The Cold Start Problem 6/7
Part 5.
22/ Twitch
Growth can’t be infinite, so at some point every company faces (or will face in case of Google or Facebook) an inevitable ceiling being a result of a combination of multiple factors: market saturation, churn from early users, dangerously high levels of spam, fraud, hate speech, insufficient levels of engagement from new users, a degraded user experience due to many users, etc. [MK: I’d add the inability to for new users to stand out, for example, for a new YouTube channel to get to 100m subscribers.]
Product improvements push this ceiling up a notch, but eventually the product hits this ceiling and bounces off it. There’s only as much that can be done. Network effects unravel as fast as they used to push the business in the right direction in the past. [MK: It’s also hard, but not impossible, to “cancel” an entire service or social network.]
When something is not growing – it’s stagnating and slowly dying. Even if the company is profitable and brings in millions of dollars. [MK: at the very least, the exit options for the founders become very limited: selling a company won’t be anything but a “fire sale” with the inevitable associated stigma.]
Consumer products are naturally limited by the size of the world’s population; SaaS tools run out of solvent SMBs to sell to and have to start courting enterprises. Bottom-up growth stops being sustainable at some point in life of a SaaS startup. Products don’t just hit one ceiling – they bump up against plateaus.
Over time the hard side may continue to be hard, and also obtain and learn to use even more power.
Hitting a ceiling often still means that the company is successful, however, more and more efforts are being spent on maintaining an existing size and plugging leaks. Diversification is one solution (not without its risks), but for single-product companies dealing with ceilings is a never-ending battle.
23/ Rocketship Growth
The Rocketship Growth Rate: first get to $1B+ of valuation (which usually implies revenues of $100m+) making the company IPO’able and work backwards to determining the growth rate with which these numbers are achievable (at least, theory) over 7-10 years. Revenue and time are the two major constraints: while investors have their own horizons, so do the key employees.
For SaaS startups the winning trajectory looks similar to the following: a product-market fit —> $2m ARR —> $6m ARR —> 18m ARR —> $36m ARR —> $72m ARR —> $144m ARR. After 10 years a 100% YoY growth rate is quite rare, but 50% is recommended.
For marketplaces the leading indicators are GMV (gross merchandise value) or net revenue, and the valuation gets derived from these numbers. For social products it’s the DAU and net revenue, etc. Once the company agrees on this metric, it works backward to set the interim targets. The growth rate needs to be believable and accepted by the team tasked to achieve it. Growth rates are naturally skewed towards the early years of a company (or a pivot), so it’s more probable to see 200% —> 100% —> 50% YoY growth trajectory rather than 50% —> 100% —> 200% due to the fact that most growth levers are exploited early in the life of a product. Growth rates in the first several years should be in the hundreds of percent for the business to achieve meaningful scale during the lifetimes of its founders and early investors.
There’s a lot of team dynamics going on during the growth period: on the one hand, growth is as motivating as addictive, and employees want the certainty of their future upside in the form of stock options being above the water.
Company ambitions, laughable initially, become serious once they get accompanied by the plans to achieve global domination. Lots of money gets spent on growth, product development and poaching expensive people – and the more painful it gets if these companies hit their ceiling too early.
If the company can’t reignite its growth after hitting the ceiling, there’s a huge damage to the staff morale, especially those with marketable skills who are aware of the fastest-growing companies in the market. The crisis self-perpetuates, makes fundraising much harder and longer, possibly making companies more risk-averse.
Networked products are in a better position than others to counter reaching the plateau as they have more opportunities to amplify viral growth via optimisations and resolving the emerged inefficiencies and imbalances. Conclusion: product-led growth is sustainable for longer than the marketing-led one. And that’s why networked products usually grow larger and stay longer.
24/ Saturation
At some point the potential customers pool depletes, and the focus shifts from adding new customers to offering more products and services or means of interaction to the existing ones to increase revenue. The only way to grow is continuing innovating.
Network saturation occurs when adding a new participant creates smaller value than the previous one. Same applies not just to acquisition but to engagement, too.
Not all engagement was made equal, and there are untapped opportunities for improving this metric in certain segments and/or adjacent networks. Adjacent networks become a growth factor when the existing networks are exhausted. The consist of users who might’ve tried the product but didn’t engage at the right level, thus not becoming a product of habitual use.
Understanding why a certain group of people isn’t using the product (underpowered phones, insufficient amount of content from top creators, etc.) is nothing new, but nonetheless important (if not cliché).
Adjacent segments bring with them a new set of constraints related to the hard side. New users need new content or drivers or accommodation hosts. For many businesses it also makes sense to solve these constraints (the hard side) with international expansion in mind.
Existing customers may benefit from the new formats of interaction to better support the desired use cases. (e.g., Snapchat Stories running in parallel with the core photo messaging app.)
New geographies are an obvious and much-abused growth option. Some products require building new networks in the new regions manually, some products are more suited to be internationalised by programmers and marketers only. Adjacent geographies share users, adding density to the initial network. This works not only for geographies, but industry segments, too.
Obviously, not all geographies share the same characteristics, so copy+paste of a business model may not be so straightforward.
The issue of saturation is not so much in the lack of ideas, but rather in the execution. Starting from scratch is an unsurmountable management challenge, especially when the existing product still grows in revenue or other key metrics. Most new product initiatives fail, so there has to be an established safety net for the company CEO and the key management to take risks in order to revitalise the business.
M&A can be a solution, but again – the seller is statistically more likely to benefit from the acquisition than the buyer.
