The "Wash my Ferrari" Problem: A Meditation on Risk
theothereight.wordpress.com
theothereight.wordpress.com
The other risk factor is pure CYA. Nobody wants to prepare a costing worksheet for a customer quote where your cost is too low because you didn't factor in the various overheads. Only after the annual financial statements are made will you find out if you lost money on the product or not. Do you want to be the guy who buys at $10 and sells for $20 only to find out before bonus-time that overhead was $90 per unit? So you factor in the cost of capital, inventory hold charges, shipping delays, warehouse pallet transfer cost, foreign exchange currency buffer, and tons of other charges. This makes it appear that your cost is high and so it is only natural to charge more for the product. Boeing isn't going to sell an Arduino + LED light for $100. It will be $500 because of risk of breakage and be $5000 by the time all the CYA risks have been tacked on by six levels of middle-managers.
The wonderful thing about CYA risk factors is that they help justify your high prices and if you are able to be lean, they give you a terrific margin. THIS is what you want to disrupt. You can charge less than the incumbents because they are playing it way too safe as nobody wants to lose their neck for selling a product at a loss. But don't charge so low that risk of breakage ends your business. The good thing is that the risk of breakage is not as large as the CYA risks in most every costing sheet that I have seen. YMMV depending on the industry/product.
In most cases insurance is going to be more expensive than simply eating the cost of breakages, it's really something you have to protect against rare-and-expensive events.
If you are high volume/low cost or low volume/high cost this can be a very different situation than medium volume/medium cost in terms of impact.
I agree. I manage an ERP system at a pharma-manufacturer. I added a YMMV precisely because of the different situation - different impact reasons. My main point was:
Incumbent Sales Price SP1 = Cost of (Base + Risk + CYA-buffers) + Predetermined Margin
Incumbent Profits PR1 = SP1 - Total Actual Cost
Startups Sales Price SP2 = Cost of (Base + Risk) + Predetermined Margin
Startups Profits PR2 = SP2 - Total Actual Cost
Sidenote: While in a perfectly competitive market, the sales price should be determined by the intersection of supply and demand curves, instead of being a predetermined markup based on cost, in most contract manufacturing environments, all the costs are known, markups are expected, and thoroughly negotiated.
Effective risk management doesn't just mean that you protect yourself against "bad things". The ability to manage risks better than your competitors is a competitive advantage. I heard it best described as the brakes on a sports car. Good brakes let you take corners faster.
Put simply, any real risk model should probably avoid being a 'single number', but a range of probabilities. This models the distribution of possibilities much better and helps you make better decisions.
Around a year ago he hired a webdesigner to redesign his website, apparently the cost paid was around $20,000. Now bare in mind that this was for website design only (logo etc already existed).
The website itself was a relatively bog standard wordpress setup with about 10 pages of copy, a basic WP template and a handful of stock art images, no custom code at all as far as I could tell; I could have done a better job myself over a weekend most likely.
After hearing about this price I was curious as to who had been lucky enough to get hired for such a contract, turns out it was a designer who had been referred to him by others in the legal business.
So the question is, has whoever did this found a goldmine of customers who are willing to pay such huge sums of money just because frankly $20K isn't a lot to these people.
Or was it more to do with the risk of providing services to somebody who specialises in suing the crap out of people?
Then our customers started getting serious and regarded these events as unacceptable, just as they became statistically certain, and impossible to manage out. So costs have gone up, processes multiplies, checks increased. And now lean startups are nipping at our heels...
Think medicine.
I guess it's something to do with wilfully gambling vs. relying on your own abilities. They seem like very different things but somehow get classified together.
I mean, which would you rather have, assuming they were paying you the same money (and using the same capital resources, e.g. both of them use a full low-power dedicated server that costs $1,200 up front in parts and $20/month in power. Both use about as much bandwidth.)
Customer A. who is hosting email, web and ftp for his family, and maybe has a dev setup so she can test out the new webapp she's developing.
or
Customer B. who is running a website with $10,000 per day worth of sales.
Assuming they both have my phone number and can wake me up at 4am and yell at me, I can tell you that I am going to want a lot more money to give customer B the same service as customer A, because first, customer A isn't going to call me at 4am very often, and if I do flub something up, I can give customer A a free month, an apology and an explanation of what happened, and she is going to think I'm okay.
Customer B? man, customer B is going to wake me up every time there is even a little networking blip. And if I screw it up? they are going to have reason to be really angry, and possibly sue me for a bunch of money. a free month is unlikely to mollify them.
(also note, insurance might cover the payout if I get sued, but they certainly won't cover the time and aggravation, even in the best case.)
So yeah, I can see how hosting customer B would be exciting, but I'd want a whole hell of a lot more money to deal with those increased expectations.
If we want to go to the car wash example, when I was driving my maxima with unrepaired body damage? my neighbour caved in the rear passenger door with her land rover. "Don't worry about it," I said, "You did not significantly lower the utility or value of my car." - I mean, the whole thing was probably worth about as much as the bumper on her land rover.
If she had done the same to the new M3 in the next space over? you can bet her insurance would be making it just perfect, for a price that could have bought my jalopy several times over.
That's the thing. Nobody reads the legal bullshit until the knives are fully out and the lawyers are at the table. When you think about it, it doesn't make sense to spend the effort until then. I mean, you're talking about thousands of dollars of effort to understand a contract, and that doesn't make any sense on a contract that is worth two hundred bucks. Setting expectations is an important part of avoiding the situation where the lawyers need to come out and understand the contracts.
