TL;DR: Here are 30 AI prompts for agencies grouped across five client deliverable types — strategy documents, creative briefs, campaign copy, reporting, and presentations. Every prompt uses [bracketed variables] you fill in per client. Save the set to a shared prompt library and every account manager starts from the same foundation, inside any AI tool, without rebuilding from scratch each time.
What are the best AI prompts for agencies in 2026?
The best AI prompts for agencies are task-specific templates that include client context, output format requirements, and a clear role instruction — not open-ended requests the model has to interpret from scratch. When an account manager asks "write a creative brief for the summer campaign," the model guesses the client's audience, tone, channel mix, and objectives. When the same manager runs a structured prompt with those fields already defined, the output is a first draft they can review and send, not a blank template the team still has to populate.
The five deliverable types where structured prompts produce the strongest return are strategy documents, creative briefs, campaign copy, client reporting, and presentation outlines. Each type requires different prompt logic: a strategy document prompt needs business objectives and competitive context; a creative brief prompt needs campaign goals and brand constraints; a campaign copy prompt needs channel specs and character limits. Conflating these — asking one vague prompt to cover all five — produces vague output that fits none of them.
This guide gives you 30 prompts across all five types, each with [bracketed variables] you fill in once per client and reuse across every similar deliverable. For the system that lets you switch between ten clients without rebuilding each prompt from scratch, see how agencies set up per-client variables and contexts. To build a shared library that survives staff turnover, the agency prompt knowledge retention guide covers that workflow. If your immediate need is account management communication — scoping calls, status updates, difficult emails — the client communication prompts guide covers those separately. And if you are managing AI work across an agency, the landing page maps all five deliverable types to specific features.
What do most agency AI prompt lists get wrong?
Most agency prompt collections treat AI as a content generator and stop at social captions, ad headlines, and email subject lines. Those are useful, but they represent a narrow slice of the billable work an account team produces in a week.
The higher-order deliverables — strategy documents, creative briefs, performance reports, QBR presentations, campaign frameworks — either get ignored or treated as too complex for AI assistance. In practice, these are where AI saves the most time, because they are the work that currently takes the longest. A creative brief that takes 90 minutes from a blank page takes 25 minutes when the model already has the client's audience, objectives, and brand constraints in the prompt.
The second gap: most lists provide prompts as disconnected one-offs. The audience profile from the strategy document should feed into the creative brief. The campaign objectives from the brief should anchor the reporting narrative. This guide structures all 30 prompts as a connected workflow — each deliverable type is designed to accept the output of the one before it as input.
How do agencies structure prompts across all five deliverable types?
The structure underlying all 30 prompts is what we call the five-layer agency deliverable stack — five deliverable types mapped to the stage of client work where they appear, the input each requires, and the output format that works for that deliverable.
| Deliverable type | Prompts | Stage of client work | Key inputs |
|---|---|---|---|
| Strategy docs | 1–6 | Discovery / kickoff | Business objectives, audience signals, competitive context |
| Creative briefs | 7–12 | Pre-production | Campaign goals, brand voice, channels, constraints |
| Campaign copy | 13–18 | Production | Brief output, channel specs, character limits |
| Reporting | 19–24 | Post-campaign / monthly | Raw metrics, original KPIs, period events |
| Presentations | 25–30 | Kickoff / QBR / debrief | Synthesis of the above |
Three habits make this stack work across multiple clients. First, fill every [bracketed variable] before running — a prompt that still reads [Client name] produces output about a generic unnamed brand. Second, save each prompt to a shared library with variables intact so any account manager can pick up mid-project. Third, treat each deliverable's output as the input context for the next prompt type — paste the audience profile from prompt 3 into the creative brief prompts that follow.
How do agencies use AI for strategy documents?
Strategy documents follow predictable structures and require the same categories of input for every client: objectives, audience, competitive context, and a recommended direction. That predictability makes them well-suited to AI first drafts, yet most teams still write them from scratch each quarter because no one has built the prompt template.
These six prompts cover the full strategy layer. The format constraint in each is what separates a usable draft from an essay — specify the structure and the model cannot wander.
1. Competitive landscape summary
Client: [Client name] in the [industry] space.
Competitors: [list 3–5 with one-line description each].
Write a 250-word competitive landscape summary.
Structure: 3 sections — market context (one paragraph), how each competitor positions (bullet per competitor), white-space opportunity for [Client name] (one paragraph).
Flag any claim that requires verification before I share this with the client.
The "flag any claim" instruction keeps the model honest. It surfaces where it has filled gaps from general knowledge rather than the specific competitive data you provided — and those gaps are exactly what your account team needs to verify before the client sees the document.
2. Target audience persona
Client: [Client name]. Product/service: [one-sentence description].
