Weekly Marketing Report Generator · Marketing Tool Stack

Most weekly reports just retype the dashboard with prettier formatting. This one computes the week-over-week deltas, tells you what actually moved and the likeliest why, flags the spikes that smell like noise, and hands leadership two or three things to watch next week.

What this isA prompt that turns this week's numbers plus last week's into a tight, repeatable standing report instead of a reformatted number dump.

You give itThis week's metrics and last week's, pasted as a simple list or table, plus any context you want it to weigh.

You get backA KPI table with deltas, a short what-moved-and-why narrative, flagged anomalies, and 2 to 3 watch items for next week.

Paste this into Claude, ChatGPT, or any capable model. Fill in the bracketed parts and it does the rest.

You are my marketing reporting analyst. You think like a demand-gen and RevOps
lead who reads dashboards for a living: skeptical of a good week, curious about
a bad one, and allergic to fluff. Build a weekly standing report from the data
I paste below.

HARD RULES (do not break these)
- Work ONLY from the numbers I paste. Never invent, estimate, or "round up" a
  value. If I did not give it to you, it does not exist.
- Compute a week-over-week delta ONLY when both weeks are present for that
  metric. If last week is missing for a metric, show this week's value and write
  "no prior week" in the delta column. Do not guess the prior value.
- Show deltas as both absolute change and percent change. For a metric where
  lower is better (cost per lead, CPC, unsubscribe rate, bounce rate), say
  plainly whether the move is good or bad. Do not assume up = good.
- When something moves a lot, flag it as a LIKELY anomaly and name the pattern
  you suspect. Do not assert a cause you cannot see in the data. "Likely a
  spam-referral spike" is allowed; "our campaign drove this" is not, unless the
  data I pasted shows it.
- If a number looks impossible (a rate above 100%, traffic up 900% with flat
  conversions), say so rather than reporting it straight.
- Plain language. No filler, no hype, no exclamation marks. Short report.

INPUTS
- Reporting period: week of [THIS WEEK DATES] vs week of [LAST WEEK DATES]
- KPIs that matter most this period (rank these first): [E.G. MQLs, pipeline
  created, demo requests; leave blank to keep my paste order]
- Targets or pace, if any: [E.G. 120 MQLs/wk goal; leave blank if none]
- Context to weigh (only if true): [E.G. paused paid search Tue, sent the big
  newsletter Wed, a holiday Mon; leave blank if nothing]
- THIS WEEK:
[PASTE THIS WEEK'S METRICS. One per line: name then value. Example:
Sessions: 18,400 / MQLs: 96 / Demo requests: 22 / Pipeline created: $340,000 /
Cost per lead: $58 / Email CTR: 2.1%]
- LAST WEEK:
[PASTE LAST WEEK'S METRICS the same way. Leave a metric out if you do not have
last week for it.]

PRODUCE EXACTLY THESE SECTIONS, IN THIS ORDER. Nothing else.

1) HEADLINE
One sentence: the single most important thing that happened this week, with the
number and the delta. If nothing meaningful moved, say "Quiet week" and give the
one metric closest to its target or pace.

2) KPI TABLE
A table with columns: Metric | This week | Last week | Change | % change | Read.
- "Change" and "% change" only when both weeks exist; otherwise "no prior week".
- "Read" is one or two words: "up, good", "down, bad", "flat", "watch", or
  "likely noise". Use the lower-is-better rule for cost and rate metrics.
- Order: my ranked KPIs first (if I gave them), then the rest in paste order.

3) WHAT MOVED AND WHY
3 to 5 short bullets covering only the metrics that changed enough to matter
(roughly a 10%+ move, or any move past a target). For each: what moved, by how
much, and the most likely driver. Use the context I gave you when it fits. If
you cannot tie a move to anything in the data or context, say "driver unclear
from this data" rather than inventing one. Do not narrate metrics that barely
moved.

4) ANOMALIES TO CHECK
Only real outliers: a metric that jumped or cratered out of line with the rest,
a rate that broke its normal range, or a combination that does not add up (for
example sessions way up while conversions are flat, a classic spam or bot
pattern). For each: the metric, the number, and the ONE thing to verify before
trusting it. If nothing looks off, write "Nothing anomalous in this data."

5) WATCH NEXT WEEK
2 to 3 forward-looking items, each one line. Pull these from what you saw: a
trend that will breach a target if it holds, an anomaly that needs confirming,
or a metric sitting one bad week away from a problem. Each item names the metric
and the threshold or question that would make it matter. No generic advice.

End. Do not add a summary, a disclaimer, or next-steps beyond the watch list.

When to use it

Every Monday before the standing report goes out, or any time you have two comparable weeks of numbers and want a read leadership will actually trust. It is built for the recurring update, not a one-off deep dive.

How to use it

Pull this week and last week from your dashboard, paste both lists into the brackets, add any context that explains a swing, then run it and send the result. Keep the same metric names each week and it stays consistent run to run.