The key part to this article is at the very end, and not addressed in the title (of course)....they extended their sales till the next day. Anyone who couldn't place an order (like me) did it later. So I'd argue it did not cost Costco $11 million...
So there are at least some lost sales by people shopping elsewhere. It's rare that a specific price is only offered by only one retailer for a product.
I suppose the difference with Costco is the sunk cost (membership fee) that means you're probably more likely to want to get your money's worth.
Especially in the era of Amazon shoving counterfeits every which way for every conceivable product, it's nice to have at least one retailer that isn't just hawking shelf space to anyone with a dollar and crap to sell.
Everything afterward is just a plus.
Kirkland Batteries are a win once again if they have your car size stocked
Other people probably just went directly to stores.
There's no way to measure lost sales with high accuracy, but that's not really the point, because it's most useful for modeling decisions and policies in a forward-looking context. Future revenue forecasts, the other term in the equation you're optimizing, are also inaccurate.
These functions are also generally "bathtub" shaped, and with the amount of money at stake, you can save quite a lot by being a little closer to the optimum, so they're worthwhile.
Store too crowded? Website slow or down? Something out of stock? Business not lost, people just come back later.
It’s very easy to postulate that people would go to another site (e.g. Amazon) immediately instead of waiting around to see if the Costco site would recover, especially since most BF deals are limited by both time and quantity. That alone makes the loss estimates feasible.
This is the part I was referring to as "not actually true". Fact: the site itself, which was loading although not working entirely, displayed a big banner saying this during the problem periods. It was hard to miss.
The rest is indeed just speculation either way. I doubt the losses are as high as estimated, but there's no way to say without inside data.