Is Netflix Still Over-Promising? The Bloodbath May Continue

Oct.26.11 | About: Netflix, Inc. (NFLX)

Netflix's (NASDAQ:NFLX) recent earnings announcement has been widely viewed as a disaster. Several Seeking Alpha authors are looking pretty smart about their predictions of the company's demise (I, II, III). Things are so bad that some are talking about a possible bankruptcy.

At risk of piling on and/or kicking a dead horse, this article will echo those sentiments and offer some new perspective on how bad things are.

Figure 1 shows Netflix's weekly stock price open, quarterly domestic paying and free/promotional subscribers and revenues from 2004 to the present. These data come from Netflix's SEC filings and refer, when applicable, to domestic subscribers only. Stock price data come from Google finance. All data in figure 1 are scaled and expressed as a percentage of their values in the first week in the series (e.g., Netflix reported 1842 thousand paying subscribers at the end of 2004Q1 and 2135 at the end of 2004Q2, so the scaled value in Q1 is 100 and the scaled value in Q2 is 116).

I want to highlight what I see as two sides of a huge potential problem in figure 1. Since the end of 2010, Netflix has been working hard to get new subscribers. It appears they ramped up their offerings of free subscriptions in the hopes of enticing people to sign up and stick around as paying subscribers. Note that figure 1 shows a steep drop in the uptake of these free subscribers in the most recent quarter. This is probably attributable to the negative public relations in the recent quarter and people not taking the company up on its free trials.

Figure 1. NFLX revenues, paying domestic subscribers, free domestic subscribers, and weekly open share price.

Click to enlarge
Click to enlarge
In the past, Netflix's promotional campaigns and free subscribers have been highly successful in generating new business for the company. This suggests that the current drop may be especially problematic for subscriber growth in the upcoming quarters.

Figure 2 summarizes this sentiment. It shows a pretty striking relationship between the number of free subscribers in the prior quarter and the number of paying subscribers in the current quarter (blue font). For the statistically interested, the r-squared of the regression line is 0.94. What is scary is what happens when we use these historical trends to predict what the 2011Q4 subscriber numbers will be.

Based on the 946,000 free subscribers in 2011Q3, we would expect roughly 20 million domestic paying subscribers (to both the DVD and streaming plans). This is shown as the red font in the figure. As a point of reference, that is at the low end of the most recent guidance offered by the company (20.0m - 21.5m, shown in black circles) in its recent 8-K for domestic streaming subscribers only. (Note that this guidance appears to be based on free and paying subscribers). To the best of my knowledge, the company did not offer a breakdown of the expected overlap in Q4 plans, but I assume that they probably think that unique users (the comparable metric to the past results) will exceed the streaming subscribers by some margin.

Figure 2. Relationship between current quarters' paying subscribers and prior quarters' free subscribers and what this implies for 2011Q4.

Click to enlarge

Click to enlarge

The decline in free subscribers is worrying, but it is only one data point and thus tough to make investment decisions off of. Figure 3 considers the issue from another angle, showing the average volume of U.S. based Google searches for the term "Netflix" in the quarter alongside the end of quarter (EOQ) domestic paying subscriber numbers. It is scaled in the same fashion as Figure 1.

I assume that people leave digital footprints on the web, footprints which, even in aggregate, can be very useful. One potential mechanism for this is that many people search for Netflix rather than typing the name directly into the address bar or using a bookmark (my mother does this all the time). If this is true, then the relative number of Google searches for the term Netflix should give some indication of the number of paying subscribers. Of course, many people access Netflix on third party devices and do not search for the site directly (and not everyone uses Google), but this does not mean that traces of subscriber behavior cannot be seen in search queries.

Figure 3. Trends in domestic paying subscribers and weekly Google searches for "Netflix."

