Prompt Chain: Full Research Report Pipeline

What This Builds

A structured multi-step prompt chain that takes you from raw findings to a complete report narrative in one Claude conversation session — generating the executive summary, section introductions, key callout boxes, implications, and appendix text as a coordinated pipeline rather than ad hoc prompts. The result is a report that reads like it was written by one person with a single point of view, not assembled from disconnected AI outputs.

Prerequisites

• {{tool:Claude.plan}} subscription at {{tool:Claude.price}} — needed for the extended context window to hold a full report's worth of content
• Your study's key findings compiled (bullet points organized by section)
• The client's business question and audience defined
• Time to build: First time takes 1–2 hours to develop your own chain template; after that, 30 minutes per report
• Cost: {{tool:Claude.price}}

The Concept

A prompt chain is a sequence of prompts where each output feeds into the next. Think of it like an assembly line: the first prompt sets the context and tone, the second drafts the executive summary (establishing the narrative spine), the third drafts each section introduction (following that spine), the fourth writes callout boxes that are consistent with the section narratives, and the fifth generates implications that flow logically from everything above. At no point do you have to re-explain the client situation — it's all in the chain.

Build It Step by Step

Part 1: Chain Prompt 1 — Context Setting

Start every report session with this identical opening. Copy-paste it as your first message in Claude:

I'm building a research report. Here is the full context:

CLIENT: [name or type]
INDUSTRY: [industry]
BUSINESS QUESTION: [what they hired us to find out]
AUDIENCE: [who reads the report — role and sophistication level]
REPORT TONE: [choose: strategic/direct/accessible/formal]
KEY DECISION: [what decision this data is supposed to inform]

Study overview:
- Study type: [survey / qual / mixed methods]
- Sample: [n=X, target audience]
- Fieldwork dates: [dates]

I will share findings by section in the following messages. Please confirm you have the context and are ready to help build the report narrative.

What you should see: Claude confirming it understands the context and asking for the findings.

Part 2: Chain Prompt 2 — Executive Summary First

Before writing any section introductions, write the executive summary. This establishes the narrative spine that everything else will follow.

Here are all the key findings from the study. First, before drafting anything, identify what you think is the single most important finding — the headline. Then draft the executive summary:

[Paste all findings, organized by section]

Executive Summary instructions:
- 280–320 words
- Lead with the headline finding in the first sentence
- 3 supporting paragraphs of context
- Close with 3 implications formatted as: "[Action verb] [recommendation]"
- Confident, assertive language — never hedge ("may suggest," "could indicate")

What you should see: A complete executive summary with a clear headline finding up front and 3 action-oriented implications.

Part 3: Chain Prompt 3 — Section Introductions

Now draft introductions for each section. Do all of them in one prompt:

Using the narrative spine from the executive summary above, write brief section introductions (90–110 words each) for these sections. Each introduction should frame the section's findings in light of the headline finding and client question:

Section 1: [Section name] — Findings: [bullet list]
Section 2: [Section name] — Findings: [bullet list]
Section 3: [Section name] — Findings: [bullet list]
[Continue for each section]

Each introduction must: (1) state what this section covers, (2) note the most important finding in this section, and (3) hint at what it means before the data is shown.

What you should see: A set of section introductions that feel connected to each other and to the executive summary.

Part 4: Chain Prompt 4 — Callout Boxes

Generate callout boxes for the most important findings:

From everything we've drafted so far, identify the 6 findings that most deserve a prominent "Key Insight" callout box on a slide. For each one, write:

1. The finding (1 sentence, with specific numbers)
2. A callout box version (2 sentences, assertive, written for a skimming executive who might only read these 6 boxes)

Do not repeat findings already covered in the executive summary — these should be secondary insights that add texture.

What you should see: 6 callout boxes that work as standalone insights without requiring the surrounding slide content.

Part 5: Chain Prompt 5 — Implications Slide

The final chain prompt generates the implications slide — the strategic payoff of the whole report:

Based on everything in this report — the executive summary, section findings, and callout boxes — write the Implications slide. This is the last slide of the report and the most important one.

Format: 5 implications, each as:
- Bold headline (5–7 words, action-oriented)
- 2 supporting sentences that link back to specific findings

Implications must be:
- Genuinely actionable (what should the client DO, not just know?)
- Directly traceable to findings in this report
- Written for [client audience role]
- Escalating in ambition: start with immediate actions, build toward strategic shifts

What you should see: 5 implications that feel earned by the data, not generic recommendations.

Real Example: Consumer Segmentation Study

Study: A 1,000-respondent segmentation study for a national insurance brand. 4 segments identified. Client is VP of Marketing.

Chain Prompt 1 (Context): Client: Regional insurance carrier. Question: Which consumer segments should we prioritize for growth? Audience: VP of Marketing, data-literate. Decision: Budget allocation across customer acquisition channels.

Chain Prompt 2 (Executive Summary): After pasting all segment findings, Claude identifies that Segment 2 ("The Planners") is the highest-value, most-acquirable group and leads the executive summary with this headline: "The insurance brand's clearest growth opportunity is Planners — a 28% segment that overindexes on digital research, pays above-average premiums, and shows 2x the likelihood to switch providers."

Chain Prompt 3 (Section Intros): Each of the 4 segment introductions is now framed relative to the Planners headline, making the deck feel unified.

Chain Prompt 4 (Callouts): 6 callout boxes highlight the most striking segment statistics — the ones that stop a reader mid-scroll.

Chain Prompt 5 (Implications): 5 implications tell the VP of Marketing exactly which channels to invest in and which segment to prioritize for next quarter's campaign.

Total time: 40 minutes of Claude work + 30 minutes of review and edits.

What to Do When It Breaks

• Implications are too generic → Add to the prompt: "Each implication must reference a specific finding by name. Avoid generic advice that would apply to any brand."
• Section introductions don't feel connected → After getting them, ask: "Rewrite section intro 3 so it references the theme established in the executive summary — that [headline finding]."
• Claude loses context mid-chain → If you're hitting context limits, summarize earlier chain outputs: "Here's a summary of what we've drafted so far: [paste exec summary]. Now draft section introductions."
• Output tone shifts between chain steps → Start each prompt with "Maintaining the same [strategic/accessible/formal] tone as the executive summary we drafted..."

Variations

• Simpler version: Run just Chain Prompts 1 and 2 — the context + executive summary. The rest you write yourself. Even this 2-prompt version saves significant time.
• Extended version: Add a Chain Prompt 0 before context-setting: ask Claude to review your raw crosstab data and identify the most statistically significant and practically meaningful findings before you organize them into sections.

What to Do Next

• This week: Build your own Chain Prompt 1 template with your firm's standard context fields and save it as a text file you can open for every new project
• This month: Run the full 5-prompt chain on an upcoming project and track time savings vs. your normal process
• Advanced: Combine this chain with a Claude Project (see Level 4 guide on Claude Projects) so the chain runs inside a Project that already knows your client — eliminating even Chain Prompt 1

Advanced guide for Research Analyst professionals. These techniques use more sophisticated AI features that may require paid subscriptions.