No, I’m not talking about TPA 5100.38 – Subsequent Event Related to VSOE.  That just says you can do your VSOE analysis after the balance sheet date, as long as you support it with data that existed on or before the balance sheet date.  What I’m talking about below is what to do when you initially have to recognize an entire arrangement (product, license, everything) ratably because of lack of VSOE of fair value for either PCS or other services.

If you don’t have VSOE of fair value, paragraph 12 of SOP 97-2 actually tells you to defer all revenue “until the earlier of the point at which (a) such sufficient vendor-specific objective evidence does exist or (b) all elements of the arrangement have been delivered”, but then goes on to provide at least two exceptions:

• “if the only undelivered element is PCS, the entire fee should be recognized ratably”, and
• “if the only undelivered element is services… (for example, training or installation), the entire fee should be recognized ratably”.

So here’s the rule:  if no VSOE of fair value, then ratable treatment over the service period.  I get it.

But what if you establish VSOE of fair value in a subsequent reporting period?  Maybe, with another quarter or two of data, your population of standalone transactions at long last now demonstrates the magic 80/15 bell shape (see discussion of other ways to get VSOE here).

Certainly, we don’t go back and restate prior periods now that we have established VSOE of fair value in a current period.  But what do we do with the millions stuck in the huge deferred revenue waterfall slowly tricking into revenue over the service period?  Do we let that waterfall keep trickling, or do we now reassess those arrangements by applying VSOE of fair value to the remaining undelivered services and recognize the residual in the current period.

Answer:  there isn’t one (which means you get to make a policy choice).  When I chat with some knowledgeable Big 4 colleagues on this topic, about half say the newly established VSOE of fair value should only be applied to new arrangements, while the other half say that you should take the residual revenue in the current period.  One firm actually has come out and explained this issue in their commentary, while others seem not to object to either approach.  In any case, before you go materially altering revenue, ensure your audit team is on board with your choice as these are the sort of judgment calls that evolve.

Personally, I can see it both ways.  Paragraph 12 does indeed say defer revenue until you get VSOE, but doing such a “catch-up” of revenue in a period where nothing else fundamentally happened with the transaction seems like “mark-to-market” accounting for rev rec.

On the other hand, perhaps the exceptions listed in paragraph 12 are supposed to apply for the “life” of the arrangement without our being able to revisit the accounting determined at the outset.

Your call.  Just be consistent.

For those who don’t know what VSOE stands for, you probably won’t be interested in this posting.  For those who have forgotten because you are in denial that it is a burning issue you eventually have to come to grips with, it means Vendor Specific Objective Evidence…not industry data, not competitive information, or any other third party evidence, but vendor specific (i.e. company specific) evidence.

I don’t disagree with too many things in the revenue recognition world, but the thresholds to what “consistent” have evolved, while they may be quantitatively straightforward, are not useful in many circumstances, which means the resulting financial statements that get produced are also not useful (and I thought usefulness was an important objective in financial statement reporting, unless I’ve read the very first Statement of Financial Accounting Concepts incorrectly).

Let’s say a  company ships millions of dollars worth of product (for which software is more than incidental) to a customer and throws in a bunch of training for which they have either:  a) not yet sold on a standalone basis (because they are too busy shipping millions worth of product and have not had time to scale their training organization), or b) sold it on a standalone basis but not at pricing “consistent” enough (because of course they will give large and varying discounts to customers who are willing to buy millions worth of product).  In other words, no VSOE of fair value for training; therefore, spread everything over the period during which the training occurs (which could be years) – this is not useful.  Depending on the terms of the agreement, it is not infeasible that the customer could depreciate the asset faster than revenue is recognized (all because the company included a few training courses in the arrangement).  It also makes the financial statements not comparable to a competitor who just happened to have a handful of standalone training sales to assert the much sought after VSOE of fair value.  For investors, I would hope the question of “Did you sell training at the same price?” falls much lower on the relevant scale than “Are you doing all you can to continue to get millions of dollars worth of product orders, even if it means giving away or highly discounting other minor elements of the transaction?”

Let me digress, because it is not useful for me to continue to spout off my feelings without summarizing how to handle the interpretations of what constitutes VSOE of fair value we currently have to deal with.  Until someone blesses a more reasonable “but can’t be more than” approach, here is some direction about establishing VSOE of fair value.