Click to enlarge
Click to enlarge
Beyond basic correlation, figure 3 shows three other interesting dynamics. First, in later periods, starting around Q3 2008, the number of searches show small peaks right around quarterly earnings release dates (these can be seen in the two peaks where the red line exceeds the blue dots around 08Q3 and 09Q4). To me, this suggests investors’ and others non-subscribers' growing interest in Netflix. Second, there is a notable deviation in the correlation between searches and subscribers near the end of the period – the number of searches for Netflix clearly exceeds the number of paying subscribers in Q4 2010, and Q1 and Q2 2011. I attribute this to the stock hitting new highs and the recent controversies, though a more thorough analysis may offer another explanation. Finally, and most importantly, the number of people googling Netflix peaked in the first week of July 2011. There has been a substantial decline since then.

What do the recent declines portend for Netflix's Q4 number of subscribers? The easiest way to get a handle on this is to look at the past and compute a linear regression of the effect of average google search volume in a quarter on the reported number of subscribers in that quarter. To account for the recent controversies and the stock price rise, I include two controls in this regression: a) the average quarterly search volume for Netflix's ticker symbol "NFLX", and b) the average quarterly search volume for the term "Netflix price" which should capture some of the interest in news stories about the price increases. (For readers familiar with Google Trends data, I used fixed scaling).

Table 1 presents the predicted number of subscribers based on these data and the regression. Generally, the regression performs quite well, and better over time. While the predicted number of subscribers were overestimated by over a million in the first several quarters of data used, the difference - shown in column 5 - steadily declines. Indeed, the predicted value for the 2011Q3 is only about 35,000 individuals off. One thing to note, however, is that uncertainty in the prediction - shown as the standard error of the prediction in column 4 - has increased substantially in recent quarters. Figure 4 shows this information graphically.

Table 1. Paying domestic subscribers and regression predicted domestic subscribers based on Google searches of "Netflix", "NFLX", and "Netflix Price."

(1)

(2)

(3)

(4)

(5)

Year and Quarter

Paying domestic subscribers

Prediction based on searches

Standard error of the prediction

Difference

04Q1

1842

3238.8

346

-1396.8

04Q2

2024

3308.6

342

-1284.6

04Q3

2135

3657.4

324

-1522.4

04Q4

2486

4006.3

310

-1520.3

05Q1

2887

4508.7

294

-1621.7

05Q2

3109

4257.6

326

-1148.6

05Q3

3423

4688.1

290

-1265.1

05Q4

4026

5130.9

286

-1104.9

06Q1

4734

5449.5

273

-715.5

06Q2

5017

5307.7

273

-290.7

06Q3

5489

5744.7

263

-255.7

06Q4

7326

6107.5

260

1218.5

07Q1

6676

6364.1

288

311.9

07Q2

6609

5806.7

287

802.3

07Q3

6845

6361.4

254

483.6

07Q4

7326

6546.2

259

779.8

08Q1

8102

6710.1

478

1391.9

08Q2

8235

6689.0

337

1546.0

08Q3

8490

7434.1

327

1055.9

08Q4

9164

8943.6

346

220.4

09Q1

10116

9304.9

248

811.1

09Q2

10375

8984.6

291

1390.4

09Q3

10835

10255.7

500

579.3

09Q4

11892

11138.7

642

753.3

10Q1

13622

12571.8

461

1050.2

10Q2

14577

12985.9

289

1591.1

10Q3

15862

14577.9

407

1284.1

10Q4

17935

16654.3

734

1280.7

11Q1

21405

23537.9

745

-2132.9

11Q2

23263

25589.8

819

-2326.8

11Q3

23789

23753.8

1276

35.2

11Q4

?

20734.5

1138

?

Click to enlarge

Note: all data in thousands.

Figure 4. Predicted domestic paying subscribers vs. actual, over time.

Click to enlarge

Click to enlarge
As can be seen, the predicted 2011Q4 results are rather dismal, and quite similar to what was obtained based from the regression in figure 2. This is another data point and is also worryingly below what appears to be the company's guidance. Of course, this begs the question: is Netflix overpromising again?

Based on these results and a number of other worrying signs coming from Netflix's balance sheet, I am not going to try and catch this falling knife. (Of course, past results are not a guide to future results). To play this, I am holding onto a small number of November 19, 2011 $55.00 Puts which I retained after taking profits on a larger short position with the same puts that I entered last week.

Disclosure: I am short NFLX.

Additional disclosure: I am short Netflix through ownership of puts.