Fear in the Evening of the Platform Lifecycle
The news and articles about big tech platforms got me thinking about where things stand in the product lifecycle and what might be next. Three areas specifically: layoffs of ˜150,000, increasingly vocal frustration with the user experience, and all the buzz on the subsidization model and alternative platforms. I am focused mainly on Google, Facebook, Twitter, and Amazon, and by ‘platform’ I mean their user offerings across search, shopping, social media, blogging, and video/audio streaming. First I needed to model the present state, and then get the future sorted.
How Events Identify Lifecycle Stage and Strategy
The way these layoffs were executed, particularly at Google, provides insight into their business and product cycles. Some workers found their access shut down or their key card flashing red at security and current and fired workers said the selection of who would go seemed random (that’s only anecdotal, but it wasn’t on merit). The exception was the targeted cuts in the 'Area120' group (top of the product funnel, speculative work to turn ideas into future products and revenue streams).
It seems likely that a cost reduction target was set and that determined how many employees would need to be fired. Think about how that contrasts with a company that maintains efficiency by removing underperforming employees and chopping projects that are unlikely to pay off. The way this was done must put fear in remaining employees that accountants are running the company and doing your job well does not reduce the chance of the axe making a visit to your desk.
One Google employee speculated that some long-time employees may have been fired because good reviews over a long period had pushed their salaries out of line. "Congratulations for doing so well here, but you made yourself a target, here's your pink slip!" The culture established by founders Larry Page and Sergey Brinn accelerated innovation by treating workers as the company’s most important stakeholders. These cuts sent a clear message that this is an era with different stakeholder priorities.
Business management legend Peter Drucker said there are only three fundamental business strategies: Operational Efficiency, Product Innovation, and Customer Intimacy. I’ve used this to map a company’s actions to the three strategies to identify their relative mix and thus their focus and vulnerability. The way the layoffs were carried out showed a focus on Operational Efficiency without concern about a chilling effect on Product Innovation. In the lifecycle of a technology product (see 2nd image below), that is another indicator of a late-maturity product stage.
The business model for growth and domination of an online category was proven and accepted years ago. Investment in innovative capabilities and subsidization of content creates a unique and valuable user experience that sparks viral growth. A critical mass of users creates a draw for other stakeholders like advertisers and content providers. The beauty is that in either direction growth of one stakeholder base tends to drive growth in another.
For example, a critical mass of content providers can draw in users who can get all their content in one place. In causal models that is a reinforcing (or ‘vicious’) cycle, and a very good thing when growing stakeholder bases!
The model subsidizes growth in the early stages. In later stages, the leverage gained from that growth is used to generate large returns (or think of it as paying back the loan used for subsidies). What interests me as a product manager is that we are moving into the evening in the lifecycle for these platforms, and given that this is a model without historical precedent, I don’t think anyone has seen this before and what dynamics will be encountered. The user experience, and thus user relationship, is subsidized on the front end, which means now these companies have to increasingly mess with users to generate revenue, but also keep them from churning.
Increasing returns to investors as the lifecycle progresses logically implies that relationships with users will decline from here on.
All platforms or products age and as they do they face rising competitive pressure. Alternatives within platform categories will step-up their offerings, like Tik-Tok or alternatives that are attracting users with a different model and/or policies, like Mastodon. Then there is the wildcard: new innovation, most likely via AI to bring substitute products that can replace existing capabilities (see Harvard’s Michael Porter- Five Forces Model for where to look for future competitive threats).
Right now, any alternative needs to have compellingly superior capabilities to overcome switching costs for the user, and no alternative has reached that level of innovation. But the clock is ticking, These platforms will need something exciting next and they’ll need users to buy into it. The model predicts that at that time the relationship with users will likely be at a historic low. That will be challenging.
Whether it is search or social media or streaming, it helps to think of these platforms as markets in order to understand how the dynamics between the provider and a stakeholder group shift over the lifecycle. Let's build a conceptual model of these platforms using the analogy of a stock exchange.
