Commentary

As Online Spending Fell Off a Cliff, Forecasters Spent an Alarmingly Long Time Suspended in Midair

FTR:Online Spending

Have we hit bottom yet?

Trying to predict the future is a thankless job, especially when money's involved: Equal parts art and science, the value of a revenue forecast necessarily remains unknown until its predictions come true, or don't - at which point the world has probably moved on to the next forecast. Still there seems to be limitless demand for "industry outlooks," cited (usually without question) in myriad Powerpoint presentations and articles in the trade press. This makes sense psychologically, satisfying our natural desire for foreknowledge and a feeling of security - but is this feeling really justified?

The short answer is: "no." Take the example of Internet ad revenue forecasts. With the benefit of hindsight, it's clear now that almost every major forecast was far too optimistic about revenue trends in 2008 and 2009. What's more, revised forecasts issued during the economic upheaval of 2008 were still alarmingly wide of the mark - not only guessing wrong on future trends, but failing to reflect well-documented developments that had already occurred.

A survey of eight revenue forecasts issued between June 2007 and December 2008 (from Veronis Suhler Stevenson, Jupiter Research, the Kelsey Group, Lehman Bros., Magna, eMarketer, J.P. Morgan, and ZenithOptimedia) reveals that, on average, the crystal ball for Internet advertising revenue growth in 2008 was about 100% higher than the actual growth rate, with an average prediction of 20% versus actual growth of just 10%, according to the Interactive Advertising Bureau. Midway through 2009, it seems likely the forecasts will be even further off. On average the analysts predicted 17.5% growth in Internet revenues, but the first quarter actually saw negative growth with a -5% drop. True, the rest of the year could see a recovery - but to yield an annual growth rate of 17.5% Internet revenues would have to grow an average 25% in the second, third, and fourth quarters.

Of course it's easy to take potshots after the fact; a lot happened between June 2007 and December 2008 that even the keenest observers never imagined, including the collapse or near-collapse of dozens of banks (beginning with Lehman Bros. itself) followed by the sudden paralysis of international credit markets, and the government's announcement in December 2008 that the economy had been in a recession since the fourth quarter of 2007. Given these extraordinary circumstances, it's not surprising analysts had to revise their earlier predictions downwards, sometimes more than once.

What's remarkable, however, is how slow the prognosticators were to change their forecasts. For example ZenithOptimedia first predicted in June 2008 that Internet revenues would grow 21% that year and 18% this year, maintaining this prediction (in revised forecasts issued in October and December 2008) through the bank failures and the official announcement of the recession. In fact the research outfit didn't alter the figures until April 2009, when it finally slashed the online growth projection for 2009 by more than half, to 8.6%.

Zenith wasn't alone in hanging on to outdated forecasts. In August 2008 eMarketer said online revenues would grow 17% in 2008 and 14% in 2009, for a total of just under $29 billion by the end of this year - but in November revised that downwards to $25.7 billion in 2009, for a more modest 10% growth rate. Then in its most recent (April) forecast, the online projection for 2009 dropped yet again, to $24.5 billion, representing growth of just 4.7%. Veronis Suhler Stevenson has also had to revise its forecasts repeatedly. After originally forecasting an 18.6% growth rate for 2009 in August 2007, vss lowered the figure to 15% in August 2008, then slashed it a third time in February 2009, to 9%.

In the case of Zenith and eMarketer (and several other forecasters) their projections seemed to lag behind reality, failing to reflect clear macroeconomic trends. Most observers would probably agree that the failure of Lehman Bros. in July, the subsequent failure or near-failure of other banks beginning in September, and the official declaration of a recession in December were the main game-changing events of the last year - but they failed to influence forecasts until April.

Why the delay? There are several likely culprits. First of all, given the heavily quantitative nature of financial analysis and investment advice, forecasters are required to ground their predictions in historical data, which allows them to posit future developments. While this may produce accurate results in periods of long-term, incremental change, in a time of sudden transformation (ironically, when forecasts are needed most) the method is more likely to come up short. In this case, after several years of slow decline in macroeconomic indicators, in fall 2008 the bottom fell out of the American economy so suddenly that even the most recent data simply didn't correspond to emergent trends.

Of course, forecasting methodology does leave some room for intuitive leaps ("listening to your gut"), depending on how much leeway analysts are allowed by their employers and clients - the "art" mingling with science. But the fact remains that there was almost no reliable data on emergent market trends in Internet advertising for fall 2008; it's no coincidence that April's belated revisions all appeared just a few days after the Interactive Advertising Bureau released its industry revenue figures for the fourth quarter of 2008 on March 30.

Analysts are understandably leery of publishing forecasts based on mere speculation, and the combination of a lack of data, institutional inertia, and the inherent conservatism of most financial advisories kept Internet revenue predictions in an almost bizarrely happy place well after the situation had clearly changed; substantive revisions had to wait six months (in some cases) from the beginning of the broader economic meltdown. All this serves to as a reminder of the self-evident truth about most forecasts, which aren't so much predictions of the near future as they are reflections of the recent past.

Next story loading loading..