Why GE Digital Failed
inc.com
inc.com
Regarding digital transformation, this should not come as surprise but there is an incredible tension and misalignment between a GE core pushing for a platform and different businesses.
The issue is that businesses have roadmaps where they need to release new products each year, and it does not take into account having to rewrite everything or huge part from scratch to get some longer term platform benefits. So in the end the adoption is an issue.
I joined GE just before Predix introduction, and no one in my business rushed to adopt it. As a matter of fact, now that GE digital is going to be divested, that was likely the smart move.
Moreover, the product I am working on, is based on the "GE platform" effort of a few years prior. Platform that has no longer any official support (the maintenance team is one guy somewhere in the world) and that has re fragmented in every different product it was used in. The lack of support is not really an issue because the framework has been battle tested but on the other hand, there is no obvious benefits coming from having an "almost-common" core between applications or devices. So as business it makes you wary to adopt the last idea coming from GE digital or GE central, event it makes absolute sense on paper, because you have no idea how long they will commit to that idea. This is really a chicken-egg problem, because no one wants to use a zombie platform, and no one wants to spend support on a platform no one uses.
Maybe on a more positive note, even if I have not seen much direct impact on GE digital products, GE SW culture has changed. When I joined, SW was kinda neglected, more seen as a liability than anything, and now at last we have moved to more modern tools with CI pipelines, focus on devops and so on. But this is more like a grassroots effort. But I don't feel anymore that we are using a sw stack and tools totally outdated.
The team I work on was never sold on the benefits of the platform, but we were told to use it. Arbitrary targets like "get 150 apps to run on Predix" were set, but because there were no clear advantages to doing so, these targets were internally de-prioritized compared to working on features. They spun the numbers to give the appearance of momentum, but over time people talked about the platform less and less, and it became obvious that the emperor had no clothes.
GE has pockets of good developers; the grassroots culture that you mention is evidence of that. I'm hopeful that Flannery's renewed interest in the business verticals will enable those developers to focus their effort in places that actually deliver value, instead of wasting it on grand top-down initiatives to become some kind of industrial AWS. The sooner that happens, the sooner the company will get out of the miserable slump that it's been in.
I have a friend that works in mechanical maintenance. Heck, i used to work for him while i was putting myself through automotive trade school. Mitsui, Mitutoyo, Kuka, and Motoman not to mention Fanuc are all household names for industry automation based on a well defined platform of experience dragging back more than 30 years. Hell, even mitsubishi controllers are massively more invested in tooling and automation than GE ever could be.
And, they listen to customers...theyre open by default. I can plug motor control data and Fanuc probe logic systems into my Infinity QS SPC software and get accurate runout and ID/OD grind data in REAL TIME. no licenses required because in machining or component part shops that are large enough, its more important to meet JIT deadlines than it is to pay licenses (they get paid after the fact, and vendors are happy to subscribe to this model.)
The only GE stuff we used was in the substation transformers used by the building to feed power to the cymbal cutters and CNC systems in the shop. never networked, because it didnt have to be.
Don't even think about ANY open source components or anything of the sort found in the software industry.
Even the fiber interconnects are open protocols (regardless of how often implemented they are).
At the low end, you have modbus for everything (yes, it sucks latency wise even over tcp, yes, people can't number crap reasonably, etc. It works in the end).
Ethernet/IP is not a bad step up (I use it with 0 rockwell components). All my VFD's can be controlled with the same PLC code because of OVDA standards.
Ethercat is also fairly nice (and higher end CNC machines often just have everything from I/O to servo's on a single ethercat bus). It's more complicated, but i expect that will change over time.
Right now I have no problems connecting my siemens, mitsubishi, etc servos, field i/o, pneumatic valves, etc together over Ethernet/IP. Even lower end PLC's from automationdirect support it just fine.
(I've done Ethercat a few times as well, it's just harder to deal with on the software side ATM unless you go with beckhoff stuff)
Short term profit seeking is a cancer upon us all, is there no way out of this hell?