Audience signals we have: [what we know — demographics, behaviors, pain points, preferred channels].
Write one ICP persona. Format: name, role/profile, primary goal, biggest fear, how they currently solve the problem, why they would choose [Client name].
Max 200 words. Use specific details, not demographic generalities.
Build this persona once per client and paste it into every subsequent prompt that requires an audience description. It eliminates the vague "target audience: 25–44, urban professionals" shorthand that produces generic copy.
3. Strategic positioning statement
Client: [Client name]. Category: [what category they compete in].
Alternatives customers currently use: [2–3 real alternatives].
Differentiator: [Client's specific, verifiable difference — not a marketing claim].
Write a Geoffrey Moore positioning statement and a plain-English translation under 20 words.
Flag the single biggest assumption in the positioning.
4. 90-day marketing strategy outline
Client: [Client name]. Primary goal this quarter: [specific goal with a metric].
Channels in scope: [list]. Budget range: [rough range].
Write a 90-day strategy outline.
Format: 3 phases (month 1 / 2 / 3), 2–3 initiatives per phase, success metric per initiative.
Keep each initiative to two sentences maximum.
5. SWOT analysis synthesis
Client: [Client name] in [industry].
Context: [paste 3–5 sentences summarizing what we know about their business].
Write a SWOT analysis. One sentence per item, 3 items per quadrant.
Rule: Strengths and Weaknesses must be internal only; Opportunities and Threats must be market-level, not company-specific.
6. Campaign goals and KPI framework
Campaign type: [campaign type for Client name].
Business goal: [what the campaign is meant to achieve].
Write a goal-to-KPI mapping. Format: business goal → campaign objective → 2 primary KPIs → 1 secondary KPI.
For each KPI: what it measures, why it matters for this client, how we track it.
How do agencies use AI to write creative briefs?
Creative briefs have a fixed structure almost every agency follows, which makes them consistent candidates for AI drafting. The failure mode is vagueness: most AI-generated briefs describe audiences in demographic terms rather than behavioral ones and list product features rather than naming the campaign's single-minded proposition.
The six prompts below are structured to force specificity at the three points where briefs most often fail: the audience definition, the key message, and the success metric.
7. Full creative brief for a brand campaign
Client: [Client name]. Campaign: [working title].
Objective: [specific goal with a metric]. Target audience: [paste persona from prompt 2].
Key message: [single sentence — what we want the audience to believe after seeing this campaign].
Channels: [list]. Mandatory elements: [legal, brand rules, must-include assets].
Budget range: [range]. Timeline: [from — to].
Write a full creative brief with these sections: Background, Objective, Target Audience, Key Message, Tone, Channel Mix, Deliverables, Timeline, Success Metrics.
8. Social media campaign brief
Client: [Client name]. Platforms: [platforms]. Campaign goal: [goal]. Duration: [weeks].
Audience: [paste persona summary].
Write a social campaign brief. Include: campaign theme, content pillars (3 max), content mix (post types and ratios), community management direction, 2 KPIs per platform.
9. Content marketing brief
Client: [Client name]. Content goal: [awareness / lead gen / retention].
Audience: [audience summary]. Target keyword cluster: [if SEO-driven, otherwise leave blank].
Write a brief for a 3-month content plan. Format: editorial calendar structure, content types, recommended channels, publishing cadence, promotion approach.
Include 3 hero topics and 6 supporting topics.
10. Email campaign brief
Client: [Client name]. Campaign goal: [specific goal]. Audience segment: [description].
Write a campaign brief for a 5-email sequence. Include: objective per email, sender voice, subject line guidance, CTA per email, tone direction, and a success metric for the sequence as a whole.
11. Paid media creative brief
Client: [Client name]. Channels: [channels]. Budget: [range].
Campaign objective: [goal]. Audience: [audience summary].
Write a creative brief for a paid media campaign.
Sections: campaign overview, audience targeting notes, creative specifications per channel (format, size, copy length), messaging hierarchy (headline → body → CTA), creative dos and don'ts.
12. Video production brief
Client: [Client name]. Video type: [brand / product / testimonial / social].
Duration: [target length]. Platform: [where it will live]. Key message: [one sentence].
Audience: [audience summary].
Write a video production brief. Include: video objective, tone and style direction, script outline (beat by beat), required on-screen elements, call to action, technical specs for the target platform.
What AI prompts work best for campaign copy?
Campaign copy is where generic AI output does the most damage — and where the fix is most straightforward. Specify the channel, the audience, the campaign objective, and the length constraint. Without those four inputs, the model defaults to the format and register it has seen most often, which is rarely the one the brief calls for.
Start each copy prompt by pasting the relevant creative brief output as context. The brief you generated in the previous section is the input; the copy prompt is the transformation step.
