How to manage schedule risk

In one of our usual catchups, I was having a chat with one of our engineering leads, casually explaining a feature request we had received from our sales organization. As a team, most of our clients are internal, and we are always happy to receive such feedback. It came in as a mild shock when the response was we wouldn’t be implementing this feature. I knew it wasn’t a matter of time; we had processes to handle long term as well as delight features. So why the no?

In the same casual tone, he went on to explain a viable scenario where with the feature implemented, the interaction between sales and finance systems would grind to a halt, and we would then need to push an Android update. In short, the feature delivered nothing more than delight value but had the potential to stop the entire business!

I moved on from this conversation with even more respect for this individual; this truly is the mark of great leadership.

In this entry, I will be discussing a bit more about risk and how you can relate to it within your engineering organization.

Be clear

Risk is such an ambiguous concept. You can’t see, touch, smell, hear or taste it. Even intellectually grasping it is a hard endeavor.

It is easy to ignore such things. After all, do you really want to be the Grimm at the project kick-off party?

Still, it’s essential to enumerate your risks and the likelihood the risk will materialize. It doesn’t need to be a long or formal document. Frankly, even if you drop the likelihood calculation, just the fact that the possibilities are on top of your mind is well worth the effort.

If you already do this, pat yourself on the back. Now, to the more difficult task. How do you know the risk has materialized?

Suppose in your risk register, you had “Developer leaves team” and “Vendor X stops supporting Y library.” You may say, well, of course, we will know when these risks have materialized. Once you dig a bit deeper, you will see how this belief is flawed in an insidious way.

Let’s take the first case. Most organizations have a reasonable notice period before you can leave, for an engineering manager, this is cheap comfort. When a team member has decided to move on from your business, all sense of initiative goes out of the window. What you now have is not the whole brained, energetic, committed dev you used to know, rather a by the book, can’t wait till 5pm, does minimum work to get it done lad. In short, your ambitious schedule is f*ckd!

Have a contingency plan

If you have a risk register and have not thought through how you will handle the risks, then in essence what you have is a glorified worry list.

I have come to see risk as having a fascinating characteristic. Once it becomes obvious, a risk will transition then what you need to do becomes obvious but mitigating it, if at all possible can only be done at a high cost. Before a risk is obvious, what you need to do is less obvious, but it can be done much more cheaply.

To illustrate the point, think of an app that failed in production because it runs out of memory. In this case, once it becomes manifest the event is happening, there is little else you can do if you are stuck in a VM type of architecture, scale it, and the application quickly eats through the new RAM. In short, what you need to do is visible but very expensive. If you had thought about this problem when the app was still in Greenfield, you then wouldn’t know if the risk would ever materialize, but solving it would have been as easy as say, choosing a different programming language, or going cloud-native.

You obviously can’t plan for all risks, but if you can select the risks with the most significant impact and the chance of happening. You maximize your chances of being able to sail through the worst of the storm or better yet, avoiding it altogether.

Did it happen to your peers

In Swahili, there is a saying

“Ukiona mwenzio ananyolewa, zako tia maji”.

Roughly translated, if you see your neighbor getting shaved, you better start wetting your own hair.

The point here is if other smart people have tried to solve the problem you are now embarking on without success, and you can not clearly articulate why they failed, there is probably an underlying structural issue.

An example here is the Kenyan cashless markets. Almost all international players come in with a firm intention only to accept online payments via credit cards, soon this softens to Mpesa, and for those that thrive, cash makes its way back into their model. I am not saying I understand why pure cashless doesn’t work or that it will not work, maybe even sooner than we think. What I am saying is, if you are entering the market now, you are well advised to consider lessons of your predecessors.

To get started, you can think of some of these risks, almost every leader in a tech organization I know has had to face some variant of.

  1. Developer leaves your team
  2. An essential vendor/library drops support suddenly
  3. The client gets new ideas as the project starts shaping up

How do you manage risk in your own organization?  Talk to me in the comment section below, my Linked in chenchajacob or my twitter @jchexv



Why I never advise anyone to freelance

I have been in the industry for a while now, and one of the questions which keep on popping up is, “Should I freelance?” or even more directly  “After all these years, why don’t you go it alone?”. The indication here is why don’t I become a dev for hire. While I no longer offer value through hands-on coding, I still feel younger entrants into the industry need to understand a bit more of why I think small scale software consulting is not a good idea.