--------------------
Tuesday, June 9th, 2015
John told me that the board of directors had drawn up monthly sales goals for him. Starting in August, he would be expected to hit his quota. I thought this was insane. Once a product exists and is stable, then a company can draw up a sales schedule. How can one reasonably do that when the product does not even exist yet? Especially if the product is a cutting-edge technology which carries a lot of unknowns? For a stable company with an existing product, deadlines need to be more than mere aspirational goals, but when building truly original technology, then the entire company is aspirational — until the technology is working, there is no proof that the technology can work.
Even if the glorious day arrived when we would finally have an app customers could install on their iPhones, that would only be the beginning of a long process. Customers are an endless surprise. I’ve worked with startups for sixteen years; I know this well. Whenever I have shown people new software, the features that seemed intuitive to me were counterintuitive to them. Real-life needs that seemed intuitive to them seemed strange to me. If John thought that we could create our apps and have them working by mid-August, and he could immediately go out and start making tons of money, then he was sadly mistaken. If the board of directors thought that, then they were being badly misled.
https://www.amazon.com/Destroy-Tech-Startup-Easy-Steps/dp/09...
Take your firm private with investors that are focused on long term goals, not short term profits.
Why bother with public markets if private markets provide sufficient capital to execute?
[0] https://www.bloomberg.com/news/articles/2018-08-09/how-elon-...
The stock market doesn't require short-term thinking. But it does create a default focus on short-term performance when leadership does not articulate a plan.
That of course meant that internal adoption of Predix was negligible unless mandated by higher ups, and if your own company isn't going to adopt and improve the product, I sure as hell don't see why anyone external to GE would.
Apart from that, GE Digital was a fundamentally mismanaged effort. The people in charge seemed more interested in talking the talk rather than walking the walk. Most people at San Ramon seemed very disillusioned and their frustrations were out in the open during any of the company-wide all-hands meetings with a Q&A component.
This is common in very large companies because an easy way to justify a new business unit it to claim it will pay for itself by charging other business units as well as external customers.
Somebody should start an ETF which disinvests in B2B companies that run bullshit ads.
Their HR website was all flash, around the time when it was obvious to any competent engineer that flash was going the way the dodo. It also didn't work. (It was one of those websites where someone spent a lot of time concentrating on graphics but didn't really know how to get it to work.)
"Bad smell" was an understatement.
I wouldn't judge a company by its HR website. It's likely a relatively turn-key product that was chosen by a non-technical department and then neglected.
For all those years in which the Archer Daniels Midland corporation ran ads on Sunday morning political talk shows, ADM got to scam the government and consumers for billions thanks to subsidies they were able to get.
Derisively, my impression is that an organisation like GE will have to go through garbage collection while one (or why not more) digital native organisations take its place, in order for the new organisation to be a digital native. But I have no idea if being a digital native is even necessary. Most GE business units seem to be doing just fine with their accumulated capital of all kinds and their economies of scale.
I don't see this kind of clarity in some other companies that try to transform themselves. I think it's quite rare.
Not long ago AI was such a taboo word that people were avoiding it at all cost, because pronouncing it was a sure way to drive any kind of funding away (funny how quickly these things change). So anything yielding tangible results (machine learning, natural language processing, image recognition, etc.) was "not AI". AI were those crazies trying to put the human brain on a chip. My point: because Netflix underwent a successful digital transformation doesn't mean it's not a digital transformation, it just means that it's possible to describe it in a much more specific way.
Same in the heavy industry: Boeing/Airbus underwent crazy digital transformations, but they are not called that way because it's possible to be much more specific: simulations when it comes to design, 3D printing on the production side, etc.
As an aside, I do think the "digital" part is not the most significant part of Netflix's transformation. In pursuit of "we deliver entertainment on demand," Netflix started creating entertainment. That is a bigger cultural shift for a company than creating a streaming service. (Although their streaming service is an impressive and significant achievement.) But if they hadn't started creating entertainment, I don't think Netflix would be as relevant as they are now, nor would I think they would have a future.
