The Cold Start Problem 2/7
Andrew Chen
4/ Tiny Speck
Slack, Dropbox – these B2B tools get adopted from the bottom up, from micro-groups up to the entire organisations. Every networked product starts with a single network. Reaching the “atomic network” – the smallest network with enough people to receive enough value out of it to tolerate using it - is the goal.
Networks usually have “sides” – content creators and consumers, drivers and passengers, buyers and sellers; one side is usually easier to attract than the other, but it’s the harder side that needs the most attention and care. It’s known that under 10% of users (usually closer to 1%) generate most of the content. This is the audience the product development efforts should be focused on pleasing.
5/ Anti-Network Effects
Anti-network effects are the negative force driving new networks’ population to zero. At inception they’re like the headwinds – they can blow you away if you don’t acquire enough mass (i.e., active users) quickly in the process. Yes, it’s that old “chicken and egg” problem. Or the “Cold Start Problem”.
In practice the initial user base has a spike of newly registered users, and then … nothing happens, then comes another spike (thanks to increased marketing efforts), but unless the network is big and useful enough to reach the minimum viable (what was the term again? Atomic?) scale, many dollars will be spent in a futile attempt to take off.
The odds are not in the favour of the founders (to state the obvious): out of thousands of products only a handful ever reach 1 billion users (i.e., a super-meaningful scale). Those who make it there, though, milk their position at the pedestal really well.
Some engagement metrics are good predictors of retention (or lack of churn). E.g., for Slack the magic number is 2000 messages sent and received, after which 93% of users stay.
The number of users the network needs to be useful is obviously network-specific, but also very engagement-dependent. Engagement metrics optimising the key aspects of user experience (e.g., passenger’s wait time to get an Uber) show both the uplift in the value created (quick arrival better experience for the passenger and more paid trips for the driver) and diminishing returns once the key aspect metric is hit. The threshold for value creation is a very important number that the product company must always meet or exceed.
For two-sided marketplaces the threshold is higher (meaning longer time to launch and higher initial investment into the hard part of the network). For instance, Airbnb needs 300 property listings and 100 reviews to see growth take off in a market. Higher investment requirements also serve as competitive moats, so it’s not all that bad in the light of the fact that copying successful winning moves usually leads to a more muted outcome at best.
Solving the Cold Start Problem requires launching a minimum viable network [MK: sorry, the author never said anything about “minimum” or “viable”, it’s just natural to add these terms here] and ensuring it’s dense enough (i.e., users are active, relevant, and reasonably engaged). Adding the features improving user experience on the core activities will increase engagement, retention and monetisation.
6/ The Atomic Network
Many atomic networks start with a limited geography (Uber) or a user segment (Slack, Facebook); the assumption is that if one is able to build one self-sustainable network, scaling up to 2, 10, 100, etc. won’t be impossible. Scaling up will be equal to copy+paste repeatedly. Each company writes its own playbook for scalable growth.
The threshold defines the sufficient audience that can sustain engagement to consistently deliver value to its members. Starting off takes a bare minimum of relevant features sufficient to create value, but not overwhelm with them. Density is more important than the potential market size (which is irrelevant if there are no customers for the core feature set).
A big mistake is thinking about scaling before the atomic network has even emerged. [MK: time and momentum are more important than money, so getting to this minimum viable size is infinitely more important than monetising the audience from the beginning.]
Growth hacking has seemingly fallen out of favour recently, but still there are certain incentives for joining a network – status (invitation-only), money (promotional credit), or bragging rights (early access).
The next big product will start out like it’s for a niche network. (Direct quote) The people who don’t “get” it are simply not in the target audience (yet). The product itself doesn’t need to change much or at all for the perception to shift: usually it’s the expansion of the atomic network that makes it increasingly useful even to initial non-believers. Mechanisms for inviting others (friends, colleagues) is what makes the network growing by introducing more members to benefit from the value created. The easier it is to grow the network for one player – the easier it is to do so for another player, too, as the same growth dynamics are at play.
Products requiring top-down implementation (make a company-wide deployment first and then teach people how to use them) are poor candidates for networked growth.
7/ The Hard Side
A minority of users [MK: that’s the 1% or even 0.1% I mentioned earlier] create disproportionate value and have disproportionate power as a result. For social networks these are the top creators the networks absolutely want to attract and retain. [MK: unless their political views go against the grain, but let’s ignore this for a while.] They can be app developers, super-spreaders (when it comes to early adoption), marketplace sellers, etc. [MK: the assumption is that there are more buyers than sellers, consumers than creators.] The counter example is that on job sites there are more sellers (candidates) than buyers (jobs), but it’s a rare example.
One-sided networks are the ones where all users are on one side (e.g., messengers), but even there one can find active users increasing the engagement of more than just themselves.
Cross-side network effects arise when more users on one side benefits the other side of the network (more customers drivers are happier, but also the prices are lower).
[MK: User-generated content is a tricky beast: top contributors/creators are usually doing their best not for the money, but for … it’s actually a tricky and very intimate question, but my guess would be that it’s the feeling of self-fulfilment and of an elevated status in the community that they care about is the underlying driver.]
The hard side is something that creates value for others to consume, and in return the “creators” of this value need the status (see my comment directly above) and the money. It’s important to understand that the 1% are putting way-way more time into their trade than their consumers. [MK: it takes me 5x time longer to write notes on the books than actually reading them.]
Every single platform must have a strategy of working with the top 1% of users from day one. The book asks all the right question: who are they, how can they be retained, how can they become even more active and/or frequent on the platform? [MK: I guess a very important question is being missed here: what is the pathway for a creator to enter this 1% of the audience? Not having such a pathway or having an obscure one (thank you, TikTok) reduces engagement and increases the churn to competing platforms.]
Advice on such topics is truly dependent on the product category; there’s no one size fits all solution, even though good product management skills are indeed transferable between industries.
If you’ve been wondering where the 1% / 10% / 100% rule came from – here’s the source. There’s also a “power law” distribution where the top 20% of top influences / creators get the vast majority of engagement.
Snapchat: the pyramid of content grows from “self-expression” (publish everything; Snapchat) to “status” (broadcast only the things that make one look cool; Instagram) to “talent” (putting an oversized effort into making something cool and well-rehearsed; TikTok). Needless to say, the pyramid represents the number of people participating as well as the frequency of their engagement with the world.
Social feedback loop in the form of likes, shares and comments is a very strong motivator for the creators to keep going and create even more content. “If this piece of content was created and no one saw it, would the creator be disappointed?” (direct quote) If the answer is “yes”, the social feedback loop is a key value to the creator.
A social network can’t exist without its core content creators; marketplaces can’t exist without sellers; messaging apps get uninstalled if people transition to other apps (Whatsapp —> Telegram, of course).
Part 3.