A client intake prompt that produces a usable social content brief
Ask an AI assistant to separate confirmed facts, missing decisions and proposed ideas before it writes posts. Give it one client’s approved material and require a source for each fact. Your intake is complete when the client has confirmed a usable message and next action—not when every box in a questionnaire contains text.
Start with a small evidence packet
Our proposed intake packet has four parts: the service being sold, the audience it serves, a small set of approved facts, and the action a reader should take. Add the person who can approve the finished brief. Keep preferences such as tone separate from facts such as service area. That distinction lets the reviewer disagree with a style choice without accidentally changing what the business actually offers.
OpenAI’s prompting guidance describes providing relevant context and constraining an answer to a chosen set of resources. For agency intake, our application of that idea is a short client-specific source packet. It is a drafting aid, not a guarantee that every response will be accurate. Check what the assistant returns against the supplied material.[1]
Use the prompt to find decisions, not fill gaps
Copy this prompt: “Use only the client material below. Return four sections: confirmed facts with their source sentences; missing decisions that would change a post; three proposed content angles; and one draft based only on confirmed facts. Mark ideas as proposals. Do not invent prices, service areas, availability, testimonials or guarantees. Ask no more than three questions first, ordered by their effect on the deliverable. Do not publish or schedule.”
The question limit is a suggested operating rule. It forces the first conversation to address decisions that matter now. For example, asking which service the client wants to promote is more useful than requesting ten adjectives for the brand’s personality. If a material detail remains unresolved after three questions, keep it unresolved. The limit is a way to focus the conversation, not permission to manufacture an answer.
Work through one fictional intake
Imagine Juniper Bookkeeping supplies this approved note: “We prepare monthly bookkeeping reports for independent design studios. Our introductory conversation is free. We are taking enquiries through our contact page.” The agency has no approved price, service territory or turnaround claim. The sample output should preserve those gaps while still making the next useful draft possible.
A supportable draft is: “Run a design studio? Juniper Bookkeeping prepares monthly bookkeeping reports for independent design businesses. Start with a free introductory conversation through our contact page.” This is fictional copy for the example. The assistant should not turn it into “Never worry about tax again,” claim that appointments are limited, or assign a price. Those additions cannot be traced to the note.
The worksheet shows how to record that result. Use the real approved destination when you work with a client; the example address below is deliberately fictional. Save the accepted brief with a version label and identify the approver. If the service or destination changes later, update the brief before asking the assistant for another batch.
Turn intake into a repeatable agency service
Sell a defined first deliverable: a confirmed brief and one reviewed draft, followed by the agreed ongoing service. Explain what client input you need and when you need it. A brief full of unresolved claims should not trigger a promise to schedule a month of content. Track the questions that repeatedly delay intake, then improve your next client questionnaire using those observations.
If you will operate several clients, Sociamonials describes the AgencyPro API as supporting separate client workspaces and optional approval routing. Its current product page requires an active AgencyPro subscription for that scope. Treat the worksheet as an operating document; do not assume that pasting it into an AI assistant creates or configures those workspaces for you.[2]
For the first week, keep the accepted brief visible beside the draft review. When a client rejects a post, classify the cause: missing fact, wrong emphasis, unsuitable tone or a changed offer. This small record helps you decide whether to change the brief, the prompt or the service agreement. Changing all three at once makes it harder to learn what helped.
Worked example: Juniper’s first accepted brief
| Field | Illustrative value | Review decision |
|---|---|---|
| Service | Monthly bookkeeping reports for independent design studios. | Supported by the supplied fictional note. |
| Next action | Request a free introductory conversation at https://juniper.example/contact | Replace the fictional address with the client-approved URL. |
| Unconfirmed | Price, territory and response time are not supplied. | Ask which details are needed now; omit them from this draft. |
| Draft J1 | Juniper Bookkeeping prepares monthly bookkeeping reports for independent design businesses. Start with a free introductory conversation through our contact page. | Trace both sentences to the supplied note; approve style separately. |
| Approval record | Example: owner approves brief B1 and draft J1 for the agreed profile. | This is a hypothetical record. Capture the real approver and version in actual work. |
When this approach needs adjusting
- Juniper, the source note, URL and approval record are fictional; no customer result or hands-on product test is claimed.
- A source packet may itself be wrong or outdated. The client must confirm facts and permission to use supplied material.
Return to this next week
Review the next draft against B1 before creating new intake questions. Update the brief only when a real fact, objective or approved destination changes. If you are building this into a service for several clients, compare the AgencyPro setup scope with the work your agency will perform.
See how AgencyPro fits your agency