Takeaway: 4Q results + mgmt commentary ≈ TWTR sunsetting CPC in favor of autoplay; a dangerous move in the context of its MAUs/inventory constraints

KEY POINTS

  1. 4Q16 = PRETTY UGLY: TWTR missed top-line results on waning Ad revenues.  US ad revenues declined on a y/y basis at an accelerated rate in 4Q.  Non O&O Ad revenue remained largely flat y/y, so most of the pressure was concentrated in its core Ad business.  Further, the implied 1Q revenue guide translates to exacerbated y/y declines any way you slice it.  Worse, the MAU metrics disappointed despite what should have been a clear tailwind toward MAU growth given the Trump rubbernecking effect, which is why we stayed on the sidelines this print.  We suspect that MAU weakness wasn’t a function of waning user interest given TWTR’s rising DAU levels, but rather TWTR’s conflicting monetization strategy (ad load).
  2. STORY REVERSION:  We had been wondering how mgmt was going to approach life after comping past the autoplay launch…outside of using headcount cuts as a preemptive excuse for inevitable declines in revenue.  The answer was just more autoplay ad load, which we suspect increased considerably in 4Q16.  Ad engagement spiked by nearly 50% q/q; we haven’t seen that level of sequential growth since the initial autoplay launch in 3Q15.  But this time we also saw the return of TWTR’s diverging MAU/ad engagement trend, suggesting that surging ad load is once again pushing its users away.  So even with the Trump rubbernecking tailwind, US MAU growth decelerated given on what we suspect was heighted ad fatigue.  In short, TWTR is back to where it started in terms of inventory constraints.  But this time, it is also layering on a demand problem following its sales rep cuts, and this is withing selling environment that has become more competitive for branded advertising dollars according to mgmt. 
  3. PLAYING WITH FIRE: Mgmt had suggested that it will be moving away from some of its legacy CPC ad products, which basically means it is going to start relying even more on autoplay now.  The problem with ramping autoplay ad load is that it comes with greater risk than ramping CPC; not just because autoplay is more intrusive, but since the user can literally shut it off in their app setting (see last slide below).   So the autoplay risk isn’t just confined to ad fatigue/MAU churn, but ad engagements automatically reverting back to pre-autoplay levels for any user who chooses to shut it off, which is basically another form of churn in the form of forfeited ad engagements.   For context, we estimate that per-user ad engagement has increased by nearly a factor of 6x following the launch of autoplay.  We suspect that increase was essentially all autoplay given its lower engagement threshold.  In short, the story may be worse than TWTR’s implied revenue guidance would suggest.  We’ll be curious to see how sell-side approaches its estimates from here.  

TWTR | Playing with Fire (4Q16) - TWTR   Ad eng vs. Price q q 4Q16

TWTR | Playing with Fire (4Q16) - TWTR   Comping Autoplay

TWTR | Playing with Fire (4Q16) - TWTR   Ad Fatigue Chart 4Q16

TWTR | Playing with Fire (4Q16) - TWTR   Ad Fatigue Slide

TWTR | Playing with Fire (4Q16) - TWTR   Autoplay Slide 1

TWTR | Playing with Fire (4Q16) - TWTR   Autoplay Slide