There is a growing terminology around feeds, connections, and associated metrics. For example, it is common on Twitter to refer to the sum of your posts on which another user took action as an 'engagement' (a like or retweet). Just FYI, I'm not trying to comply with terms used elsewhere!
· ‘Market’ mirrors the economic reality that these platforms negotiate connections between stakeholders in real time. Non-monetary connections like presenting an unpromoted link from a search or showing a post from a connection compete with monetized connections that the platform handles.
· Since tech loves acronyms, let’s call it PaaM (Platform as a Market)😂.
A market may have one to many stakeholder groups that participate
· Stakeholder groups include users, advertisers, retailers, resellers, content providers, and creators.
· It may not feel warm and fuzzy, but stakeholders are market commodities and a connection is the atomic transaction of the platform.
· Examples of a connection are being shown a friend’s post or being served an ad. This is the ‘atomic transaction’ of a platform- its building blocks
· A single user action may generate several connections
· The platform acts like a market maker in a stock, mediating the connections that are made
Market Maker represents the platform provider’s algorithms or other methods to respond to user actions by serving up connections — this is the black box that is often the target of public criticism if the user experience declines (Market Maker exists in my virtual model only- it represents the proprietary approach that each platform uses to do this function).
By balancing flows in the model, we can assume a subsidy flow TO users in a growth phase will require a revenue-supporting flow FROM users in a later harvest phase.
· moving from growth to maturity, market maker de-prioritizes the user’s preferred connections in favor of prioritizing revenue-producing connections (i.e. serving more ads or other paid placements).
· The market-making function likely applies Algorithms that use data the provider has on each stakeholder. For example, a creator like a streamer may have tier and target audience attributes that determine how the streamer’s connections (in this case, video content) get distributed to users.
· another consideration for the market maker are bids from commercial stakeholders (ex. advertisers) to increase the chance of selection or to improve ranking.
Platform Model and Technology Product Lifecycle
The conflicts with stakeholder groups in this business model don't appear until the lifecycle progresses to maturity. First, let me explain the lifecycle concept and then apply it to these platforms.
Since the 1990’s the most popular model for technology product cycles is the one defined by Geoffrey Moore in Crossing the Chasm and Inside the Tornado. When he first published, enterprise software was primarily sold as a one-time right-to-use license and installed in a company’s data center, so things are a bit different now😆. What I still find useful is Moore's identification of shifting dynamics between the company and its users, suppliers, and partners over a product lifecycle. This shift has implications for profitability and operational emphasis which we have seen from the recent news that led me to write this article.
What’s important with Moore’s model are the changing dynamics with customers, competitors, and partners as the lifecycle progresses.
The subsidization that is used to grow the base of stakeholders requires access to capital, but eventually, that bill needs to be paid. That can be done by feeding more ads to users, auctioning content to users, charging premiums for placement, or even replacing external content or products with in-house substitutes. The implication is that user experience will be best during the growth stage and will increasingly decline after that. The primary force maintaining user experience in maturity is the threat to the provider of mass churn.
This model has put innovative software in our laps for free, but there’s no free lunch.🍕
The declining user experience is generally driven by the platform’s need to abide by our preferences less in order to generate revenue. What users can do is close the information gap to better understand the financial pressures on a platform. The more financial pressure on a company, the more likely user experience will decline.
For example, Elon Musk took on a massive amount of debt to purchase Twitter. Consider his recent $300 million interest-only payment on that debt. For two years I’ve had a Twitter Developer’s license that provides me access to the full backend API and I’ve used it to build NLP analyses that I’ve written about here on Medium. Musk just announced that API access will now only be available via paid license. This is due to the pressure to find any extra revenue to service the debt. The trouble is many value-added services or bots use the free API's, as do researchers who provide analysis of social media or contribute open-source NLP tools. Musk will generate some incremental revenue by charging for API access, but everything he is cutting out may decrease the value of the platform to users, resulting in churn. It is not clear to me that the net of his action with API's will be a positive for Twitter.
At the End of the Lifecycle, What Comes Next?