(Of course, this is why most contracts are as one sided as legally possible; there is no advantage to giving quarter, as the counterparty won't really read it until the relationship has soured and they are actively hostile.)
The issue hits hard in medical record IT. Losing control of 500 or more folks' medical records gets your name in lights here, and you're required to try to notify everybody who might be affected. http://www.hhs.gov/ocr/privacy/hipaa/administrative/breachno...
What kind of insurance could possibly cover a startup against the reputation cost of this? Not insurance that any startup could afford. Plus, the liabilities for misuse of the leaked data (identity theft, employment blacklisting of sick people, you name it) are unlimited.
So, a business that holds medical records for people is inherently a Ferrari car wash, unless the entrepreneurs can somehow persuade their hospital customers to bear the reputation risk. That's very hard.
If I were a photagrapher, I would likely want to charge more for a wedding then say the engagement pictures (or other portraits) which have a similiar amount of effort. If I have a problem with a portrait I might need to call them back in to reshoot. If theres a problem with the wedding pictures I'm stuck.
The workaround is better preparation, more people, more equipment, etc. All this costs money.
I think this model is just wrong and charging any customers more due to profiling should not be practiced by any company. Assuming that your formula is correct. Don't raise the price because of the probability and risk. You should train your employees and improve your quality of service, so that the probability is low enough, now you charge back the same $5 for a fararri carwash.
No, The consequences of faulty equipment or service are fairly acute–i.e. someone dies or something is damaged right away. I think long-term consumer risks like obesity and lung cancer are different than what the author is referring to.
Also, most class actions suits against McDonalds have been thrown out [1]. Malpractice lawsuits and the like are very numerous, on the other hand.
* >> I think this model is just wrong and charging any customers more due to profiling should not be practiced by any company*
That's absurd. The entire insurance industry is based on this. As is any financial product that takes your credit score into account.
[1] http://www.bloomberg.com/news/2010-10-27/mcdonald-s-obesity-...
"Please don't submit comments complaining that a submission is inappropriate for the site. If you think something is spam or offtopic, flag it by going to its page and clicking on the "flag" link. (Not all users will see this; there is a karma threshold.) If you flag something, please don't also comment that you did." - http://ycombinator.com/newsguidelines.html
I think I've only seen 2 or 3 spam posts on dead. And a few 'special people' whom are hellbanned. In a nutshell, the S/N does not get noticeably worse with showdead on.
That's a bad way to contribute to a discussion.
I didn't even realize your point was solely about the example. I thought you were objecting to the entire argument. Bad examples are more common than termites in a snowstorm and you should focus on the meat of the article.
To me not finishing something is a measure of persistent captivation. It's not the first time I fell below my minimum. After I wrote the the comment, I figured people would harass me about not finishing and that perhaps I was being unfair to the author if I missed something redeeming at the end. I finished it, wished to change the comment appropriately, but it had already been downvoted. I should have known that it's very easy to look antagonistic when you're not saying something positive but I did say explicitly what I objected to. I really was just hoping to help by nature of sharing my opinion and was totally open - hopeful even - that someone would change my mind rather than just being looked upon as some asshole who just wanted to be holistically critical.
Clearly if the amount of the coverage (again, in this artificial scenario) is less than you can "fix" a Ferrari for, then perhaps they should turn Ferrari's away if they are very risk-averse. (again, all this is in the artificial model.)
But if the insurance-company says, "We can increase your coverage to (cost of fixing the ferrari) for a difference of (difference)" then the only way they can possibly pay that difference is by lowering their margins on every car wash, or by discriminating and factoring that difference only into the appropriate washes.
all this is more correct than your claim, isn't it? (again, in the proposed model.)
incidentally, I consider the proposed model to be extremely braindead. A car wash is not actually going to pay for any damage, and if it did it certainly would never actually pay for a museum-quality paint job, and if it DID then the price would not be ten bucks per car wash, a shop that can afford these kinds of lavish customer catering and appeasement would be starting at fifty dollars or something.
What actually happens is there is a sign saying they disclaim all responsibility for any damage. What also actually happens is normal price discrimination.
so I think a clever answer to "Why are you charging me more just because I have a ferrari" is "only ferrari owners would care about a small scratch from a CAR WASH, and we don't get many of you, so we have to charge you more to pay for that there sign there (point to large disclaimer sign.) Ten bucks please."
I figured the probability isn't handled by the car wash itself but instead essential to derive the price of the quote. Surely they must give an insurer the volume of cars coming through, yes? What seemed unlikely to me is that the car wash must also show the types of cars getting washed. This would be something handled by the insurance company after making certain assumptions that factor into the risk.
Assuming that each car wash did pay damages (which as you mentioned they don't), do you think they would charge premiums for washing all luxury cars? The price to insure against damages seems like it should be fixed per car (even if based off volume). One month might be oversaturated with $30,000 cars and another month might have more $20,000 cars. Will the cost to insure the car wash against potential damages these months not be the same even when a higher saturation of more expensive cars incurs more risk for the insurer?
My answer to this forms my basis of why I didn't like the example, even artificially, which is that the price to insure every car is the same based on the likelihood of all cars getting damaged and the price to fix them (monthly insurance cost/amount of cars). The insurance cost has no respect to which cars go through the wash - probably because it's too much of a hassle for the insurer to police which cars are actually visiting and making a more accurate calculation of risk. Instead, they assume every wash has a certain amount of luxury cars - even if they don't. Obviously this wouldn't hurt margins if they accepted the cars that are covered but it doesn't stop them from raising margins, which they still might do.