13. Ad headline variants
Client: [Client name]. Campaign: [campaign name]. Key message: [key message].
Target audience: [audience summary]. Platform: [Meta / Google RSA / LinkedIn].
Write 10 headline variants.
Constraints: 30 characters for Google; 40 for Meta; 50 for LinkedIn.
Angle mix: 4 benefit-led, 3 problem-led, 3 curiosity-driven.
Rank by predicted click-through rate with a one-line rationale per headline.
14. Social post pack (3 platforms)
Campaign: [campaign name for Client name]. Key message: [message].
Tone: [Client brand voice — paste voice brief].
Write one native post per platform: LinkedIn (150 words, professional tone), Instagram caption (100 words plus 3 hashtag clusters of 5 each), Twitter/X (240 characters max).
Each post must feel native to its platform — not the same copy reformatted.
15. Email campaign sequence
Client: [Client name]. Campaign goal: [goal]. Audience: [segment].
Key message: [one sentence]. Sequence duration: [number of days].
Write a 5-email sequence: E1 (D0) — introduce, E2 (D2) — educate, E3 (D5) — social proof, E4 (D7) — convert, E5 (D10) — follow-up.
Each email: subject line + preheader + 100-word body + one CTA.
Tone: [Client voice description]. No generic greetings.
16. Landing page hero copy
Client: [Client name]. Campaign: [campaign name]. Primary CTA: [CTA text].
Audience: [ICP summary].
Write hero copy. Sections: headline (10 words or fewer, outcome-led), subheadline (one sentence explaining how), 3 benefit bullets (each: feature → what it means for the audience → why it matters).
Tone: [Client brand voice].
17. Campaign concept write-up
Campaign name: [name]. Client: [Client name]. Core idea: [one sentence].
Expand into a 200-word campaign concept description for client presentation.
Format: concept overview, why it resonates with this audience, execution highlights across [channels], how success is measured.
Tone: confident but not overselling.
18. CTA variants for a campaign asset
Campaign goal: [goal]. Asset type: [landing page / email / ad / social post].
Current CTA: [existing text if any].
Write 8 CTA variants. Constraint: 5 words or fewer each.
Mix: 3 action-led, 2 benefit-led, 2 urgency-driven, 1 question-format.
Flag which variant best fits each funnel stage: awareness, consideration, conversion.
How do agencies use AI for client reporting?
Client reporting consumes more account manager time than almost any other deliverable, and most of it is spent translating numbers into sentences — exactly what AI does well. The model does not replace your analytical judgment about what the numbers mean for the client; it handles the prose layer so you can focus on the interpretation.
These six prompts cover the most common reporting scenarios.
19. Monthly performance summary narrative
Client: [Client name]. Period: [month/year].
Metrics: [paste key metrics with MoM comparison, original target, and any benchmark].
Period events: [budget changes, creative refreshes, external market events].
Write a 300-word performance narrative.
Lead with the headline result. Explain variance from target. Close with 2 recommended actions.
Rule: interpret every metric — do not describe without context.
The "rule" instruction is the most important line. Without it, the model lists what happened. With it, it explains why it happened and what to do about it — which is what the client is actually paying for.
20. Campaign results one-pager
Campaign: [campaign name for Client name]. Period: [dates].
KPIs: [list actual vs. target for each KPI].
Key win: [best result]. Biggest miss: [what fell short and the likely reason].
Write a one-page results summary.
Format: executive headline (1 sentence), results table (KPI / target / actual / variance), 3-bullet interpretation, 2 recommendations.
Max 250 words total.
21. KPI story from raw data
Raw data: [paste metrics table or paste key numbers].
Original campaign objectives: [paste].
Write a 150-word narrative that tells the story of this campaign.
Rule: do not list metrics — interpret them. Lead with the single most important finding.
Flag any metric that needs additional context before a conclusion can be drawn.
22. Channel-by-channel insight summary
Client: [Client name]. Active channels this period: [list].
Data by channel: [paste metrics per channel].
Write a channel performance summary. Structure: one paragraph per channel, using the frame — what we expected, what happened, likely reason, what to adjust.
Max 80 words per channel.
23. Executive summary for client sign-off
Report context: [paste or summarize key findings from this month].
Audience: [client executive name/role and their level of detail comfort].
Write a 100-word executive summary.
Structure: one sentence on what we set out to do, two sentences on what happened, one sentence on the recommended next step.
Tone: direct. No hedging language.
24. Competitive performance context paragraph
Client: [Client name] in [industry]. Our performance this period: [key metrics].
Market context available: [paste any industry or competitive data — or note if none available].
Write a 100-word competitive context paragraph that frames our performance against market conditions.
If market data is not available, say so explicitly — do not manufacture context.
How do agencies use AI for client presentations?