If your plan is "provide streaming video" you'd call it that instead, thereby showing you know what you're trying to do, what success means for the project, and how it will produce business value.
"We believe by transaction counts we’re in the top five of e-commerce companies in the world. We know we’re behind Apple and Amazon, but we may be as high as third."
Domino's has wildly invested in their tech: https://www.marketplace.org/2015/09/24/business/corner-offic...
It was fascinating watching it play out. They put a hard deadline of when stores needed to upgrade from the decades-old DOS system, and in the case of my store at least, forcibly transferred ownership to a different franchisee when the original owner refused to make the investment.
The new system was a completely bespoke store management system. POS, call routing, back office management, driver dispatch, scheduling, inventory management, commissary ordering, line management, and online ordering all integrated into a single system.
It even had built in adaptive labor management, using historical volume data to estimate expected scheduling needs, as well as live trend data that told you when to cut drivers. Which was rarely accurate, because it used average delivery times as an input, but corporate used average delivery times as an audit metric, so most stores gamed the dispatch system to goose their corporate audit scores (with the side effect of breaking the accuracy of all the labor management functions of the new system).
On the backend, it enabled a lot of neat data analysis that corporate could use to identify shady shit at franchises, like 2 minute long phone calls, an order that was punched in but never completed, yet inventory coming up short for that exact "canceled" order.
Having worked at a store that actively resisted the changes, then moving to a store where the franchisee embraced the new capabilities and shifted incentives to increase adoption, gave me an incredibly vivid view of how change management impacts projects. One of the experiences I could have asked for that earlier in my career.
a. enabling employee experiences on Mobile (typically intranet apps) b. enabling Customer experiences (apps that enable customers for customer experience - eg. Mobile banking, flight booking, etc) c. partner ecosystem apps
one of my customer had 50+ employee apps & has over 160K employees. Apps were built on IBM's platform and were managed by airwatch agent. This strategy of the mobility program had multiple issues arising mostly from end user perspective.
1. App discovery was a challenge (high pain, low or unknown gain - for users to discover any app at all) 2. installation fatigue (imagine downloading multiple company apps) 3. erratic app lifecycle management led to higher upgrades 4. airwatch agent sucks battery 5. Intrusive nature of airwatch agent on personal devices
Our approach to solve this challenge:
1. Provided a microapp architecture based one app container (a sandboxed cordova + enterprise plugins baked in the container for hosting multiple microapps. monolith was leading to development of multiple binaries otherwise). 2. added a universal inbox with actionable links (this created a way of soft discovery of new microapps) 3. SSO enabled all microapps 4. cloud based console for managing the micro app lifecycle was provisioned. This gave a mechanism to hot load web assets and retire them. 5. Enterprise grade security was added on the container & this led the customer make provisioning of app with airwatch optional. 6 making web apps of office productivity apps available to be in line with the info sec guidelines.
Observation post roll out (we are now ~30 months into the program): Adoption - over 100K users enabled. Over 50k+ monthly active users
a. making airwatch optional - led to reduced user resistance for adopting the app b. Simpler life cycle management of just the microapps - made it less cumbersome for the users c. cutting down of time to build & manage subsequent apps
Full Disclosure: I am part of the founding team of DronaHQ - a containerised low code platform, which has been implemented at customer end.
I am quite amazed the closed mindedness of some of the people commenting here
1) Incumbent IT/Dev teams eating millions of dollars creating roadblocks every where for the newer teams
2) Expected to bill the newer platforms to existing business divisions
3) Red-tape and bureaucracy all over the place - Loads of approvals needed to do something simple like setting up dev environments
4) Top down planning for 'transformation'. No inputs are ever sought from new teams on anything
To sum it up -> doomed before even starting
With senior leadership help firewall and replace can work. That is, firewall support for existing systems to old team and make new team build new setup. But it is really hard -- old guard will fight to kill new process after every mistake and new teams will try to kill existing systems prematurely.