Quite simply, paragraph 10 of SOP 97-2 (Software Revenue Recognition) defines VSOE of fair value as either “the price charged when the same element is sold separately” or “for an element not yet being sold separately, the price established by management having the relevant authority”.

Don’t get too excited about using your authority to set pricing because the fact that you may actually have that authority is only one aspect.  And it is not just the list price, but the actual price (after planned discounts) that has to be established.  Here is why it is rarely used:

1. it only is applicable for elements that you have not yet introduced separately, so it can’t be used to establish VSOE of fair value for that element you’ve been discounting all over the place for years;
2. it must be “probable” that the price you assert will actually be the price when you happen to sell it separately, so don’t plan on doing a last minute change to the price based on competitive pressure or other market conditions that you would normally do as a sound business decision (and the longer it is expected to take to sell it separately, the harder it will be to defend the probability);
3. you need to have a track record (preferably for similar products or services) of setting expected prices internally (usually marketing does this) and having those prices play out in the market place.

It is much more common for companies to establish fair value using one of methods that look at the price when the element is sold separately.  Generally, there are two acceptable methodologies in practice (or three if you have a particular firm as your auditors, and it is ridiculous to me that firms disagree on a concept that can have such an impact on financial results – comparability should not depend on who your auditors are!).

At any rate, let’s talk about the “bell-shaped curve” and the “stated price” methods. 

Reference to a bell shape is very literal as you want to price your elements in such a way that a large number of transactions (i.e. standalone deals for that particular element) fall within a consistent band (forming a shape like a bell if you remember your graphs of data sets from statistics class).  Some people call this the “80/15” test because the current interpretation of what could be sufficient is having 80% of the transactions fall within +/-15% of the median of the data.  A little stats lesson:  median = mid-point; mean = average.  Don’t use the average to calculate the range as that could mask the price you actually charge most frequently (which is what you are trying to get to).  On the other hand, don’t get discouraged if you get in the 70’s instead of in the 80’s (especially if the band of pricing is still narrow and/or there is an upward trend).  If you are in the 60’s, then it will be a challenge to assert you charge a consistent price.  However, some interpret consistency as being a “substantial majority” (generally, at least two-thirds), so maybe 66.7% is enough!  This is actually close to the mathematical three sigma rule (or empirical rule) which states that 68.3% of a set of values in a normal distribution lie within one standard deviation from the mean.  My point:  see what you have and don’t give up unless it is very clear you can’t come close to approaching 70%.

There are two versions of the stated price method:  one audit firm says just state the price in the order for which you need VSOE and that’s the fair value (i.e. quote what the price of that element will be if they buy it again).  You still have to support the idea that the price is substantive, and ensure you can demonstrate the likelihood of that future transaction even happening, but it generally is that easy.  However, if you don’t stick to that price when the time comes, that means you did not actually have VSOE of fair value back when you relied on that stated price to establish it.  This can lead to restatements to past periods, so it’s a pretty high risk endeavor, not to mention that it locks both the supplier and customer into a price sometimes years in advance (which, from a business perspective, is not usually desired).

The second “more accepted” approach to the stated price method is to state what the price of that element will be if the customer buys it again outside of the current arrangement (same as above), but use a population of stated prices in various orders to build support for the notion that to price the element at that level is indeed the customary practice.  And if you have at least one standalone transaction to back test that assertion, you may have VSOE of fair value.  Some explain this as applying the bell-shaped approach to stated prices in bundled arrangements.

So…lot’s of interpretations, most of which are difficult to apply if you are a start-up, introducing new products/services, pricing into new markets, or varying pricing for other valid business reasons.  Companies that are successful in establishing VSOE of fair value first recognize that the current interpretations are what they are, select an approach that works for them, put in place measures to limit flexibility in negotiations on elements that typically remain undelivered in an arrangement (e.g. support, maintenance, training, professional services), and drive towards consistent pricing behavior, usually with a strategy of negotiating the pricing, if necessary, on elements that typically are delivered up front (e.g. equipment, software licenses).

Learn more:

• Software Revenue Recognition
• Advanced Revenue Recognition Topics