I recently bought a Nest Hello to replace my perfectly good, near new, Ring Video Doorbell. The experience got me thinking about sunk cost in IT and how significantly it strangles the business and costs companies ridiculous amounts of money.
When I first saw the Nest Hello, I had no interest. I had recently purchased and installed my Ring. I was happy with it, and the Amazon Alexa integration was great. I had no need to change. A few weeks later I decided to replace my home security system because it’s a cable provider system and like everything from a cable provider it’s a shit service at caviar pricing because ‘Hey, you have no choice you sad F’er.’ That’s the beauty of the monopoly our government happily built and sustains for them. I chose to go with a system from Nest, because I already have two of their thermostats, several of their smoke detectors, and a couple of their indoor cameras. I ordered the security system components I needed, and a few cameras to compliment it, then I looked back into the Nest Hello.
The Nest Hello is a much better camera, and more feature rich device. More importantly it will integrate seamlessly with my new security system, and existing devices, eliminating yet another single use app on my phone (the Ring app.) The counter argument for purchasing the device was my sunk cost. I’d spent money on the Ring, and I’d also spent time and hassle installing it. The Nest might require me to get back in the attic and change out the transformer for my doorbell as well as wire in a new line conditioner. Not things I enjoy doing. The sunk cost nearly stopped my purchase. Why throw away a good device I just installed, to get a feature or two and a better picture.
I then stepped back and looked at it from a different point of view. What’s my business case? What’s the outcome I’m purchasing this technology to achieve? The answer is a little bit of security, but a lot of piece of mind for my home. I live alone, and I travel a lot. While I’m gone I need to manage packages, service people, and my pets. I also need to do this quickly and easily. This means that seamless integration is a top priority for me, and video quality, etc. is another big concern. Nest’s Hello camera feature set far better for my use case, especially when adding their IQ cameras. Lastly for video recording and monitoring service, I would now only need one provider, and one manageable bill rather than one for Nest and one for Ring. From that perspective the answer became clear: the cost I sunk wasn’t providing any value based on my use-cases, therefore it was irrelevant. It was actually irrelvant in the first place, but we’ll get back to that.
I went ahead and bought the Nest Hello. Next came another sunk cost problem. My house is covered in Amazon Alexa devices which integrate quite well with Ring. I have no fewer than 8 Alexa enabled devices around the home, garage, etc. Nest is a Google product, so it’s best integration is with Google Home. Do I replace my beloved Amazon devices with Google Home to get the best integration?
First a rant: The fact that I should even have to consider this is ludicrous, and shows that both products are run by shit heads that won’t even feign the semblance of looking out for their customers interests. Because they have competing products they forcibly degrade any integration between the systems rather than integrating and differentiating on product quality rather than engineered lock-in. I despise this, it’s bad business, and completely unnecessary. I’d guess it actually stalls potential sales of both because people want to ‘sit back and see how it plays out’ before investing in one or the other.
I have a lot of sunk financial cost in my Alexa devices. There’s also some cost in time setting them up and integrating them with my other home-automation tools. That in mind I went back to the outcome I’m trying to achieve. My Alexa/Ring integration allowed me to see who was at the front door, and talk to them. My Alexa/Hello integration will only let me view the video. What’s my use-case? I use the integration to see the door, and decide if I should walk to the front door to answer. If it’s a package delivery, I can grab it later. If it needs a signature, I’ll see them waiting. If it’s something else, I walk to the door for a conversation. Basically I only use the integration to view the video and decide if I should go to the door or not. This means that Alexa/Hello integration, while not ideal, meets my needs perfectly. I easily chose to keep Alexa which provides the side benefit of not providing the evil behemoth that is Google any more access to my life than I already have. Last thing I need is my Gmail recommending male potency remedies after the Google device in my bedroom listens in on a night with my girlfriend. I’m picturing Microsoft Clippy here for some reason.
I’m much more comfortable with Amazon listening in and craftily adding some books on love making for dummies to my Kindle recommendations while using price discrimination to charge me more for marital aid purchases because they know I need them.
Ok, enough TMI, back to the point. Your technology sunk cost is killing you, mmkay? When making technology decisions for your company you should ignore sunk costs. Your rational brain knows this, but you don’t do it.
