Metcalf, Google Actions, and Why Marketers Hesitate on Voice

Wed, Mar 11, 2020

Read in 8 minutes

From Bret Kinsella in his recently published ThinkTank article. For more of Bret's great work, head over to

Metcalf, Google Actions, and Why Marketers Hesitate on Voice

“Both Amazon and Google are creating ecosystems around their voice assistants and that has included deliberate outreach to third parties. Whether you embrace the voice web or voice app store model as the better analogy for what is happening with voice assistants, you still arrive at a conclusion that the ecosystems need content and services (for those of you who think it’s just a 1P product, we will address that in a bit). And, given the visions expressed by Amazon and Google around Alexa and Assistant, they will not do all of this on their own. This is where third parties come into the equation.”

Metcalf’s Law, Networks and Google Actions Mysterious Disappearance

“Metcalf’s Law says the value of a network is square of the number of clients in the network, i.e. (n2) where n is clients or in this case users. The core principle has been applied to everything from Ethernet (its original application) to social networks, fax machines, and email. If you are not familiar with Metcalf’s Law here is a simple way to understand it. Suppose you have email but no one else does. That means n = 1. The value of the email network is 1. If a second person acquires email, then the value is (2)2 = 4. There is not big value there because you may not have to email that person very often but there is value. “

“And, look at how this grows. Go to five people and the value is 25, ten people and it is 100. You now can send email to many more people and they can communicate with you. When its 100 people, the value climbs to 10,000. Don’t worry about the value denomination. Just focus on the fact that more participants offer an increasing amount of value that is not linear but the steeper slope of a square of the user base. “

“Voice assistant ecosystems from Amazon and Google are what we typically call two-sided networks. The consumer side doesn’t benefit directly from other consumers joining, but does when enterprises (let’s call them brands) join that have useful content or services. Similarly, brands benefit from more consumers joining the network that they can connect with as opposed to other brands. To be sure, consumers benefit indirectly from other consumers joining because that makes the entire network more valuable to brands which will cause more of them to join. But, a simple equation could be VC = B2 where VC is value to the consumer and B is the number of brands. Similarly, you then have VB = C2 where VB is the value to the brand and C is the number of consumers. “

“To make it more complicated, we have the first-party assistant capabilities which add value but that is linear. This is just a product feature equation. When the voice assistant adds a capability such as weather, then everyone in the network on the consumer side can now ask for the weather. Going back to the math, the value is “n” and not “n2” of each new first-party voice assistant feature. Sometimes it also displaces a brand skill which could create disincentives for supporting the ecosystem. So, for this analysis, we will leave it out. “

“Right now there is an imbalance in the number of consumers using voice assistants and number of brands supporting them. Amazon and Google are actively adding new features, but they will get even more benefit from hundreds or thousands of brands creating content and services for their tens of millions of global users. The scope of what Amazon and Google (and indeed Samsung and probably Apple but who knows what they are thinking) want to provide to consumers is so vast that they cannot build it all themselves nor do they control all of the means of fulfillment. Thus, the third-party ecosystem building activities are central to the overall assistant strategy. “

“This is where the mysteriously disappearing Google Actions incident can really cause damage to the desired network effects. Brands that have not yet built an Action may choose to delay supporting the platform until they have a better understanding about how reliable it will be as a voice assistant network. Brands that already support the network with Google Actions may question the efficacy of their investment and use that rationale to reduce forward-looking investment. Consumers may become frustrated when their favorite Actions are suddenly unavailable and break the habit formation process. “

“There are not many brands supporting Google Assistant yet. Research by Voicebot of 200 leading consumer brands in the U.S. in 20 different product categories found that only 21.5% were supporting Google Assistant with their own Action. Google doesn’t want to provide these brands or other service providers with an excuse to delay their support of the Assistant ecosystem. However, this week they have done just that not only because of the incident but also because of the response. “

Weeks to Resolve an Issue

“Google seems to not have noticed these issues for some time. Even a full day after the first incident was detected, internal Google dashboards apparently showed no clear issues in their monitoring software and there was no concrete communication back to skill developers that had submitted support tickets the day before. Notably, none of Google’s 0P capabilities (this would be known as 1st party skills in the Alexa ecosystem) were impacted so third parties immediately assumed the slow response was a lack of commitment to the ecosystem. “

“This slow recognition of the issue was followed by an email to some support tickets suggesting it may take weeks to resolve. Would Google be so cavalier if search through Google Assistant suddenly went missing? Most third-party developers think the prioritization would change instantly. “

Not All Content and Services are Created Equal

“The bigger problem is that not all content and services are equally valued by users. Starbucks is more valued than the latest trivia game. Similarly, the presence of other brands and services makes it more likely that a request to Google Assistant will be fulfilled. Many of those most valued brands make deliberate decisions on what channels to support. Google may be introducing risk that these brands will focus their attention elsewhere. It is now 2020 budget season. Imagine how this could impact not just a brand executive, but an agency head that is counting on budget approval for a big brand voice project for 2020. Ooops.”

Why it May All Blow Over, or Not

“This may all blow over because from what we can tell, consumers aren’t using Google Assistant third-party Actions very much. Maybe they won’t even notice. The problem is that the brands and agencies are noticing and they will make decisions based on how they expect these sectors to evolve. It may be they simply become more circumspect about Google Assistant or it could extend into skepticism about the entire voice category.

“The agencies in particular are in a tough spot. Most of them have some responsibility for maintaining Google Actions for their clients once in production. When a Google Action fails and cannot serve consumers, they have to explain what is going on to their clients. This takes time but also can undermine the confidence these customers have in their partners and the technology which again puts budget commitments at risk. “

Single Point of Failure Risk and Channel Priorities

“The Great Google Action Disappearance also reinforces the fact that Google has single points of failure in its systems. Voice Insider #30 chronicled how the 5-hour outage of the single global instance of Dialogflow caused issues worldwide. Are we today facing a similar problem or is it just a bug that wasn’t caught in regression testing? Neither is a good outcome and both bring baggage with them.”

“A slow-motion reaction could be the result of Google assuming that these are not mission critical apps and that consumers have other channels to access the content or services. That may well be true. However, it undermines the intent of assistants taking on more responsibility and makes them look like a cut supplement to mobile and web as opposed to a potential replacement. It also undermines the argument to monetize through voice assistants if you don’t know much about uptime or resilience. “

“If there were brands making large sums of revenue every day on Assistant, the rancor would surely be higher. But, most brands are not generating revenue off of Google Actions and if they are viewed as unreliable, they are not likely to make it a core strategy. So, if you treat third party outages as not critical issues because no one is losing money, everyone will recognize the risk and not invest to make money or integrate closely with their products and services. And, of course, you undermine the ability of all ecosystem partners to benefit from the fruits of Metcalf’s Law.”

“You can read more about the outage here. You will note that many large brands are affected as well as a number of independent developers. Hopefully, this will be resolved soon, but at least it provided an opportunity to discuss some bigger picture issues that are important to the future of the industry.”


Imagine, for a moment, the impact of standards that enable DNS-like accessibility and platform interoperability on the above scenario of doubt and hesitation. Imagine building once and using many. Imagine a developer community unleashed through “open” voice.

Imagine the impact of the Open Voice Network. #OpenVoiceNetwork