mistake I did
I am generally good at taking some vacation every year to reset . I did not do that this year and it certainly is proving to be a bad idea . I know I am not alone – and it’s not going well for others who didn’t take the time off either . Almost everyone I know in my team and in my network who has taken the time off are more productive than I am .
12. Do something else outside work
Last year, it was mostly playing cards online daily with my friends . That has come down a lot this year . But 4 days a week, I take my puppy to training for IGP competition which we hope to start competing next year . I try hard to block that time off from all work – and it literally has been the best decision I made this year . It brings a much needed balance . I am sure that if I hadn’t decided on that – and also not taken vacations – I would have completely burnt out half way through the year . I am fascinated by the range of hobbies my friends have picked up new this year- Ironman , wood carving , singing , equestrian etc . In every case their experience mirrors mine – and their businesses have had a positive impact .
Staring at data is a big part of my job – but it’s very rare that data alone gives me direction on what to do next . Data needs to be put into the context of what I feel (and what others feel) and then some decision gets made . So in reality – I am not really data driven , I am more “data enabled” when it comes to my decision making process .
What I feel – perhaps what can be called my intuition – is based on my past experience . So I often wonder how useful it will be to depend on intuition when it comes to decisions about future . That led me to think about my feelings a little more – and that led me to three (overlapping) possibilities on why I decide to go forward with some decision
1. I like and trust the people who will execute on it
This doesn’t happen unless I know them really well . And amongst the people I know – only a few fall into this category when I think about it more . With such people , I feel strongly that they are so driven that they will make it happen irrespective of challenges I can anticipate . The reality unfortunately is that my success rate is only that of a coin toss . While some data comes into play – it’s really not data driven or data enabled if I am honest about it . I will however add that when everything else is “iffy” – I trust my judgement of people and make bets on it . In such cases – at least so far – it’s been better than coin toss odds for success .
2. I understand it from first principles
These usually turn out to be my best decisions – I understand the problem well from the ground up , and consequently I have a framework to evaluate solutions . All the examples I can think of have ended well – but I am sure there is some bias in my thinking, so let’s say 80% success rate . I can use data to validate my assumptions and mental models – so these are data enabled decisions .
3. I can see the potential tweaks needed to make it work
These are usually things like redesigning the process , having a different leader for the team , resetting the business case etc . I think this is where experience comes in handy – because it’s essentially pattern recognition that is helping me . To increase my odds, I also tap into my network for their experience once I figure out the pattern . Interestingly , this is the category where historic data comes in handy . Quite often – it’s staring at data that gives me a starting hypothesis on what needs to be tweaked .
The time dimension
I try hard to be thoughtful about the decisions I make that have large and/or long term impact . That needs time to deliberate . I conserve my time, energy and brainpower to make such decisions by routinely delegating whatever I can to my team . But even then – a third of the time , I will have to make snap judgments with limited time to deliberate .
As I look back at examples of such decisions – I see an interesting trend . When I have delegated and conserved my time and energy – my snap judgments generally turn out to be ok more often than not .
What is the net net ?
I am convinced that we don’t really need human decision making if it’s purely data driven – such decisions should be automated ( with manual over rides and other precautions on ethics/security etc taken care of ) . Humans (generally) should only have to care about data enabled decisions .
What’s the weakest link here ?
There are two ways to think about data enabled decision making . One is using data to find answers to questions you defined . The other is defining questions based on data . The former is largely a solved problem already . The latter is what keeps us employed 🙂
And Vijay Says... 