“Rational thinking dictates that we should ignore sunk costs when making a decision. The goal of a decision is to alter the course of the future. And since sunk costs cannot be changed, you should avoid taking those costs into account when deciding how to proceed.†https://blog.fastfedora.com/2011/01/the-sunk-cost-dilemma.html
You have sunk cost in hardware, software, people-hours, consulting, and everywhere else under the sun. If you’re like most these sunk costs hinder every decision you make. “I just refreshed my network, I can’t buy new equipment.†“My servers are only two years old, I won’t swap them out.†I have an enterprise ELA with them, I should use their version. These are all bad reasons to make a decision. The cost is already spent, it’s gone, it can’t be changed, but future costs, and capabilities can. Maybe:
- That sparkly $400,000 SDN rip and replace will plug far more cohesively into the VP of Applications ongoing DevOps project allowing them to launch features faster resulting in millions of dollars in potential profit to the company over the next 24 months.
- The new servers increase compute density lowering your overall footprint and saving you on power, cooling, management, and licensing over time starting a quarter or two down the road.
- Maybe that feature that’s included for free with your ELA will end up costing you thousands in unforeseen integration challenges while only solving 10% of your existing problem.
This issue becomes insanely more relevant as you try and modernize for more agile IT delivery. Regardless of the buzzword you’re shooting towards, DevOps, Cloud, UnicornRainbowDeliverySystems, the shift will be difficult. It will be exponentially more difficult if you anchor it with the sunk cost of every bad decision ever made in your environment.
“Of course your tool sounds great, and we need something exactly like it, but we already have so many tools, I can’t justify another one.†I’ve heard that verbatim from a customer, and it’s bat—shit—freaking—crazy. If your other tools suck, get rid of them, don’t let those bad decisions negate you from purchasing something that does what you need. Maybe it’s your vetting process, or um, eh, that thing you see when you look in the mirror that needs changing. That’s like saying ‘My wife needs a car to get to work, but I already have these two project cars I can’t get running, I can’t justify buying her a commuter car.’
Most of our data centers are built using the same methodology Dr. Frankenstein used to reanimate the dead. He grabbed a cart and a wheelbarrow and set off for his local graveyard. He dug up graves grabbing the things he needed, a torso, a couple of legs, a head, etc. and carted them back to his lab. Once safely back at the lab he happily stitched them together and applied power.
Data centers have been built buying the piece needed at the time from the favored vendor of the moment. A smattering of HP here, a dash of Cisco there, some EMC, a touch of NetApp, oh this Arista thing is shiny… Then up through the software stack, a teaspoon of Oracle makes the profits go down, the profits go down… some SalesForce, some VMware, and on, and on. We’ve stitched these things together with Ethernet and applied power.
Now you want to ‘DevOps that’, or ‘cloudify the thing’? Really, are you sure you REALLY want to do that? Fine go ahead, I won’t call you crazy, I’ll just think… never mind, yes I will call you crazy… crazy. DevOps, Cloud, etc. are all like virtualization before them, if you put them on a shit foundation, you get shit results.
Now don’t get me wrong. You can protect your sunk costs, sweat your assets, and still achieve buzzword greatness. It’s possible. The question is should you, and would it actually save you money? The answer is no, and ‘hell no.’ The cost of additional tools, customization, integration and lost time will quickly, and exponentially, outweigh any perceived ‘investment protection’ savings, except in the most extreme of corner-cases.
I’m not promoting throwing the baby out with the bathwater, or rip-and-replace every step of the way. I am recommending you consider those options. Look at the big picture and ignore sunk-cost as much as you can.
Maybe you replace $500,000 in hardware and software you bought last year with $750,000 worth of new-fangled shit today, and $250,000 in services to build and launch it. Crap, you wasted the sunk $500K and sunk $1 million more! How do you explain that? Maybe you’ll be explaining it as the cost of moving your company from 4 software releases per year to 1 software reease per week. Maybe that release schedule is what just allowed your Dev team to ‘dark test’ then rolling release the next killer feature on your customer platform. Maybe customer attrition is down 50% while the cost of customer acquisition is 30% of what it was a year ago. Maybe you’ll be explaining the tough calls it takes to be the hero.