Block one hour a week and run every app through the same table: is it up, what broke, what does it cost, who signed up. Then mark each app build, fix or ignore. You stop deciding from scratch every Sunday night, and you finish knowing exactly where the next hours go.
Running a SaaS alone means you are the product team, the marketing team, the support team and the admin team. Run several and the load stacks. Portfolio operators are described as stitching together Stripe dashboards, separate analytics accounts and spreadsheets, and the same report says that at 10 or more apps the overhead can swamp a solo founder. A weekly table is the cheap fix for the stitching part.
Block the time first, then size it with one formula
Pick one fixed hour, the same day each week. Put it in your calendar like a meeting with someone you cannot cancel on.
Then check that the hour fits. Minutes per app times number of apps has to leave room for one fix.
Example: say you give each app 10 minutes and you run 4 apps. That is 40 minutes of checking, which leaves 20 minutes of a 60-minute block for one fix. Swap in your own numbers. If the checking alone is longer than your block, shorten the check on apps with no users, or you are running more apps than your hours allow. More on that near the end.
Copy this table and fill in one row per app every week
| App | Up? (yes/no) | Errors or support messages this week | Errors last week | Cost this month | New users or signups | This week: build / fix / ignore |
|---|---|---|---|---|---|---|
| App one | ||||||
| App two |
Keep it in one note or one sheet. One place to look beats five dashboards. Do not add revenue to it. This table is about whether each app is alive and what it costs you.
Check the apps in this order every week
- Open each app and do the first thing a user does. Load the home page, sign in, or submit the main form. A page that loads can still have a broken form.
- Read the error log and the support inbox for each app. Write the count in the table, not a feeling.
- Compare errors to last week. This week minus last week. A positive number means worse, even if nobody complained yet.
- Write what each app cost this month. Hosting, domain, API and tools.
- Mark each row build, fix or ignore. The rules for this are below.
- Do the fixes first, then build time. Never the other way round.
- Write one line of notes per app. Next week you start from the note, not from memory.
This order is my working approach, not a tested method. Run it for four weeks, then move the steps around to fit your apps.
Set up two tools once so the weekly review is a read, not a hunt
The checklist is easy to keep if the numbers come to you. You need one error tracker and one uptime check, and that is all. Guides for solo founders suggest alerting loudly only on critical flows and sending everything else to a weekly review, because too many alerts get ignored. The same guide's weekly questions are the ones in the table: are errors up or down, are pages slower, is there a new type of error, did uptime issues repeat, and is alert noise too high.
A workable setup per app, on free tiers:
- One error-tracking project per app, so errors from different apps do not mix. Sentry is the one these guides name.
- One uptime check on the main URL. UptimeRobot is the one they name.
- One alert, on the critical flow only: sign in, payment, or the main form.
A guide for monitoring errors and downtime in an AI-built app walks through the same idea for apps you built with a coding agent. If an alert fires and turns out to be noise, change or delete it before next week. Noise is a bug in your monitoring.
For the agent side of this, read how to monitor AI agents in production when you run several apps alone.
Mark each app build, fix or ignore with these rules
| What the table shows | Mark it |
|---|---|
| Down, or a user-facing flow is broken | Fix. Before anything else this week. |
| Errors up from last week, nobody complained yet | Fix. Small, this week. |
| Up, no complaints, no new users | Ten-minute check only. Ignore the rest. |
| Two apps want build time | The one with real users or a waiting fix goes first. |
| Costs money, no signups for a quarter | Ignore, and decide whether to keep paying. |
I go deeper on the call between building, fixing and leaving things alone in build, fix or ignore: how to decide each week. The table above is the short version.
Check cost every week and audit every quarter
Small recurring charges are the ones nobody looks at, and each one is paid every month until you cancel it. Keep the cost column in the table up to date each week, and block a longer audit once a quarter.
Two numbers to compute with your own totals:
- Cost per app per month = total monthly spend on tools and hosting divided by the number of apps. Example: say you pay 120 a month across 4 apps. That is 30 per app. Compare each app's own bill to that average and see which one carries the weight.
- Idle share = subscriptions you have not opened in two weeks divided by all subscriptions. Example: say 3 of 12 are idle. That is 25 percent, and the audit starts with those 3.
A guide to auditing and cutting software subscriptions recommends cancelling anything you have not reached for in two weeks. Treat that as advice and re-subscribe if you miss the tool. When three or four specialized tools do one job between them, consolidate before adding another.
Quarterly audit, in order:
- Export one month of card or bank statements for the account that pays for your tools.
- List every recurring charge: tool, app it serves, monthly cost, last time you opened it.
- Mark each line keep, cancel or consolidate.
- Cancel what you marked and put the next audit date in your calendar.
- Write the new monthly total in the table.
If agents are part of your stack, cap what they can spend too. I wrote up how in AI agent spending limits.
Mistakes that cost you the week
| Mistake | What it costs |
|---|---|
| Checking only the app you like | The quiet app breaks unnoticed. The first sign is a user message. |
| Alerting on everything | Alert fatigue. The real alert gets ignored. |
| One error project for every app | You cannot tell which app broke. |
| Paying for monitoring before the free tiers run out | An avoidable monthly cost on every app. |
| Spreading costs by feel | The app that earns nothing keeps getting funded. |
| Skipping the note at the end | Next week starts from memory and you repeat the same checks. |
When the hour does not fit, you have too many apps
If you cannot finish the table in the block three weeks in a row, the problem is the number of apps, not your discipline. Move the quietest app to a ten-minute check, or retire it. A portfolio that fits your hours beats a bigger one that is always on fire.
Do this in the next ten minutes
Open a note and paste the table. Add a row for each app you run. Fill in the Up? and Cost this month columns right now, from memory and your last card statement. Then pick the day and hour for your weekly block and put it in your calendar.
Send me the app that keeps breaking
If one app in your portfolio eats your week and you want a second opinion on what to automate, email me with the app and the manual task. I read every message.