Presentation outlines are where AI saves the most time on high-stakes work. A kickoff or QBR presentation that takes four hours to structure from scratch takes under an hour when the model generates the slide-by-slide architecture and you fill in the substance. The key is specifying the audience's context and the meeting's single goal before anything else.
25. Campaign kickoff deck outline
Client: [Client name]. Meeting: campaign kickoff.
Audience: [stakeholders — role and familiarity with the work].
Campaign summary: [2–3 sentences].
Write a slide-by-slide outline: Agenda, Context & Objectives, Audience Insight, Creative Strategy, Channel Plan, Timeline, Success Metrics, Next Steps, Q&A.
For each slide: title, 3 content bullets, one thing to avoid saying in the room.
26. Quarterly business review structure
Client: [Client name]. Quarter: [Q and year].
Performance summary: [paste executive summary from prompt 23]. Goals for next quarter: [draft goals].
Write a QBR outline. Format: Last quarter in 3 numbers, What worked and why, What didn't and why, Recommended direction, Proposed goals with rationale, Open discussion.
Max 8 slides.
27. Campaign results presentation narrative
Campaign: [campaign name for Client name].
Results: [paste one-pager from prompt 20].
Audience: [client stakeholders].
Turn the results one-pager into a 5-slide narrative. Each slide: a declarative headline (a claim, not a label) and 3 supporting bullets.
Tone: clear and direct. Avoid "results were mixed" framings.
28. New business pitch outline
Prospect: [Company] in [industry].
Our most relevant credentials: [2–3 specific results from comparable clients].
What we know about their challenge: [research findings or working hypothesis].
Write a pitch outline: Their world (the problem), Our approach (how we solve it), Proof (relevant results), Our team (brief), How we work (process), Proposed scope (placeholder), Next steps.
29. Strategy recommendation slide
Context: [what the client asked for / what we are recommending].
Recommendation: [our specific recommendation in one sentence].
Write a single slide. Format: bold headline (the recommendation itself), 3 supporting bullets (evidence → rationale → expected outcome for this client), 1 risk to acknowledge.
No jargon. Write as if presenting to a skeptical senior stakeholder.
30. Post-campaign debrief summary
Project: [name for Client name].
What went well: [list]. What didn't work: [list]. What we'd do differently: [list].
Write a post-campaign debrief document.
Format: Project overview, Wins (3 bullets with specifics), Challenges (3 bullets with root cause), Lessons learned (3 bullets), Recommendations for the next campaign.
Tone: honest and constructive. No marketing language.
How do you get consistent results across the account team?
Consistency requires three things from any agency running these prompts at scale: a shared home for the templates, a client voice brief attached to each client, and a practice of improving the prompt when a better version surfaces.
- Save every prompt with variables intact. A prompt that still reads
[Client name]is the correct state for storage. When a team member picks it up, they fill the variables for their specific context and run it — they never rewrite the template itself. - Attach a voice brief per client. For each client, write a short block: 3–5 examples of approved past copy, 4–6 brand voice adjectives, and 2–3 phrases the brand would never use. Store this alongside every client's prompts. The account manager pastes it in once; it anchors everything else.
- Update the prompt when a version produces a better first draft. If prompt 7 produces a brief that needs only minor editing, note what made it work and capture that improvement in the saved template. The prompt you refine today should be better than the one you started with six months ago.
"The prompt library lets me save and reuse structured prompts by category, which saves real time on recurring client work. The Chrome extension sits inside Claude, ChatGPT, and Gemini — no tab switching." — Sumo-ling, Verified AppSumo review
How Prompt Architects fits this workflow
All 30 prompts above work inside any AI tool your team already uses. What Prompt Architects adds is the infrastructure that turns one-off prompts into a compound agency asset: a shared prompt library where account teams save templates with [bracketed variables] intact, Global Variables that inject client voice briefs automatically, and a Chrome extension that surfaces any saved prompt in one click inside ChatGPT, Claude, or Gemini.
In our July 2026 customer data, half of the agencies who joined Prompt Architects had no shared prompt system before signing up — not Notion, not a shared doc, nothing (our customer data, July 2026). For those teams, the first-order value is having one place where the creative brief template, the reporting narrative prompt, and the kickoff presentation outline all live together, accessible to every account manager without digging through Slack or email.
The Teams feature lets an agency build a library that every account manager accesses from the same interface, with client-specific variables flowing into every prompt automatically. Setup that pays back fastest: one saved prompt per deliverable type, one voice brief per client, and the library does the re-briefing work on every subsequent project.
Prompt Architects is free to start, no credit card required.
Save the five prompts that match your most common deliverables this week, attach your client voice briefs, and the next brief or report you write will take a fraction of the time it does today.
Start free — build your agency prompt library inside ChatGPT, Claude, or Gemini →