The Prompt
You are a senior B2B demand gen leader writing the weekly paid media exec summary that goes to your VP of Marketing and is forwarded to the CMO. You know that exec summaries lose readers in the first sentence if they lead with methodology, and you know that data without a "so what" is just noise.
INPUTS
I will paste the week's paid media performance data below. Include: total spend, total leads, blended CPL, channel-level breakdown, and how each compares to target and to prior period.
{PASTE_PERFORMANCE_DATA_HERE}
{OPTIONAL_PASTE_CONTEXT_HERE}
(Examples: "Q2 in flight, pipeline coverage at 85% of plan," "New ABM motion launched two weeks ago," "Holiday week, traffic expected to be lower." Leave blank if no context to add.)
WHAT I NEED FROM YOU
Produce an exec summary in this exact order. Total length: 250-400 words.
1. The Headline (1-2 sentences)
The single most important thing the VP needs to know this week. Not "spend was up." Something like: "LinkedIn drove a 22% lift in MQLs at flat CPL — the strongest week of Q2 so far." Lead with the win or the problem, not the data.
2. Wins (top 3, bulleted)
Three things that worked this week. Each one: what happened (with the number), why it mattered, and whether it's a one-week event or a pattern.
3. Misses (top 3, bulleted)
Three things that didn't work. Each one: what happened (with the number), what we know about the cause, and what's being done about it. Do not hide a miss — execs trust honesty more than they trust spin.
4. Looking Ahead (2-4 sentences)
What's planned for the next week or two and what the team is watching. Frame as forward-looking, not retrospective.
JUDGMENT RULES
- Use plain English. "Click-through rate" is fine; "CTR" is fine the second time; "the eCPM hovered around our benchmark" is jargon. The VP shouldn't have to translate.
- Numbers belong in the wins and misses, not the headline. The headline is a story; the bullets carry the proof.
- Do not pad. A 3-bullet "wins" section with one weak bullet is worse than a 2-bullet section. Cut the weak one.
- Tone is direct, confident, and grounded. Not breathless. Not hedged. "LinkedIn outperformed Google this week" — not "LinkedIn appears to have outperformed Google in what may indicate an emerging trend."
- If the week was bad overall, lead with that. "This week was below plan" is a better headline than burying the issue in bullet 3.
- If you don't have enough data to call something a win or a miss, say so. "Insufficient signal yet on the new ABM motion — flagging for review at the 4-week mark" is better than fabricating a story.
OUTPUT FORMAT
Return as {OUTPUT_FORMAT}.
If "slack": a single message formatted for #demand-gen or #paid-media channels, with bold headers and clean spacing.
If "markdown": full structure with ## headers and bulleted wins/misses.
If "email": opens with a 1-line subject suggestion, then the body in plain text suitable for Outlook or Gmail.
Begin.
How to Use It
This prompt is designed for Claude (Sonnet or Opus) and a Monday workflow: paste last week's channel-level performance data, set the output format (slack/markdown/email), and let Claude write the first draft. In side-by-side testing, Claude produces sharper summaries than ChatGPT for exec audiences — Claude's tendency to open with the actual business story ("LinkedIn drove a 22% MQL lift at flat CPL") rather than methodology ("this week we analyzed spend across four channels") is a consistent difference. GPT-4 class models produce competent summaries but require more post-editing to strip the analyst-speak.
The lead headline discipline is what separates this prompt from a standard reporting template. The prompt explicitly instructs Claude not to lead with data — to lead with the story. "This week was below plan" is the correct headline for a bad week; burying that in bullet 3 while opening with impression volume is the kind of thing that makes VPs stop reading exec summaries. Claude follows this framing reliably; the instruction is in the judgment rules.
Three inputs: the performance data (channel breakdown with actuals vs. targets vs. prior period), an optional context field (where the quarter is, what motions launched, any known anomalies), and an output format selector. The output format matters because Slack, email, and a written doc all have different length and formatting norms — the prompt handles all three without rewriting. If your exec prefers Slack threads over email, change one word in the input and the entire output structure shifts.
Example Output
Example output coming soon — currently running this prompt against live data and will publish the redacted output once it's ready.
Variations
Two variations of this prompt are worth knowing.
Variation 1: Biweekly Deep-Dive Version
A longer-form exec summary format for biweekly pipeline reviews — includes a month-to-date section, channel-level trend commentary, and a strategic recommendation for the next 30 days. For when your VP wants more than a weekly snapshot.
[PROMPT GOES HERE]
Variation 2: Channel-Isolated Summary Version
Scoped to a single channel — useful when one channel had a significant week and deserves its own focused summary before the full cross-channel rollup. Common use case: LinkedIn launches a new campaign type or Google loses Brand keyword coverage mid-week.
[PROMPT GOES HERE]
Get one new prompt every Monday.
Plus the system behind it. Free. Built for in-house demand gen managers at B2B SaaS companies.
Subscribe free →Frequently Asked Questions
Does this work with ChatGPT or is Claude required?
Claude is the recommended choice for exec-audience summaries. The consistent differentiator in production is Claude's "lead with the story" execution — Claude reliably opens with the headline insight rather than a description of methodology. ChatGPT produces usable summaries but requires more editing to achieve the VP-readable tone this prompt targets. For one-off use, either works. For a weekly workflow where summary quality directly shapes executive perception of the team, Claude is the right call.
What data format should I paste in?
A table works best — channel rows, with columns for spend, leads, CPL, and comparison data (vs. target and vs. prior week). Pasting a Google Sheets selection, raw CSV, or a markdown table all work. The prompt also accepts narrative descriptions of performance if you're summarizing from memory, but structured data produces sharper summaries. Include whatever down-funnel metrics your team tracks (MQLs, SAOs, pipeline contribution) and the prompt will incorporate them into the wins and misses logic.
How long should the exec summary be?
The prompt targets 250–400 words. That's calibrated for VP reading behavior: long enough to tell the full story, short enough to read in 90 seconds. If your exec consistently pushes for shorter summaries, trim by cutting the weakest bullet from wins or misses before the meeting — don't shorten the headline or the forward-looking section, those carry the most information per word.
What does the {OUTPUT_FORMAT} variable do?
It controls the entire output structure. "slack" formats for a #demand-gen or #paid-media channel message with bold headers and clean spacing. "markdown" produces a structured document with ## headers for each section. "email" adds a one-line subject suggestion and formats the body for Outlook or Gmail. Replace the variable text with whichever you need before running the prompt.