Network Rehydration Test
Judge a network’s defensibility by how quickly users can reconstruct it elsewhere
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 97%
Costolo contrasts contact-based networks with interest and professional graphs. A messaging service can rebuild a user’s network quickly by importing phone contacts, so users can reproduce nearly identical value on a competitor within minutes. Twitter’s graph is harder to recreate because it includes distinctive accounts selected around interests rather than people already stored in a phone or email account. That makes Twitter resilient to would-be replacements, but it also makes onboarding difficult because there is no obvious starter kit for each user’s interests. The Network Rehydration Test asks how easily users can reconstruct the same nodes and relationships elsewhere. The slower and more judgment-dependent that reconstruction is, the stronger the network’s defensive moat—and often the harder its initial growth problem.
Origin
Dick Costolo used Twitter, LinkedIn, WhatsApp, Telegram, and Facebook to explain the tradeoff between graph resilience and easy onboarding.
Core principles
- 01Network size alone does not determine defensibility
- 02Easily imported relationships create vulnerable networks
- 03Unique connections take longer to reconstruct
- 04Acquisition friction can strengthen retention while limiting growth
- 05Evaluate resilience and onboarding together
How to run it
- 1
Define Nodes and Edges
State precisely who or what participates in the network and what relationship connects each pair.
Pro tip Distinguish contacts, friends, interests, professional ties, and transactional histories.
Watch out Calling every relationship a social graph hides important differences.
- 2
Locate External Copies
Determine whether equivalent relationships already exist in phone contacts, email, another platform, or a public directory.
Pro tip Include both authorized imports and technically feasible scraping.
Watch out A large network is fragile if its structure is already copied elsewhere.
- 3
Estimate Rehydration Time
Calculate how long a typical user would need to rebuild enough relationships to recover the product’s core value on a competitor.
Pro tip Measure time to useful equivalence, not time to create an account.
Watch out Do not assume users must reproduce every edge before switching.
- 4
Measure Judgment per Edge
Assess how much discovery, history, or personal judgment was required to create each valuable connection.
Pro tip Unique interest and professional relationships usually require more curation than address-book contacts.
Watch out High curation protects the network but can make new-user activation difficult.
- 5
Balance Defense and Onboarding
Create recommendations, starter packs, or guided discovery that reduce activation friction without making the entire graph portable.
Pro tip Help users discover value before asking them to construct a full graph.
Watch out Over-automating the graph can weaken the distinctiveness that makes it defensible.
In the wild
A Twitter user may follow publications, specialists, and distinctive personalities who are absent from the user’s contacts. By contrast, a messaging competitor can import the user’s phone contacts and quickly reconstruct the same network.
→ Twitter is harder to replace, while contact-based messaging is easier to adopt and easier to replicate.
A new professional network asks users to curate trusted experts based on years of collaboration. Because those trust relationships are not available in an address book, a competitor cannot reproduce the graph through a simple import.
→ The community gains defensibility but must invest heavily in recommendations and onboarding.
Common mistakes
Equating Scale With Defensibility
Millions of users do not create a strong moat when their relationships can be imported instantly into a rival product.
Ignoring the Onboarding Tax
A unique graph may resist competition while also leaving new users unsure whom to follow or why the product matters.
Measuring Account Creation Instead of Value
The relevant metric is how quickly equivalent network utility can be rebuilt, not how quickly a profile can be opened.
Is it for you?
Best for
It is best for founders and investors assessing the defensibility and growth constraints of networked products.
Not ideal for
It is not ideal for products whose value comes primarily from proprietary technology rather than user relationships.
From the transcript
“there is no like starter kit for your graph for the interest graph on Twitter, like there was for Facebook.”
“You import them from your phone and you've rehydrated the exact same network in a matter of minutes.”
“The beauty of a LinkedIn or a Twitter is they're difficult networks to build.”
From the episode
Twitter (with Dick Costolo)
Dick Costolo