Sleeping With the Enemy Part IIPosted: April 5, 2012 Filed under: Corporate strategy, economics, incentives, networks, Vertical integration 8 Comments
In an earlier post, I noted Target’s costly decision to end its on-line outsourcing arrangement with Amazon’s cloud service and take all its work in-house. The short-term costs were considerable, both in direct outlays and in performance degradation, and the long-term benefits were hard to pin down. Vague paranoia rather than careful analysis seemed to have driven the decision. I pointed out that firms often seemed unwilling to “sleep with the enemy,” i.e. purchase critical inputs from a direct rival, but the case for such reluctance was weak.
A few months ago, an apparent counterexample popped up. Swatch, the Swiss wristwatch giant, decided unilaterally to cease supplying mechanical watch assemblies to a host of competing domestic brands that are completely dependent on Swatch for these key components. These competitors (including Constant, LVMH, and Chanel) sued, fruitlessly, to force Swatch to continue to sell to them. The Swiss Federal Administrative Court backed up a deal Swatch cut with the Swiss competition authorities that allows Swatch to begin reducing its shipments to rivals. The competition authority will report later this year on how much grace time Swatch’s customers must be given to find new sources of supply, and these customers may appeal to the highest Swiss court. For now, Swatch’s customers are scrambling for alternative sources of supply in order to stay in business. The stakes are especially high because overall business is booming, with lots of demand in Asia.
(Before getting to the strategy questions, I have to get one thing off my chest: Why would anybody choose to buy an expensive mechanical watch in the 21st century? Higher priced, less reliable, less accurate, less durable, requiring more attention to keep running, and with identical visual aesthetics–mechanical watches seem completely inferior to quartz movements [or even the old Bulova tuning fork technology]. I imagine mechanical watch users wearing monocles and writing with quill pens as they groom their powdered wigs.)
Having completed my off-topic rant, does the Swatch episode provide an object lesson justifying refusal to sleep with the enemy? Would the bevy of small Swiss mechanical watch producers have been better off all along to have been vertically integrated into designing and manufacturing the mechanisms? Was Swatch’s mechanism business really analogous to Amazon’s cloud services? A few points, in no particular order, that might help us distinguish cases where sleeping with the enemy might be dangerous.
1. Adding incremental production capacity for Swiss movements is more expensive and difficult than adding server capacity in a cloud computing system. Nick Hayek, the head of Swatch has stated publicly that he wants to allocate more of the firm’s production capacity to its own brands such as Omega, Longines, and Breguet. He has professed weariness with Swatch being treated as a mechanical parts “supermarket” by its rivals/customers, “forced to deliver to everyone whatever they want.” (Apparently, Swiss competition law, until the settlement, treated Swatch as a monopoly supplier because of its 70% share of all Swiss movements and constrained management’s freedom to tell customers to buzz off.) The situation is very different with Amazon’s cloud services; the company acts as if it believes that there are strong increasing returns to serving more of the market and it aggressively solicits outsourcing contracts.
Lesson: Sleeping with the enemy is riskier when the input’s production capacity is scarce and costly to expand.
2. Swatch’s “monopoly” position in manufacturing Swiss movements is a legacy of its rescue and consolidation of a collapsing industry in the 1980s. In past years, Swatch has repeatedly urged its rival/customers to invest in their own production capacity, to little avail. By contrast, Amazon has nothing like a monopoly in cloud services and Jeff Bezos has gone out of his way in interviews to discourage companies from trying to build their own infrastructure, likening it to a firm having its own power plant.
Lesson: Sleeping with the enemy is riskier when the supplier is not committed to the outsourcing business.
3. Swiss watch manufacturers are limited in their ability to find alternative sources of supply by their need to claim that their watches contain “Swiss movements.” This designation is only legal for products actually produced in Switzerland, preventing the buildup of a lower-cost supply source in places like Poland or Taiwan. Thus, the original decision to sleep with the enemy was probably pretty rational–the present value of free-riding on Swatch’s capital investments likely exceeded the disruption costs the firms now face.
Lesson: Even when sleeping with the enemy is risky, it may still be worthwhile to play out the string as long as possible.
Hmmm. Do you think the Target/Amazon and Swatch/Constant, etc. situations would make good instructional cases?
The two are diametrically opposed examples. The Swiss movements is restricted by a country while the amazon cloud is not.
Anyone could setup a fully function, identically functional and reliable service as Amazon AWS. No one has the patents, skilled labor, and long term operational experience of Swatch. Again, complete opposite scenarios.
And finally.. Swatch’s complaint is that that want to have their cake (distributor) and eat it too ( direct sales channel). The two are completely different businesses. While having both under one roof may allow the patent management to consolidate the margins of each into one profitable margin, they must be considered separate for the needs of monopolistic regulation and real business accounting.
In short, I think that this is a horrible example of “sleeping with the enemy”.
I’m not sure if we disagree. The premise of the post was that both the Target-Amazon and Constant-Swatch examples featured one company sourcing a key input from a direct downstream rival. Both the small Swiss watch companies and Target were “sleeping with the enemy” by this definition. The evaluation of that strategy in the two cases might differ due to the factors mentioned by both of us.
