# AI operating system readiness scorecard

Twenty statements in five areas that tell a marketing agency whether it is ready to run client work on an AI operating system, with a score, three bands and the first fix for your lowest area. RevSmith, updated 4 November 2026.

The online version, which works out the score in your browser and sends nothing, is on [What is an AI operating system for marketing agencies?](https://revsmith.in/ai-operating-system-for-agencies). This file is the same scorecard to print or copy.

## How to use it

1. Answer for the agency as it runs today, not as you plan to run it. If you hesitate between two answers, take the lower one.
2. Score each statement: **No = 0**, **Partly = 1**, **Yes = 2**. "Partly" means true for some clients, or some of the time.
3. Add up each area (0 to 8) and the total (0 to 40).
4. Read your band, then the first fix for your lowest area.
5. Better still, ask two colleagues to score it on their own. Where your answers differ by two points, you have found a process nobody owns.

It takes about ten minutes.

## The twenty statements

Each statement is written so that Yes means ready.

### Client context

| # | Statement | Your points (0, 1 or 2) |
|---|---|---|
| 1 | Each client's goals, scope, brand rules and contacts are written down where the team can find them. | |
| 2 | A new team member could learn a client's history without asking the account lead. | |
| 3 | Decisions and promises from client calls are filed under the client within a day. | |
| 4 | You know which client information must never reach another client, and it is kept apart. | |
| | **Client context subtotal (0 to 8)** | |

### Repeatable work

| # | Statement | Your points (0, 1 or 2) |
|---|---|---|
| 5 | At least one job, such as the monthly report, is done the same way for most clients. | |
| 6 | The steps of your repeated jobs are written down as checklists or SOPs. | |
| 7 | You can name the three repeated jobs that take the most hours each month. | |
| 8 | Repeated jobs add up to at least ten hours a week across the team. | |
| | **Repeatable work subtotal (0 to 8)** | |

### Tools and data

| # | Statement | Your points (0, 1 or 2) |
|---|---|---|
| 9 | Your client work lives in tools that can be connected, not only in files on laptops. | |
| 10 | You hold the access you need to each client's ad, analytics and social accounts. | |
| 11 | For each number you report, the team agrees which tool is the source of truth. | |
| 12 | One person owns your tool stack and can connect a new app without a developer. | |
| | **Tools and data subtotal (0 to 8)** | |

### Approvals and risk

| # | Statement | Your points (0, 1 or 2) |
|---|---|---|
| 13 | You have a written rule for what a person checks before anything reaches a client. | |
| 14 | You have decided who may send to clients, publish posts or change ad spend. | |
| 15 | Your client contracts allow AI to work on their data, or you have asked each client. | |
| 16 | You keep a record of what was sent to each client, which version, and who approved it. | |
| | **Approvals and risk subtotal (0 to 8)** | |

### Scale and economics

| # | Statement | Your points (0, 1 or 2) |
|---|---|---|
| 17 | At least five people do client work in your agency. | |
| 18 | You serve at least ten active clients or retainers. | |
| 19 | You know how many hours each client takes a month and what that time costs. | |
| 20 | One person can give a pilot a few hours a week for a month, with a budget agreed. | |
| | **Scale and economics subtotal (0 to 8)** | |

## Your total

| Area | Your points | Out of |
|---|---|---|
| Client context | | 8 |
| Repeatable work | | 8 |
| Tools and data | | 8 |
| Approvals and risk | | 8 |
| Scale and economics | | 8 |
| **Total** | | **40** |

Your lowest area is the one with the fewest points. If two areas tie, start with Approvals and risk if it is one of them, then Client context.

## What your score means

| Score | Band | What to do next |
|---|---|---|
| 0 to 14 | **Not yet: fix the basics first** | Write down how your most repeated jobs are done, put each client's goals, scope and contacts in one place, and decide who checks work before it reaches a client. [The agency operations guide](https://revsmith.in/guides/agency-operations) shows where to start. Score again in a month. |
| 15 to 27 | **Ready for a pilot on one workflow** | Run one repeated job with a clear owner for a month, with a person approving every output. Time it before and after, and note each run's AI cost. [Score your tasks in the automation audit](https://revsmith.in/ai-for-marketing-agencies). |
| 28 to 40 | **Ready to run on an AI operating system** | Give each client its own context, connect the tools that hold client data, and move repeated jobs over one at a time, with approval on anything that reaches a client. Put the ten vendor questions in the [decision matrix](https://revsmith.in/templates/ai-os-decision-matrix.md) to every tool on your list. |

Whatever your band, fix a low Approvals and risk score before any AI agent touches client work.

## Your lowest area: the first fix

| Lowest area | What it usually means | First fix, this month | Read next |
|---|---|---|---|
| Client context | What is true about each client lives in people's heads and inboxes | Write a one-page record for your three biggest clients: goals, scope, brand rules, contacts, open promises. Keep it where the whole team can see it. | [Client onboarding checklist](https://revsmith.in/templates/client-onboarding-checklist) |
| Repeatable work | Every client is served differently, so there is nothing to automate yet | Pick the job you do most often, write its steps as a checklist, and run it the same way for every client for a month. | [Agency operations](https://revsmith.in/guides/agency-operations) |
| Tools and data | Numbers cannot be traced, and access is borrowed or shared | List each client's accounts, who owns each one and your access level. Agree one source of truth for every number you report. | [Marketing agency tech stack](https://revsmith.in/guides/marketing-agency-tech-stack) |
| Approvals and risk | Nobody has decided what may leave the agency, or who says so | Write one paragraph: who may send to clients, publish posts or change ad spend; what is checked first; where each approval is recorded. Check your client contracts for AI terms. | [AI agents for marketing agencies](https://revsmith.in/guides/ai-agents-for-marketing-agencies) |
| Scale and economics | Too few repeated jobs to pay back setup, or nobody has time to run a pilot | Log the hours each client takes for one month. Then name one pilot owner, a budget and an end date. | [Agency profitability](https://revsmith.in/guides/agency-profitability) |

## Example: one agency's answers

An example, not a customer: a 12-person performance agency with 18 retainers.

| Area | Answers (statements in order) | Points |
|---|---|---|
| Client context | Partly, Partly, No, Yes | 1 + 1 + 0 + 2 = 4 |
| Repeatable work | Yes, Partly, Yes, Yes | 2 + 1 + 2 + 2 = 7 |
| Tools and data | Yes, Yes, Partly, Partly | 2 + 2 + 1 + 1 = 6 |
| Approvals and risk | Partly, No, Partly, No | 1 + 0 + 1 + 0 = 2 |
| Scale and economics | Yes, Yes, Partly, Partly | 2 + 2 + 1 + 1 = 6 |
| **Total** | | **4 + 7 + 6 + 2 + 6 = 25** |

25 sits in the band **Ready for a pilot on one workflow**. The lowest area is Approvals and risk (2 of 8), so before the pilot starts the agency writes down who may send to clients and where approvals are recorded. Its pilot is the follow-up after client calls: 36 calls a month at about 45 minutes each is 27 hours, the job repeats the same way for every client, and it touches client context, tools and approvals at once.

## Notes

- This scorecard is a guide to where to start, not a benchmark. The example answers are illustrative, not survey data.
- It applies whichever tool you choose, including none. RevSmith makes this scorecard and builds an AI operating system for marketing agencies; we would rather you score "Not yet" and fix the basics than start a pilot that cannot succeed.
- Questions or corrections: founder@revsmith.in.