25/ The Law of Shitty Clickthoughs
Every marketing channel degrades over time. All product metrics go down over time, too. That’s why products hit their ceilings.
Email clickthroughs went down over time, too (there are too many emails that are sent as part of engagement tactics of online services but are rarely worth reading). [MK: On Coursenotes I get 34% open rate at most.] The same applies to banner blindness as well as other forms of advertising (including “native” ads).
Since all interactions between users and software are complex, i.e., require multiple steps, degradation of engagement with at least one step (e.g., collaboratively editing a document with change notifications sent via email that rarely gets read) may become a severe issue, often not dependent on the product at all. Issues like this can cascade and affect both the new and existing users.
This explains the major channel’s marketing payback period that tends to grow over time and with additional budget: 6 months become 12 months, then 24 months, etc., with the product becoming a hostage to the marketing spend.
The recommendation given in the book is very cliché but goes like this: find new marketing channels within the SEO/SEM strategy for B2C and add direct sales to B2B. Multichannel marketing is the essential part of the solution, and its components need to be rebalanced every few years.
“Networked products can be more efficient by growing without spend, by optimising their viral loops.” (Direct quote) Twitch: focus more on creators, give them better tools and watch the creator base grow resulting in higher user engagement. Getting to a billion-user base by buying a significant chunk of it is unmanageably expensive.
26/ When the Network Revolts
The hard side can grow in importance to the company, but their interests may run counter. This is most often about the money (commissions charged to sellers, benefits to car sharing drivers), the new moderation or API usage policies. The members of the hard side can join efforts and try collective bargaining – sometimes leading to the death of a product (Twitter’s Vine).
It makes sense to proactively manage the power members of the hard side, because most likely they’re earning their living from the platform and their commitment is far from simply emotional.
The power members cause the majority of the engagement. And they also deliver the highest value to the members (in terms of quality of service, product, or content) – as long as the platform allows them to improve their emotional and monetary outcomes.
Over time platforms move to professionalisation – both in terms of the existing hard side that evolves in its quality (often – thanks to the education provided by the platform or its partners), and in terms of the new hard side members that are attracted to the platform. (E.g., made-for-TV content on YouTube.)
Professionalisation leads to targeting new customer segments, too.
“The most successful members of the hard side are most likely to professionalise, and they also have the most expertise on how to become successful.” (Direct quote) Thus, all things being equal, these members are the most suitable candidate for the investment by the platform. Or they may become VC-funded startups themselves.
Professionalisation, nonetheless, creates a risk of the company’s increased dependence on its hard side members, which inevitably leads to a confrontation down the line. This is OK.
Market saturation also means that that the new members will be less engaged and relevant than the previously acquired audience. At the same time, not giving the hard side a chance to increase their revenues leads to churn and negative economics.
27/ Eternal September
Every network starts with a “netiquette” – a set of rules of how to behave and what to do and not to do. Starting with a small set of close friends (speaking about social networks) is easy and engaging; one increasingly has to moderate themselves as their network grows and they can get in trouble for posting something that was acceptable for a smaller audience (i.e., a collapse in context).
Context collapse occurs when too many networks (personal, professional, hobbies, etc.) are brought into one. The hard side gets confused as they no longer have a homogenous target audience.
There are networks with fake names and real names (LinkedIn, Facebook) where online words have real-world reputational consequences. Since networks amplify the message (good or bad), if the risk gets too high, creators will reduce their involvement, negatively affecting their users as a result.
New network users expect different value than the early users. New sellers may start selling something that the early sellers would never stock – thus creating an existential conflict leading to the emergence of anti-network effects.
When users can create sub-networks (e.g., small Slack channels), this allows maintaining the context and avoiding (or postponing) the collapse. At the same time, discovery and rediscovery becomes a challenge: having to find that one Slack channel among a hundred of similar ones may hurt user experience.
Spam and scams negatively impact the retention metrics, too. [MK: at FunCorp we relentlessly fight both, as our rule is “No user has ever been better off by having been scammed or spammed”.] Allowing users to report spam goes a long way in fighting it. [MK: it’s sad Facebook’s moderation is very error-prone and leads to lots of false positives.] Being able to flag content, a restaurant or an Uber driver is important to keep the network healthy. This is self-governance, and it’s good.
28/ Overcrowding
Too much content without the proper means of discovery is similar to the problem of a full wardrobe with nothing to wear. YouTube started with manual curation, and this is a de-facto rule for the initial stages of most products.
MK: the book mentions app discovery via app store features; while this is definitely cool, the vast majority of the new users don’t stick and as such look like a deadweight.
For creators the overcrowding problem sounds like “how to stand out?”. And for the new users – “how to get in?”. Some networked products like YouTube favour the creators who have been on the platform for a while. Eugene Wei calls it the “social capital inequality” [MK: love the guy!].
What happens if the most controversial and opinionated users get rewarded by positive feedback loops on the platform? Often these loops need to be broken (with possible political effects like the de-platforming of Donald Trump).
Giving new creators access to the new audience (partially to give everyone some time under the sun, and also for the platform to estimate the quality of an individual content item) can alleviate the “why bother” concern of the hard side. Every piece of content has its viewer, and it’s up to the recommendation algorithm to maximise the number of properly matched views.
Suggesting a connection via the “people you may know” kind of functionality does a good job of increasing the network density. Acting on implicit signals (i.e., the observed behaviour) help avoid the bias of explicit signals (the stated behaviour).
Sometimes the hard side starts being poorly differentiated, so platforms decide to limit the number of the hard side members to a lower number while maintaining the quality of the overall offering.
Part 7.