I don’t follow your last point about Swatch at all. Many firms produce an input both for internal use and for sale to outside customers. Setups like that are very common in the chemicals business, for example, and are not at all a matter of trying to have one’s cake and eat it too. The difference here is that a) Swatch decided it wanted to cut back on its outside sales of the input and b) alternative supply sources are hard to come by. Swatch decided it wanted to eat more cake and have less.
Balfour.com is realizing a significant savings on costs and a significant downtime reduction from it’s importing of EC2 instances into an in house VMware and OpenStack cloud setup.
It depends on your existing expertise and infrastructure. A large company such as Target can easily absorb the necessary professionals and deploy a similar stack that saves then a LOT of money against the AWS billables.
1) As ahzz notes, some of this is just about vendor relations. Cloud technology is being commoditised and if you’re big enough then you can get better deals by switching away from Amazon. I’m not sure that this contract is truly “taking things in-house” – a lot of the expertise is being sourced the Sapient/IBM through contracts that still look a lot like outsourcing.
2) AWS contracts are notoriously stacked in Amazon’s favour, esp. over uptime. I’d imagine Target is big enough to get better terms than many, but the contracts reflect that EC2 zones do go down and when they go down Amazon’s priority (naturally enough) is to keep Amazon.com running smoothly. This again is not so much about “Sleeping with the Enemy” as much as Amazon basically not being as good an outsourcing partner as their PR suggests. If you search Quora for “EC2 downtime” you’ll get great advice about engineering around Amazon’s failure modes – but I can well imagine that for some services (and I can even imagine some scenarios for Target offhand) that those particular workarounds are not ideal.
I guess overall, I’d contest the idea that careful analysis wasn’t present here. I’m much more aware of the nitty-gritty around AWS than when you made your first post, because I’ve had to explore it for a new business opportunity. So then I had no reason to dispute your post, but now I can seem more commercial reasons why a company might want to switch away from AWS.
Thanks for the specific information. The perils of relying on news media accounts is that the reporting usually can’t go that deeply into the nitty-gritty.
I do note that Target had big and unanticipated problems in making the changeover, so their decision could not have been a slam dunk. And most analysts thought the decision was based on the concern about sleeping with the enemy, not operational problems. That was the thrust of the ZDnet coverage back in 2009.
Steve – I don’t have any concrete evidence to say that Target were not in fact obsessed with the “sleeping with the enemy” problem. So I can’t gainsay you there. You can build a good instructional case out of this for students around this, I’m sure.
Here’s a Target statement quoted on ZDnet that bolsters your point:
““Amazon has been an important strategic partner since we re-launched Target.com in 2001, and the strength of Amazon’s technology and fulfillment services has been a contributing factor in Target.com’s success. However, to deliver a customized multi-channel experience for Target’s guests, we believe it is in Target’s best interest going forward to assume full control over the design and management of Target’s e-commerce technology platform, fulfillment and guest services operations.””
“Full control” – sounds kind of paranoid, perhaps.
So let me pose another question to you. If you were Target and you wanted to develop new fulfilment and guest services that Amazon did not provide in 2009 – is that a concern about sleeping with the enemy, or is it a typical customer-vendor outsourcing relationship problem?
I think it’s entirely right to question whether Target’s new platform will actually provide them with competitive advantage over other store retailers (let alone Amazon itself) but my point about the nitty-gritty was that I don’t think it’s outlandish to say that Amazon isn’t as flexible as ZDnet reports make out. As such you can definitely imagine a chain of events like this:
1) Target does analysis on the future of retail for the next 5-10 years.
2) Target identifies new approaches in fulfilment and guest services that are going to help it prosper in the future.
3) Target asks Amazon about it’s roadmap for service development.
4) Amazon responds with a roadmap that doesn’t match Target’s requirement, because they have lots of customers who have other priorities.
5) Negotiations ensue, but costs are a sticking point.
6) Target’s IT group makes the claim that they can deliver the future at a lesser cost than Amazon.
7) Board decides to back the IT group and give it a go…
I guess my train of thought is inspired by some insights I recently had into some outsourcing deals which highlighted that sometimes the vendor and the customer are just evolving in different directions – but for partnered service design that can be a huge issue.
Absolutely, the typical sourcing scenarios have to be analyzed in these situations. A classic Williamsonian analysis dealing with cospecialuzation and mutual adaptation would be relevant, among other things. The less separable the customer’s unique needs are from the specific choices of the supplier, the more likely it is that the customer will need a tailor-made solution, necessitating either a non-standard contract/understanding or a vertically integrated setup.
Perhaps the thought experiment is to consider two outsourced solutions that have identical roadmaps and capabilities, but one of them is a downstream rival. How big a premium could the non-rival supplier charge and still get you to sign up with them? That kind of ceteris paribus exercise conceptually gets at the issue.
An interesting sidelight here is your suggestion that Amazon’s service isn’t especially customer friendly. There’s a possible pattern here, where the company is obsessed with great service in its dealings with end users (consumers) but takes a decidedly less accommodating attitude in its B-to-B relationships. That may be an inappropriate carryover from its tough stances with suppliers and distribution partners to the new cloud business; their organization and culture may be classifying clients as “companies” rather than “customers” and so applying the wrong set of norms.
I’m thinking the contrast might be a bit forced. Swatch seems more trusting your prisoners than sleeping with your enemies.