There are a range of forces that could change the platform landscape, from AI to regulation to advocacy among stakeholder groups (such as suppliers or warehouse workers with Amazon). Instead of those drivers of change, I am focused on current alternative platforms as well as possible new components or offerings that might shift balance in the platform ecosystem.
The current structure and model allows a handful of individuals to constrain innovation like turning off a water tap. Can alternative solutions change the dynamic between platforms and users and how? Would protection of user data ensure low switching costs, thus sparking new development of alternatives?
Through Medium’s new partnership, I was introduced to Mastodon. I looked deeper into its model and it is taking a different approach in almost every aspect of its business. It uses a different organizational structure (non-profit), funding (crowdsourced), decentralization (fediverse compliant), as well as granular user control over feed and distribution. Mastodon isn’t positioned to be a mass-market leader, it’s more niche, but the experimentation with changing up the model and creating a different dynamic with users is something we need more of. When considering alternatives, we can’t simply be lured by shiny objects, but need to look deeper at what a company is doing differently.
Tik-Tok is a good example of a shiny object that may not be any different. At first, it knew a little too much about what the user wanted (hmm, hope this isn’t Huawei) 😆. Now, I’ve seen the familiar increase in complaints about user feeds returning too much junk. Sustainably better user experiences won’t happen if the business is using essentially the same subsidization model.
The EU took a leading position in the protection of user rights to data as well as privacy and helped raise user awareness on data rights. One of the major switching costs in moving to an alternative is losing your data and the connections and content rights it may represent. It’s clear to me that decoupling user data from proprietary platforms is the most promising initiative that could shift the balance in the online platform ecosystem.
This article isn’t just one big pitch for my personal data ideas, but it happens to be an area that I've spent some time on. How would a company introduce a user data component that protects users' rights? The tech giants would likely see it as a threat and squash it, or simply refuse to support it. That is why I brainstormed features that could be included that would incentivize big platforms to be compliant with it.
Rethinking the Online Platform Architecture
I wrote an article about User-Driven Identity and Personalization and I built a light proof-of-concept in Python. To ensure arms-length, third-party operation, the system would need to be administered like phone number clearinghouse systems, do-not-call lists, or domain name registries for the web (ICANN specifications). The concept is the system could be operated by one or more companies but with compliance to published specs and a service-level agreement (SLA). This would be a first step to allow user preferences, connections, and links for content rights (like NFTs) to be managed separately from search and social media platforms.
To make the existence of this platform a benefit and not a threat to large, existing platforms, I’ve identified extended personal attributes like affinities, traits, and learning styles that platforms can use to drive far more productive and personalized interfaces in the next generation of online platforms.
Users provide authorization before the systems they identify can check out their personal data. Switching platforms is simple, like porting a phone number but much faster. The user could simply drop the authorization for their old provider and add authorization for their new provider.
The way private data collection works today, most users would never volunteer extended information and would resent it if scraped without their knowledge. Once we have a system where users control who gets to use their data, we can start to drive a far more personalized experience, and the platform providers will be the ones who benefit from that, along with users of course.
In 2020 when we suddenly had to do everything over the digital channel, I began integrating what I know of neuroscience with what I know of UI/UX. I thought about tools that a user could use to capture extended attributes that could drive more personal and productive customer support systems in particular, but it applies to any online app. Users need to trust the independence and security of the system that holds this personal data before we could experiment with some really cool, personalized UIs.
An independently-operated personal data repository with support for personalized, productive next-generation UI's could protect users' data rights and keep switching costs low while also enabling deep personalization that would be a benefit even to large, existing platforms.
I remember using Google Search and Gmail when it first came out and I was also early on Twitter. This model has put a lot of innovative products in the hands of the public for free, and these platforms have transformed the world. But we can't escape the truth that user experience only declines over the life of the product. That puts all the platform providers under increasing pressure to retain users. Google basically makes money from search and Cloud. I get the sense that they are going to need some products that users just can't live without, or alternatives may start to truly threaten their base.
