The Devil in Groupon's Details

It’s another bad day for Groupon: not only is Andrew Ross Sorkin using the company as Exhibit A in his opposition to the JOBS Act, but more worryingly the Wall Street Journal is now reporting that the SEC is examining the earnings revision which Groupon announced yesterday.

Vipal Monga has explained exactly what the problem is here, but his story is very hard to access online, so I’ll try to summarize. The issue at hand is that of refunds, and how they’re accounted for.

Let’s say that Groupon has managed to sell 240 coupons for “cool sculpting”, at $500 apiece. That’s a total of $120,000. The coupons expire on September 19, in six months’ time. Let’s also assume that, as per usual, Groupon keeps 50% of the proceeds, and gives the other 50% to the merchant. In this case, it would keep $60,000 for itself, and remit $60,000 to Dr. Aron Kressel. But Dr. Kressel wouldn’t get all the money up front. He gets one third, or $20,000, immediately. He gets another $20,000 after 30 days. And then he gets the final $20,000 after 60 days. That’s May 18.

Now, Kressel might not get all of his $60,000. Let’s say that some of the people who bought a coupon turn up for their initial consultation before May 18, and are told that they’re not medically suitable for the treatment and therefore can’t have it. Those people — let’s say there are 20 of them — are eligible for a full refund from Groupon. So Groupon gives those people back their money, $10,000 in all, and holds back from Kressel his $5,000 share of that money. As a result, Kressel’s final payment is not $20,000 but rather $15,000, and he ends up getting paid $55,000 in total by Groupon.

And at the same time, of course, Groupon’s own revenues from the deal are also reduced to $55,000: the economics of selling 240 coupons and refunding 20 of them before May 18 are basically the same as the economics of selling 220 coupons and refunding none of them.

After May 18, however, things change. At that point, Kressel is paid out, but Groupon still has the Groupon Promise. As a result, if anybody gets turned away from Kressel’s office after May 18, Groupon eats the whole refund. Let’s say that appointments become easier to come by after May 18, and a further 50 people end up being told that they’re not eligible for the procedure after that point. Remember that Kressel has already been paid $250 by each of those people, and doesn’t need to repay the money if he finds them ineligible.

Those 50 people still get their refunds from Groupon — a total of $25,000. But in this case, all of that $25,000 comes out of Groupon’s share of the revenues, and none of it comes out of Kressel’s cut.

So what’s the situation on September 19, when the deal expires? 240 coupons will have been sold, for an up-front total of $120,000. 70 of those coupons will have been refunded, bringing total revenues down by $35,000 to $85,000. And of those revenues, Kressel will have received $55,000, while Groupon will have received just $30,000 — a 65/35 split in favor of the merchant, rather than the 50/50 split originally envisioned.

And in fact it’s possible for Groupon to lose money on the deal, if there are enough refunds after May 18.

How is all this accounted for?

The way that Groupon does its accounting, it adds up its share of the gross revenues — that would be $60,000 in the cool sculpting example — and books it as revenue immediately, minus the quantity of refunds it expects to have to issue after applying a model which tries to predict such things. If you look at Groupon’s new 10-K, you’ll find this chart (click on “Notes to Financial Statements” and then “Accrued Expenses”):

The line you want to look at here is “refunds reserve” — the number which was $13.9 million in 2010, and $67.5 million in 2011. If you add up all of the deals that Groupon issued in 2010 — that’s some $745 million in total — Groupon reckons that it’s going to have to refund $13.9 million, or 1.87%.

Then, in 2011, a lot of things changed at Groupon. It sold a lot more deals than in 2010, for starters. It also moved into higher-priced deals, things like cool sculpting, which are more likely to be refunded. And it started selling travel deals, too, which are also more likely to get people asking for refunds, especially if they turn out not to be able to book travel on the days they want.

So in 2011, out of $3.985 billion in total revenues, Groupon reserved $67.452 million for refunds. Now note these are the revised figures, which were released after Groupon realized that its initial estimates for refunds were too low.

But do the math, and it turns out that $67.452 million is just 1.69% of $3.985 billion — the anticipated refund rate actually fell from 2010 to 2011. This does not make much sense, since by all accounts — including Groupon’s — it should by rights have gone up, quite substantially.

Now there’s an easy way of dealing with this problem, which doesn’t involve any predictive algorithms at all. Here’s Monga:

Forensic accountant Howard Schilit told CFO Journal that the mistake reflects a misapplication of accounting rules, in particular those outlined in financial accounting standard 48, as set by the Financial Accounting Standards Board. The standard dictates how companies are allowed to estimate revenue for refundable products.

Under the rule, companies are allowed to set aside reserves against potential refunds based on reasonable estimates. But Schilit argued that Groupon couldn’t “reasonably” estimate the refunds because it is so young and follows a relatively new business model. Lacking that historical perspective, the company shouldn’t have recognized any revenue until after the end of their refund period.

“Everything would have to be deferred revenue until the end of the refund period,” he said. “Either [Groupon's executives] didn’t know they had to defer, or they wanted to continue to show as much revenue as they could.”

This, then, is probably what the SEC is investigating at Groupon. If it sells 240 coupons for cool sculpting, should it book $60,000 in revenue? Or $50,000? Or $30,000? The fact is that Groupon doesn’t know how much if any money it’s going to end up making from that deal until the deal expires in September. So there’s a case to be made that the company shouldn’t book any revenue at all until September, just to be on the safe side.

What happened with the earnings restatement is that Groupon discovered that the refund reserve it had been using was too low; when it increased that reserve, it ended up losing more money than it had originally reported. But should it have booked any revenue at all, so long as that revenue was subject to potential refund? I have a feeling that the SEC is going to be asking Groupon that question in quite a pointed manner.

This post first appeared on Reuters Opinion.