One operating standard, not one per business

Three ventures in, this works. Four is where it stops.

Each venture gets its own checklist and its own idea of finished. Each is a different business with different customers and different risks. Different things need different handling. Nobody argues with that, and nobody should, because at three it is true.

The failure is not dramatic, and nothing explodes, nothing gets lost, nobody complains. You just stop opening the quiet ones, because opening one means relearning its private language first. That is a tax you will not pay on a busy afternoon.

It does not cost you a business. It costs you attention, in the gap between deciding to check something and being able to.

What I run instead is one written operating standard for everything, one documented playbook per venture in a shared repository, and a scorecard per venture scored against the same checks. 850+ domains, 150+ agents in production, 100+ Base44 apps. None of that survives a different dialect per venture.

The rules

One standard, one file, every venture, no exceptions granted at the start. The standard is the checks you run against anything you own. Not the work itself. The same checks, the same shape of documentation, the same definition of done. If a venture cannot be scored by it, that is information about the venture, not a reason to fork the standard.

A standard is not a mission statement, not a brand guide, not a set of values, not a document you wrote once and have never once failed against. If nothing has ever scored badly against it, you have written a description of what you already do. That is worthless. The standard earns its place the first time it tells you something you did not want to hear.

Keep the standard short enough to hold in your head. Eight to fifteen checks. Every check answerable yes or no by someone who is not you. Anything that needs a paragraph to decide is not a check, it is an opinion. Opinions do not survive the fourth venture.

Per-venture detail lives in a playbook, not in the standard. One playbook per venture, all of them in the same repository, all of them in the same section order. The standard says every venture has a backup. The playbook says where this one's backup is and how to restore it. Keep those two apart. Do not let the standard rot into a pile of special cases.

Score every venture against the same checks, and keep the scores where you can see them side by side. A scorecard per venture, same columns, same wording. The value is not the score. It is that a venture failing check four looks identical to another venture failing check four. You fix one thing across the portfolio rather than the same thing eight times.

Do it now

Open a new file, call it your operating standard, and put it where every venture can reach it.

Write the checks and aim for ten, using this shape without leaving a single one of them unchanged:

1. There is one documented playbook, in the shared repository, updated within 90 days.
2. There is a restorable backup, and it has been restored once as a test.
3. There is one named owner for anything customer-facing.
4. Money in and money out are both readable without asking anyone.
5. Every scheduled task has an off switch and a stated end condition.
6. No credential exists in only one person's head or one person's laptop.
7. The public price matches what the payment processor actually charges.
8. Someone who is not you can complete one routine task from the playbook alone.
9. There is a written definition of done for the venture's main recurring job.
10. The scorecard for this venture was updated within 30 days.

Now build the scorecard as one table, one row per venture, one column per check, a yes or a no in every cell. Fill it honestly and fill it today. Do not improve anything yet.

Then do the part people skip. Find the check that the largest number of them fail. Fix that one check everywhere before you fix anything else anywhere.

A worked example

Say you run three businesses that share nothing but you. A domestic cleaning company with a small team. A vending-machine round across offices and gyms. A small online shop selling refurbished bikes. There is also a bike-hire trial from last spring, with its own site, that nobody has touched since.

That makes four rows. You write ten checks in your own words and score all four in one sitting.

Check 3:   Each business has one named person who answers its customers.
Check 6:   No login lives only in the owner's head or on the owner's phone.
Check 7:   The price a customer sees is the price the card reader or checkout takes.
Scorecard   1    2    3    4    5    6    7    8    9    10
Cleaning    yes  no   yes  yes  yes  no   yes  yes  yes  yes
Vending     no   yes  yes  yes  yes  no   no   no   yes  yes
Bikes       yes  yes  yes  no   no   no   yes  no   no   yes
Bike hire   no   no   no   no   no   no   yes  no   no   yes
Fails most: check 6, a no in all four rows

Every check is a yes or a no that somebody else could answer.

The failure you will hit

Two things. The first is a scorecard on which everything passes everything, which means your checks describe what you already do and measure nothing. Rewrite them harder. The second is the first genuine exception. A venture that truly cannot meet check three, for a real reason, and you will want to write a second standard for it. Do not. Mark the cell as a no. Write one line under the scorecard saying why, and leave it failing in public. A permanent honest no is cheaper than a fork. A fork is how you get back to a process per business, slowly, while feeling organised.

Not for you if

You run one thing and you intend to keep it that way. A single standard across a single venture is just a checklist with extra ceremony, and you do not need the overhead.

One standard, applied everywhere, is less work than five that drift. Consulting & Strategy is one of the Vault's categories.