Earlier this week I argued that earnings seasonality and web traffic data produce a weak Q3 revenue signal for Blue Nile (NILE), an online jewelry retailer. There is another online retailer for which seasonality and web traffic data produce the opposite signal: photo publisher Shutterfly (SFLY).
Shutterfly will announce its Q3 sales in late October, as they have historically. The announcement will describe sales from July 2012 through September 2012. SFLY has given guidance in the range of $89.5-$91.5 million for Q3, and accordingly, the average analyst estimate is $91.5 million. However, there are two reasons to believe that the Q3 announcement will considerably exceed these expectations, as signs point to Q3 revenue above $100 million.
- Seasonality. Take a quick look at SFLY's quarterly revenues over the past eight years:
The seasonal patterns are obvious: Q2 revenue is always greater than Q1, Q3 revenue is always greater than Q2 and Q4 is always greater than them all. This has everything to do with seasonal gift giving of photo-related merchandise (e.g., Christmas, Mother's Day, etc.). The most conservative estimate of Q3 revenue can be inferred by the fact that in the company's history, Q2 revenue has never exceeded Q3 revenue, implying a lower bound on Q3 revenues in the neighborhood of $99 million. Using the typical percentage increase from Q2 to Q3 of 5.53% projects revenue of $99.02 x (1 + 5.53%) = $104.49. Again, compared to analyst expectations of $91.5 this would be a welcome surprise.
- Web traffic. Like Blue Nile with jewelry, Shutterfly sells all its products (photobooks, prints, cards, etc.) online. This means that we can use observable web traffic data to infer Shutterfly's potential customer base, much like counting how many people walk into a grocery store. And according to multiple sources Q3 has been a breakout quarter for SFLY web traffic.
Compete shows a similar trend:
To formally investigate the relationship between shutterfly.com traffic and Shutterfly revenues I downloaded the history of daily shutterfly.com traffic from Alexa as far back as I could go, which was Q4 2007. The daily web traffic is expressed as the number of Internet users per million. I then averaged the daily web traffic within each quarter so that my unit of observation was quarterly (just like my revenues). This gave me 19 observations of quarterly web traffic (from Q4 2007 to Q2 2012) that correspond to 19 observations of quarterly revenues over the same period. Using these observations, the correlation between quarterly revenues and quarterly web traffic was 87%. The rank correlation (Spearman) was slightly lower at 80%. In other words, web traffic does an excellent job at forecasting what announced revenues will be. With this is mind, I considered a linear regression model where the dependent variable (the one I want to forecast) is quarterly revenues (Revenues) and the independent variable (the one I know ahead of time) is quarterly web traffic (Traffic):
Revenuesq,t = a + b*Traffic
Where a is the intercept b is the coefficient on Traffic, using those 19 observations gives an estimated a of -$157.94 (t-stat -4.07) and an estimated b of 0.410 (t-stat 5.79) and an R-square of 76%. Using those coefficient estimates and the average web traffic in Q3 2012 of 811.01 users (per million) forecasts Q3 2012 revenue of -$157.94 + 0.410*811.01 = $174 million. Now I of course don't believe revenues will be this high but you can see why the regression is estimating a figure this high. If you look at the Alexa graph you will see Q3 2012 traffic is high enough that is resembles Q4 2010 traffic (this is the 2010 holiday season when revenues were $166 million). Looking at the Compete graph produces similar conclusions: Q3 2012 traffic is approaching that of Q4 2011 traffic. Suffice to say, I believe the web traffic data signal a revenue figure well above $100 million and, thus, well above the guidance and expectation of $91.5 million.
Some readers may be concerned that Alexa (or Compete or Google) provides a poor measure of site traffic. Keep in mind that if the Alexa numbers were poor I wouldn't have found a historical correlation of 87%. Garbage in would mean garbage out and I would have found a correlation close to 0%. It would also be highly unlikely that I find similar Q3 patterns with the Compete and Google data.
Other readers may believe that the increase in web traffic is coming from the Kodak Gallery acquisition which began migrating users from Kodak (EKDKQ.PK) to Shutterfly in July. I see this more as a clarification than a concern, as it is not clear at all why past Kodak customers would add no significant marginal revenue when augmented to typical Shutterfly traffic.
Additionally, our Alexa traffic reading for shutterfly.com increased 3% month over month from June to July, well before the results of the migration of kodakgallery.com were accessible to users (August). Extra traffic from the acquisition did not meaningfully spike until August 6th, about 3 weeks into the integration. Interestingly enough, since that point, the average daily traffic count on Alexa is 875, implying approximately $200 million in revenues (if Kodak converts purchased at the same rate as organic Shutterfly users) assuming the entire quarter was ex-transition. I won't presume such an aggressive conversion rate, but the numbers speak for themselves.
Reiterating the seasonality argument to the bottom line point: adding new users from Q2 to Q3 should only increase the likelihood that Q3 revenues are higher than Q2.
Recommendation: Given the seasonality and web traffic data, I expect revenues above $100 million for Shutterfly and thus a positive surprise at the October announcement. I recommend buying before the Q3 announcement or before Shutterfly provides guidance that Q3 sales will be higher than expected.