Let us begin with a story. Years ago, I walked over to a clients office with a proposal in hand. I had just started a new consulting firm and even had two devs in employ. This was going to be our first major client. On my proposal, I had the devs profiles, the company profile, and what we thought we could offer.

The meeting was fairly long with the director of the client firm quickly going through what his vision of the product would be, basically a custom ERP. After our quick paced session, he asked, “So what will this cost us?”.  I am against such estimates, so I explained we would need more time. Further given the amount of pre-work required, field visits, interviews point estimation, and so forth. We would need to charge just to develop the scope of work and associated project documents.

The mood turned somber as the director took it upon himself to explain just how stupid it was for me to have the guts to ask for payment for the work needed to estimate the real work. New in negotiation, I capitulated and did as he wanted me to do. In the end, they ended up going with another contractor, and we took the cost.

Is this why I ask you not to freelance? No, at least not the only reason. The industry is simply stacked against you. Let me explain further.

Low barriers to entry

There is no other industry where the cost of entry is a laptop, headphones, and a decent internet connection.

Your first response may be, sure, but what matters is not these physical assets, its what is between my ears.

Perhaps this is true, let’s look at it another way. If you are top 1% of the population in terms of intelligence, then a city like Nairobi with 3M residents, there are 30,000 people JUST like you if not better.

The truth is, you are not that special. Continuing to delude yourself in this way maybe fun even satisfying but will prove disastrous to your pocket.

The field will continue to flood with X consulting, X tech, X designs kind of companies. Profits will be suppressed.

The tyranny of time

Time is a limited and precious resource. You can not buy it, rent it, get a loan of it or in any other way extend it. What you got is what you got. Yet all activities need time to gain reality.

Guess what takes a LOT of time? Product development.

Real life coding is nothing like what you get in a teaching environment, there are no neat solutions. Past your fresh start, most of your time will be spent reworking existing code.

Vi has built its entire product successfully based on the idea you spend more of your time editing the code rather than adding new code.

Factor in the time you spend fixing bugs, and you are only coding 30% of the time.

The news gets worse, Business Development is an even bigger beast. Clients rarely know what they want to be built. More often than not, they feel their problem and have a sense tech may be able to help. It will be countless meetings and brainstorming sessions before you have what looks like an addressable problem and even more time to architect the solution.

Optimistically, you will be spending maybe 5-10% of your time coding, aka time you can bill the client. Do you see how this can become a problem?

Scale works against you

We all learned about economies of scale somewhere in Econ 101. The basic idea being our costs per unit reduces with the more units we sell. Did anyone bother to tell you about diseconomies of scale?

Some businesses get stuck in an unfortunate situation where every new unit sold costs more. As a solo tech consultant, you fall squarely in this bucket.

Technical engagements tend to offer little opportunity to apply prior knowledge. Yes, I know you can carry over libraries you have built, but unfortunately, you can not carry over business domain concepts which are likely what the client cares about.

You could always hire more devs, putting aside the question, why would they work for you and not themselves, experienced devs are hard to find, and when you recruit them, the cost will be significant.

Poor bargaining power

The old saying “No one was fired for hiring IBM” still holds its ground to date.

For established clients aka where all the money is, are characteristically risk averse. They have a working business model and don’t want to jeopardize it. Further, after sales support will always be a key concern. In short, procurement departments are allergic to one man shows!

There is some reprieve here, small businesses will be more willing to engage. Competition here will be stiff, even worse, you may find yourself encountering competitors who don’t really care about the bottom line. After all, most people in this industry are here because of the promise of freedom, a chance to work from the beach. It will be tough to underbid such an individual.

So what works?

Advisory services are clearly a profitable industry, so how can you be successful here?

Here is a little known secret, successful advisory firms are NOT paid for performance. If you were to call PwC, they would offer you not a solution to your problem, a hefty hourly rate. You, of course, you have no option, an audit is a regulatory requirement, and you are better off with a known name.

What about mutual funds? Well, always read the fine print. The fees are charged against your investment, not against what you earn. In short, they make money whether their decisions pan out or not.

You could get over this huddle by charging per hour as well. A time tested model is convincing the client to embed you in their team. So you take on tasks just like any other dev. This, to me, looks and feels just like regular employment with none of the employment benefits.

