#5 - How serious is the GPT threat to Google?
Conclusions
The main question is the efficient frontier of trust/ROI
Google has potentially $220B revenue at risk
The release of ChatGPT has threatened Google's revenue model…
Microsoft allegedly will be including the technology in their search…
As covered in #4 AGI Impact on Google, Microsoft, Salesforce, Amazon, ~80% of Alphabet’s revenue comes from advertising.
Press coverage that the CEO is overseeing this initiative….
CEO Sundar Pichai has been in meetings to "define Google's AI strategy" and has "upended the work of numerous groups inside the company to respond to the threat that ChatGPT poses." - CNET
…Alphabet is stating it is delaying release out of safety concerns. While there are valid concerns, I speculate that safety is not the cause of the delay.
Alphabet is a company that has delivered information results to the entire world for over 20 years. The company must have one of the best taxonomies of “safety” ever created. Feeding GPT results through this safety engine is not a CEO level oversight problem.1 There is a more serious threat...
Is this a serious threat to Alphabet?2
I suspect the issue is the advertising model. Let’s breakdown the problem….
see appendix below re: Google revenue equation
I view this as an efficient frontier between user trust and ad ROI . Truthful answers create user trust. It is an efficient frontier between search truth and ad ROI. Sacrifice either and over the long term the model doesn’t work.
(# below tie to Problem Structure Diagram above)
1.11 Will the model suggest the search user take an action different than the ads are suggesting? In the past search results may have done so but it took a lot of clicking and searching to find the truth (aka research). New world example: You ask the “search engine” what the best electric car is to buy. It says GM has some good looking cars but the verdict is out on how long their batteries will last relative to a Tesla. At the same time the search serves an ad for a GM truck. You can’t imagine a happy GM Chief Marketing Officer in that situation.
Some would argue this is nothing new. I would disagree. In the past Google could claim to the unhappy CMO the algorithm returns the best links. In this new case Google is creating the actual answer. Google now has to debate whether GM batteries will last as long. That doesn't scale. Does the CMO decide to move ad $ away? Yes if ROI isn't there.
There are some solutions. Regardless, it is important to remember Google search owns ~84% of search dollars3. Its valuation requires this. Any of these changes are likely to chip away at that efficient frontier and valuation. It could be death by a thousand exceptions in moves away from the current efficient frontier. To be clear, I am not saying it can’t be done, I am saying it could be less lucrative than the current $1.1 trillion valuation warrants.
1.2 There is an upside case of the value for advertisers. With that comes a potential retargeting issue. For example, imagine asking the search chat a question: Which electric car to buy if you want a set range and reliability?
Say you end up buying the car. Is the search engine claiming credit for returning that recommendation? If so, there needs to be a database tracking every product returned in each search query. How does it show to the CMO it recommended that car X amount of times? This is not impossible but does add different complexity. In this world is this paid on an impression basis or click? This is what I suspect the CEO team is determining4.
2 I see this as minor. Yes, if a search player does not offer this technology their volume is likely to drop. That said, I suspect all will do so.
Conclusion:
The threat to the ad model’s efficient frontier (trust vs. ROI) is the real threat here.
Appendix
I think this diagram does a decent fast way of depicting Google’s revenue equation
Safety is important. I am stating from a governance standpoint of a $1.1 Trillion company this is not a complex issue. Safety is a derivative of an issue Alphabet has dealt with its entire life. Generally, I would suspect the company could throw 3 VPs from different departments to oversee and form a recommendation the CEO would decide. As I elaborate, I suspect this is a complex threat to the revenue model that the CEO -rightfully- needs to weigh in on ~20 smaller decisions that impact a larger decision.
I do not consider the barrier for Alphabet to be technology. See #3 Input article.
All of this does not have to be determined immediately. That said, competitive strategy is like chess or war. If you start taking too many actions in one direction and you realize too late it can be hard to undue those moves.