Create a copy-paste AI prompt bank for a [AGENCY TYPE] serving [CLIENT TYPE]. Our current services are [SERVICES], and our main operating constraints are [CONSTRAINTS AND APPROVALS]. Build prompts for these priority needs: [PRIORITY WORK]. Ask up to 3 clarifying questions only when a required input is missing. Otherwise, provide 8 distinct prompts, each 100 to 160 words, that an account manager, strategist, creative, or analyst could use. Select the most relevant tasks from discovery synthesis, client briefing, positioning options, campaign concepts, content calendars, paid-media analysis, reporting narratives, meeting agendas, change-order scopes, QA checklists, and client emails. Each entry must include: a short title, a copy-paste prompt with its own bracketed fields, its exact output format, and a one-sentence use case. Build prompts around agency realities: separate facts from assumptions, distinguish client-approved claims from proposed language, identify dependencies and decision owners, and avoid promises about results. Campaign and content prompts must require audience, channel, objective, offer, constraints, and measurement definition. Reporting prompts must require the reporting period, baseline or comparison period, data source, and caveats about attribution or incomplete tracking. Do not invent performance data, customer quotes, legal approvals, product capabilities, competitor claims, budgets, or case-study results. For regulated categories, require client or legal review before publishing claims. Do not include confidential client information beyond the supplied inputs. Before answering, check that no prompt can turn an unverified assumption into a client-facing fact. Flag the places where a human approval, data pull, or source check is needed.
Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [AGENCY TYPE] | State the kind of agency, such as brand, paid media, PR, web, or full-service. | B2B demand-generation agency. |
| [CLIENT TYPE] | Describe the industries, company sizes, or buyer groups your agency serves. | US software companies with 50 to 500 employees selling to operations and finance leaders. |
| [SERVICES] | List the services your team actually delivers and any services you do not offer. | Paid LinkedIn, landing pages, email nurture, webinar promotion, analytics reporting, and campaign strategy; no media buying outside LinkedIn and Google. |
| [CONSTRAINTS AND APPROVALS] | State budget, timeline, brand, legal, platform, privacy, and approval constraints. | Clients approve all claims and creative before launch. Do not use customer logos without permission. Most budgets are $8,000 to $25,000 per month. |
| [PRIORITY WORK] | List the agency tasks you want the prompt bank to cover first. | Discovery briefs, campaign concepts, LinkedIn ad analysis, monthly reporting, and client meeting agendas. |
How to use
- Add the agency's actual service boundaries, client approval process, and required metrics before generating the bank.
- Assign an owner to validate each output type, such as strategy for positioning and analytics for performance narratives.
- Do not paste raw client exports or unreleased plans unless your tool and contract permit it; summarize sensitive data instead.
- Use this follow-up: "Turn the reporting prompt into a version for an executive client who wants a five-bullet email and a one-table appendix."
Variations
Campaign brief
Use this at the start of a campaign before creative production begins.
Create a campaign brief for [CLIENT AND OFFER] aimed at [TARGET AUDIENCE]. The objective is [BUSINESS OBJECTIVE], the channel mix is [CHANNELS], and the constraints are [BUDGET, TIMELINE, BRAND OR LEGAL LIMITS]. Return a two-page brief with audience problem, single-minded proposition, proof points marked approved or unverified, message hierarchy, creative territories, funnel path, measurement plan, dependencies, and approvals needed. Do not invent customer evidence or benchmarks. Flag missing conversion definitions, landing-page requirements, and claims needing client review.
Performance narrative
Use this to turn a verified campaign data pull into a client report draft.
Turn these verified campaign results into a client-ready performance narrative: [RESULTS DATA]. Compare [REPORTING PERIOD] with [COMPARISON PERIOD] using [DATA SOURCE]. The stated campaign objective was [OBJECTIVE]. Produce an executive summary, metric table, drivers of change, limitations, next tests, and decisions requested. Separate observed facts from hypotheses. Define attribution limitations and do not claim causation from correlation. Check that percentage changes, totals, and metric definitions agree with the supplied data; flag missing spend, tracking, or baseline data.
Scope change
Use this when new client requests require additional time, cost, or approvals.
Draft a clear scope-change document for [NEW REQUEST] on the [PROJECT NAME] engagement. The signed scope includes [CURRENT SCOPE], while the requested addition requires [NEW DELIVERABLES]. Include rationale, deliverables, exclusions, client inputs, timeline impact, revision limits, assumptions, price or effort placeholders, and approval language. Keep the tone constructive and factual. Do not invent contractual terms, fees, dates, or commitments. Check that the request is not quietly absorbed into existing scope and flag anything that requires account-lead or legal review.
Tips
- Give the model your actual metric definitions; leads, qualified leads, pipeline, and revenue are not interchangeable reporting terms.
- Mark claims in source material as approved, substantiated, proposed, or unknown before asking for client-facing copy.
- Use AI to expose dependencies in a plan, such as creative approvals, tracking setup, landing-page access, and customer proof points.
- For reporting, provide raw metric labels and date ranges so the draft cannot silently compare mismatched periods.
FAQ
Can an agency use AI to write client deliverables?
Yes, for structured drafts and options. A responsible owner should validate claims, strategy, metrics, and contractual language before anything reaches the client.
Can AI analyze campaign performance?
It can summarize verified data and suggest testable hypotheses, but it should not infer causation without adequate evidence or complete tracking.
How do we protect client confidentiality?
Use approved tools, minimize pasted data, remove identifiers where possible, and follow the confidentiality terms in each client agreement.