Large networks such as Toptal or Upwork may help. They also embed you in a team. In my opinion, the net effect is the same as above.

In fact, the only viable way I see out is to work on a product and use your consulting fees to get by as you wait to land on a product market fit. Once this is achieved, shift all your attention and resources to the product business.

Have you managed to make a freelance operation work? Talk to me in the comment section below, my Linked in chenchajacob or my twitter @jchex


How to collaborate with your Operations colleagues


Twiga tech team currently has 25 members, the business in whole has 400+ employees and rapidly growing. This means for every techie we have 16-20 colleagues in other departments. Even more importantly, Twiga is a Tech enabled business, that is, the tech team plays a supporting role. Working well with other groups is not a luxury for the team, it’s a necessity.

As you can expect, there is bound to be some friction when operations and tech have to work closely together for business success.

As a tech team, we have great appreciation and respect for our operations colleagues. It is not easy to sell to over 2,000 vendors daily from a rapidly growing customer base of 6,000 customers and a network of over 13,000 farmers. Clearly, there are a lot of people doing a lot of things right.

In my experience, I have come to find out the areas of friction originate from two major areas:

  1. How the departments interact with the business
  2. How the people within the company interact with each other

Interaction with the company

Twiga is in the business of aggregating retail demand in African cities to provide low-cost access to better quality food.

To this end, Ops(Operations) and Tech(Technology) contribute in fundamentally different ways.

Imagine if you will, a company-wide meeting has been held, and all teams are reporting their numbers. Ops goes first. The crew talks about 100 tonnes moved from farm gate to vendor, 10 million Shillings in revenue and perhaps 4 million shillings in operating costs. Now, as the person accountable from tech, would you follow that up with, we delivered 100 story points? Even worse, 10KLOC (Thousands of Lines of Code)?

In these cases, you should use different approaches. For starters, try to identify simple to understand metrics. For example, our system uptime was 99% is more useful than our Apdex score was 98%. It doesn’t matter that the latter metric communicates more information. You can make it better if you can find business-oriented metrics, for example, we processed 10,000 orders and 9,000 deliveries without a single failure. Or 100% crash free users.

Given the flexible nature of the work, you must compensate by aggressively marketing what you plan to do, why you plan to do it, and when it will be done. Once it’s done, show don’t tell. Fairly regular tech demos where you invite your colleagues from other departments should help.

The moment a new business process is introduced, you can be positive pressure will immediately mount to have it reflected on the system. Know that businesses processes take time to mature, what seems set in stone right now will likely evolve into something barely recognizable once it hits the actual reality of operations. You want to give the process time. Otherwise, you will end up scrambling to rework your architecture every time the process bumps on a wall.

Interaction with people

Ops move fast! Almost all stock brought in is out within 48 hours with the notable exception of bananas which requires ripening.

Tech, on the other end, moves in two weeks sprints. There is progress in the middle of the sprint of course, but before you show this out, take a long look at an incomplete building or a car out at the garage. If the owner is nearby, their horror as the mechanic hammers, pulls and twists the various parts is palpable. Better to only share what is at least understandable.

Given this reality, I suggest you do far more planning than I would usually advocate for a tech team. You want your colleagues to co-create with you. Invite them to all brainstorms, you will be pleasantly surprised at the insights the non-techies can provide. The joint sessions will also allow them to peek into how you work, hopefully, in the process, they will also learn to appreciate the complexity of building a modern application.

Take the time to educate your colleagues. What you consider common may register as jargon to your colleagues. Don’t for a moment take the statement

“I am not a techie, I don’t understand this thing you guys are the geniuses”

or any of its variants. When it’s said, don’t take it as a statement of your superior cognitive status, interpret it for what it really is,

“You are a cost center who I am still not sure why I need.”

Make a great effort to understand your colleagues, get them to know how what you are doing will benefit them.

Finally, you want to be running an efficient support function. Most issues which will come to tech will likely be transactional in nature. For example, I can’t log in, I can’t see my records, etc. Transactional issues have an exciting asymmetry, they are usually easy to solve yet cause a lot of pain for the person experiencing them. This means you can create a whole world of value just by quickly handling them.

How do you handle collaboration between your tech team and the operation team? Talk to me in the comment section below, my Linked in chenchajacob or my twitter @jchex