But even when the old system is decommissioned there is no need to fire the old team. It does happen, but because everyone who is capable of finding a new project moved there and whoever is left cannot or doesn't want to do anything else.
Big businesses have money.
They can create small teams easily.
Creating a short term team with not much money is in the power of every company in the fortune 500, heck, it's in the power of the top 10,000 companies...
And yet we do not see these teams doing these transformations. Shure we see teams creating "friendly digital employee pals" and "Machine learning driven employee survey feedback tools" but what we do not see is small, underfunded teams rapidly changing the face of business.
Because... it's hard. In reality, although executives love the message "small team, no money, fast" the translation that they should be using is "a way of wasting a bit of money and time doing something that will definitely fail"
I had an opportunity to work in a small digital transformation project at one of these "big businesses". When they had me come in every day and do basically nothing for 2 weeks before they could check all the boxes to ship me an engineering computer, I got the first big glimpse of how big business bureaucracy can doom these sorts of lean teams to failure.
Though I've never worked on these sorts of enterprise transformation projects (at least not at this scale), it's pretty interesting how the internal dynamics/politics can affect them and cause most of them to fail. Normally, my takeaway is that you can learn a lot more from failed projects. But this sort of project ends in failure so often, I think the reverse is actually true (for once).
Consider the (resoundingly) successful example of Disney's Next Generation Experience (NGE) project team, which developed what was eventually known as MyMagic+ at Walt Disney World.[0][1] They were basically given carte blanche to interrogate every aspect of the vacation experience to reinvent it, settling on using RFID-equipped "MagicBands" that would replace tickets, photo cards, credit cards, etc. There wasn't an aspect of the vacation experience, or a department within the Parks division, that wouldn't have been radically changed by NGE. When the board approved the $1bn project, Bob Iger told the executive in charge of it "this better work." There was infighting, turf wars, and everything else; it's not that people wanted it to fail, but everyone had their own idea of how to define success. And yet, despite everything, it succeeded beyond even Disney's collective imagination.
These projects might be damned infuriating when you're working on them, but they're fascinating to read about from afar.
0. https://www.fastcompany.com/3044283/the-messy-business-of-re...
>Then, one year later, in 2014, GE released a press release that said it was generating more than $1 billion in revenue from productivity solutions, highlighting Predix.
Like the article says, the vast majority of this $1 billion came from counting the revenue from any product that utilized Predix/GE Digital resources, no matter how much they contributed. I also never heard of any outside company that utilized GE Digital resources, but that doesn't mean that it didn't happen.
I've had team members quit a few years ago to join GE Digital, lured by crazy salaries. Too good to be true.
For GE, this was a spreadsheet initiative. Some SVP added a few lines to spreadsheet, opening headcount for a few thousand people. Same SVP would lose interest a few years down and would simply delete those lines again. Nothing more to it, this is corporate 101 on this level (see Siemens, HP, others for the same stuff).
Sucks for people that adjusted their lifestyles to those salaries, but it was clear, from the beginning. Hiring a ton of mediocre talent with top pay always mean trouble.
Even GE still struggles to understand software. It gives me hope that software engineers will become the next class of power players and the whole thing isn't going to get subsumed into finance like everything else. Even Apple is falling prey these days to finance-thinking.
Source: Multiple family members have worked for GE
If you have a factory moving on a pre-electricity prime-mover (probably a coal engine) and distribution system (probably axes and belts everywhere) and some one comes and pitches you electricity, you don't see the advantages. You see the costs of putting electric motors in all your machines, hiring all these expensive electric engineers and technicians, and basically rebuilding your factory for electricity. I see a similar dilemma here. Most industrial units will have to be rebuilt to take advantage of industrial IoT.
Wtf does that even mean.