Self-driving cars as the new toolbar

There is a pretty simple and economically motivated reason why the likes of Google and Apple would want to power autonomous vehicles: More free time equals more time on their services ergo more revenue.

Google’s “distribution” business (of which I was a part between 2006–2009) had a clear mandate — increase the default access points to Google search. Initially this was done via deals to pre-bundle Google Toolbar into PCs and browsers, or to make Google Search the default on a browser, as was the case with Firefox, or portals. This evolved into Google search on mobile operator portals and mobile search pre-loaded on mobile handsets which then evolved into Google’s own browser (Chrome), and OS platforms (ChromeOS and Android).

The economics for these deals were straightforward: A revenue share on ads clicked from searches through those access points or an up-front “bounty” for pre-loading the software. Our friendly finance team would input some numbers and out would come the parameters we could negotiate on.

The model included assumptions on:
- Default placement and implied LTV of that placement (very few people change their default search engine or browser) which led to
-Average number of searches per user per country
-Multiplied by average number of ads shown per search
-Times CTR on ads in that country
-Times average $ value per ad in that country
-Which generated Google’s likely take from that placement
-Minus a cannibalisation % on likelihood that such a placement would cannibalise searches directly on google.com
-Which led to the total revenue that could be split from such distribution deal. How much each partner could get then depended on their scale (breadth of distribution) or importance of that geo (for example increasing market share in a country like Russia or China where Google was not as dominant)

This was a great growth business for Google and helps explain why Google was reportedly willing to pay Apple $1 billion for default placement on the iPhone.

It also explains why Android was so strategic for Google as mobile accelerated. A Google-controlled OS with a full web experience (rather than the compressed web Nokia or RIM were pushing) was known to generate more searches and, once mobile ads began to mature, more direct revenue for Google to share with operator and OEM partners *IF* they conformed to what was deemed a pure Google experience on the device (“GMS”).

While Google does not carve out revenues via Android, in their lawsuit against Google, Oracle claimed that Google had made $31 billion in revenue and $22 billion in profits from Android.

Which takes us to cars. When noise about Google’s self-driving car project (which was launched in 2009) began to surface my initial reaction was “of course! The new toolbar/Android.” Yes there is likely to have a massive social impact from autonomous vehicles but the cost to power such a service needs to be absorbed by either ownership…or in the case of Google a massive revenue machine that could benefit from it.

Data from the US census bureau estimates that the average US commute is over 25 minutes while New Yorkers spend an average of 35 minutes in gridlock.

Via Washington Post https://www.washingtonpost.com/news/wonk/wp/2016/02/25/how-much-of-your-life-youre-wasting-on-your-commute/?utm_term=.52d17319b984

To work and back that means almost an hour behind the wheel. That is an hour where you are forced to focus on the road and not your phone (hopefully!), and listen to audio media (powered by radio advertising). Interestingly in 2006 Google had purchased DMarc with a view to bring the efficiency and targeting of AdWords to audio ads but pulled the plug on the efforts in 2009. Pandora has a nice audio ad business having generated $1 billion from audio ads in 2016. But $1 billion is nice, but not the type of “moonshot” business that Google should be going after.

Taking the long view, an hour a day assuming 5 day workweek and 260 workdays per-year gives us 1,300 hours per year per worker. IF (and it’s still a big if) autonomous driving allowed these drivers to focus their attention away from the road what would they likely do? More specifically what could they do if said vehicle had a full-fledged internet connection, a massive screen and access to the full array of services and sites you usually accessed from your phone or laptop?

The likely answer is that you would use some of that time to browse the web, do some searches, watch some videos, catch up with posts on Facebook or Instagram, send some messages. And, as with the rest of the internet, such behaviour would most likely benefit the maker of the car (or the software powering the car) that you are being comfortably driven in).

As shown below, radio ad revenues is down and digital is now surpassing TV (with Google and Facebook accounting for the majority of that digital revenue).

From The Financial Times https://www.ft.com/content/6c6b74a4-3920-11e6-9a05-82a9b15a8ee7

I’ve never seen it, but I have to assume that a new version of that Excel sheet we used to use for toolbar monetisation exists somewhere inside the Googleplex applied to cars and it reads something like this:
-Car utilization (which under a shared resource model is likely higher than the inefficient use of a car only 5% of the time. Lets say 25% utilization)
-Times average length of trip (lets say 25 minutes as per average commute)
-% of time spent on a Google-owned service (40%? 75%?)
-% of that time that is monetizable (via a search ad, YouTube ad or app micro-payments. Lets say 25%)
-Average revenue per interaction (This one probably has massive sensitivities but lets say $1)

(25 minutes * 40% on google service * 25% shown an ad * $1 per interaction) = $2.5 per “driving session”
If the car is “utilized” 25% of the day that’s 6 hours or $15 per car per day or $56k/year assuming 260 days of use per year.

Given that the average car in the US sold for $33k (and not taking account the non-recoverable engineering costs plus lobbying costs to get the autonomous car on the road) the self-driving car, with Google Auto Services “GAS” — get it?), has, in itself a nice ROI across a lifetime of say 5 years ($280k).

There are 250 million cars on the road in the US so lets assume Google’s cars (and software) only manage to reach 1% penetration across a replacement cycle… that’s 2.5 million cars.

That’s 2,500,000 cars times $56,000 in revenue per car per year or $140,000,000,000 in revenue from the Google fleet per year in the US alone.

That’s more like a moonshot number…. A lot of the assumptions below are likely incorrect and the numbers excessively high (or low) but my main point is around the potential scale of the Google’s revenue from self-driving cars and the belief that autonomous vehicles could well be the next toolbar.

On top of this comes the societal impact of self-driving vehicles, the unlocking of cash from under-used assets and the retooling or infrastructure and urban environments that come as side-benefits of autonomous vehicles.