The Inventory Method: How to Find Your Single Points
You have ten people. One handles invoicing. One runs the CRM. One manages the supplier portal. One holds the AWS keys. One knows how the custom report works. If any of them disappears tomorrow, the work stops. That is the definition of a single point of failure. Business continuity automation starts with an inventory that takes two hours and a spreadsheet.
List every role. List every system. List every credential. List every process that only one person can execute. Do not ask for documentation. Ask for the name of the person who does it today. If a cell has one name, you have a single point of failure. Count them. That number is your exposure. The metric is time to recover, not documentation produced.
A ten-person company typically finds twelve to twenty single points. The owner usually holds three to five. The fix for each costs less than the loss from one day of downtime. The arithmetic is simple. If invoicing stops for one day and you bill $5,000 daily, the loss is $5,000. A shared mailbox and a delegated rule cost $6 per month. The recovery time drops from one day to fifteen minutes.
The One Inbox Problem: Email as a Bottleneck
One person owns the info@ address. One person owns the support@ address. One person owns the billing@ address. When that person is sick, on vacation, or leaves, messages pile up. Customers wait. Vendors escalate. Revenue stalls. The fix is not a shared password. The fix is a shared mailbox with delegation rules.
Set up a shared mailbox in your email platform. Grant full access to two people. Create rules that tag messages by type: invoice, support, sales, vendor. Create a second rule that forwards any message older than four hours to a Slack channel or a second inbox. The arithmetic: if your team receives thirty messages a day across three addresses and the owner is out for eight hours, ten messages sit untouched. At an average deal value of $2,000 and a 10 percent close rate from inbound, that is $2,000 at risk per day. The shared mailbox costs zero extra licenses on most plans. The setup takes thirty minutes.
Test it. Send a test message to each address. Verify the tag appears. Verify the forward triggers at four hours. Verify the second person sees it. Document the test result in your inventory spreadsheet. That is the only documentation you need.
The One Laptop Problem: Device Dependency
The developer has the only laptop with the production database dump. The designer has the only laptop with the Figma library. The accountant has the only laptop with the QuickBooks file. A spilled coffee, a lost bag, or a failed drive stops the work. The fix is not a backup drive in the same backpack. The fix is cloud-synced workspaces with automatic sync.
Move the database dump to a shared cloud folder with version history. Move the Figma library to a team workspace. Move the QuickBooks file to QuickBooks Online with multi-user access. Enable automatic sync on each laptop. The arithmetic: a developer laptop replacement takes two days including shipping, OS setup, environment restore, and credential re-entry. At a billable rate of $150 per hour, that is $2,400 in lost capacity. A cloud-synced workspace costs $12 per user per month. The recovery time drops from two days to the time it takes to log in on a new device.
Test it. Have each person log in on a spare device. Verify they can open the database, the design file, the accounting file. Time the login. Record the time in your inventory. That is your recovery baseline.
The One Account Problem: Credential Silos
The AWS root account uses the founder’s personal email. The Stripe account uses the CTO’s phone for 2FA. The domain registrar uses an email address from a former employee. The payroll provider requires a hardware token on the controller’s keychain. When the person leaves or loses the device, you cannot administer the service. The fix is not a password spreadsheet. The fix is a team password manager with shared vaults and emergency access.
Move every credential into a team vault. Organize vaults by function: infrastructure, payments, domains, payroll, marketing. Grant access by role, not by person. Enable emergency access that triggers after a waiting period and notifies a second owner. The arithmetic: a locked AWS root account takes three to seven days to recover through support. During that time, you cannot launch instances, modify security groups, or access billing. If your infrastructure costs $800 per month and you cannot shut down idle resources, you waste $200 per week. A team password manager costs $4 per user per month. The recovery time drops from days to minutes.
Test it. Have a non-owner use emergency access to retrieve the AWS root credentials. Verify they can log in. Verify the notification reaches the second owner. Record the time. That is your recovery baseline.
The One Person Problem: Knowledge Concentration
One person knows how the monthly close works. One person knows the API integration with the fulfillment partner. One person knows the unwritten rules for the custom discount matrix. When that person is unavailable, the process halts or produces errors. The fix is not a wiki nobody reads. The fix is recorded walkthroughs stored in the same place the work happens.
Record a screen capture of each critical process. Narrate the steps. Explain the exceptions. Store the video in the project folder or the process documentation space. Link the video from the task template in your project tool. The arithmetic: the monthly close takes six hours when the expert does it. When the backup does it without guidance, it takes fourteen hours and produces two errors that each cost four hours to fix. That is twenty-two hours versus six. At $75 per hour, the difference is $1,200 per month. A screen recording tool costs zero. The recording takes one extra close cycle. The recovery time for the backup drops from fourteen hours to seven hours after one guided run.
Test it. Have the backup execute the process using only the video. Time it. Count the errors. Record both numbers. That is your recovery baseline.
Building Recovery Time Into Your Operations
You now have an inventory. You have a fix for each single point. You have a recovery baseline for each. The next step is to make recovery time a standing agenda item. Put it on the monthly operations review. Review the inventory. Update the count. Re-test one single point each month. Rotate through the list. The arithmetic: twelve single points tested once per year means one test per month. Each test takes thirty minutes. That is six hours per year per person. At $75 per hour, the cost is $450 per year. The alternative is discovering a broken recovery path during an actual outage. The cost of that discovery is measured in lost revenue, not hours.
Assign ownership of the inventory to a role, not a person. When the role changes hands, the inventory transfers. The new owner runs the full test suite in their first week. That is the onboarding task that proves the system works. No handoff document replaces a live test.
If you want an outside eye on your inventory and your recovery baselines, AI consulting for your operation can run the assessment in a single session. If you want to see the tooling we build for this exact problem, review what we build. The goal is not replacing staff. The goal is ensuring the work continues when the person cannot.
Single points of failure do not disappear as you grow. They multiply. The ten-person company that fixes twelve today becomes the fifty-person company with sixty. The inventory method scales. The fixes scale. The recovery metric stays the same. Time to recover is the only number that matters when the work stops. Start the spreadsheet today. Count the single names. Fix the cheapest one first. Measure the recovery time. Repeat next month.