Guide
Bus factor in a two-person project: check who can recover it
If one of you disappears for a month, can the other keep the project running? You can find out in 30 minutes, without locking anyone out.
Not legal or security advice. Account recovery rules differ by provider. Check each provider's current help pages before you rely on a backup plan.
What bus factor means
Bus factor is the number of people who could leave before a project stops. In a two-person project, it is usually one. That is not a problem by itself. It becomes one when you have never checked.
"Leave" does not have to be dramatic. Illness, a new job, a holiday without a laptop or a lost phone with the two-factor app on it all count.
Two different questions
- Knowledge: does the other person know how it works? Where is the code, how is it deployed, which service sends the emails?
- Access: can the other person actually do it? Log in, deploy, renew the domain, pay the bill.
Knowing is not the same as being able to. Read access to a repository does not mean someone can deploy it. Being listed as a backup contact does not mean the provider will let them in. Test the two separately.
Step 1: build the dependency table
List everything the project needs to run or to get paid. Copy this table and fill one row per resource.
| Resource | Who knows how it works | Who is the owner or admin | Who pays | Recovery route | Second person tested? | Gap and next action |
|---|---|---|---|---|---|---|
| Domain | ||||||
| Code repository | ||||||
| Hosting or deploy | ||||||
| Email and DNS | ||||||
| Payments or bank | ||||||
| Social and ad accounts |
Do not write passwords or recovery codes in this table. Write where the recovery route is, for example "password manager, shared vault" or "phone number of A".
Step 2: run a read-only rehearsal
For each row, the second person tries to prove they could take over. Use only steps that change nothing.
- Log in as yourself. Can you open the admin page of the account with your own login? If you cannot, mark a gap.
- Find the recovery path. Open the provider's help page and read what recovery needs: a phone, an email address, an ID. Write it down.
- Check who can see the bill. If you cannot see the invoice, you cannot tell when the renewal is due.
- Check the date. Write down the next renewal date for the domain and each paid plan.
Do not revoke access, rotate production secrets, change a password or log the other person out. A rehearsal that breaks something is not a rehearsal.
A worked example (illustrative)
Two founders. A set up the domain, the hosting and the payment card. B wrote half the code and has read access to the repository.
| Resource | Owner | Payer | B tested? | Result |
|---|---|---|---|---|
| Domain | A | A | No | Gap: B cannot see the renewal date. |
| Repository | A | n/a | Read only | Gap: reading code does not prove B can deploy. |
| Hosting | A | A | No | Gap: B has no login. |
All three rows show the same pattern: B knows the project, but cannot act. The next action is to add B as a second admin and to move the billing email to a shared project address. This example is made up to show the method. It is not a real project.
Step 3: close the gaps
- Two admins per account wherever the provider allows it. If it allows only one owner, write down how ownership can be transferred and test that route on a spare account, not on the real one.
- A shared project email for billing and recovery, not one person's inbox.
- A password manager with a shared vault for the accounts that cannot have two admins.
- A short runbook: three lines per resource for how to deploy, how to renew and who to contact.
Step 4: repeat it
Run the table again when someone joins or leaves, when you add a paid tool and before each renewal. Fifteen minutes each time is enough once the table exists.
Do it with AI
Turn your tool list into the table
Paste your list of tools, without any passwords, into Claude or ChatGPT.
I run a two-person side project. Below is the list of tools we use, with who signed up and who pays. Do not ask for passwords or recovery codes. Build a table with these columns: resource, who knows how it works, owner or admin, payer, recovery route (describe, do not invent), second person tested (leave empty), gap and next action. Then list the five gaps most likely to stop the project if one of us is unavailable for a month. Tools: [PASTE HERE]
Frequently asked questions
What is a good bus factor?
Two or more for every resource that matters: the domain, the code, the deploy and the money. For a two-person team that means each of you can do each job.
Is single point of failure the same thing?
Close. A single point of failure is one thing or person the project depends on. Bus factor counts the people. Both ask the same question: what happens if it is gone?
Is a password manager enough?
It is a good start for accounts that cannot have two admins. You still need to test that the other person can open the vault and that the recovery route works.
Should I share my passwords with my co-founder?
Do not send passwords in chat or email. Use a shared vault in a password manager, or add them as a second admin so each person has their own login.
Related
- Access audit: who could lock you out today?
- Handover checklist for when someone leaves.
- Team charter template: who owns, pays for and can recover each account.
- Who owns the domain when co-founders split
Run the access audit to see which accounts have only one person with full control.