thinkhard.SI

Prompt repository Research & learning

Benchmark your numbers against figures you collected from peers

You have gathered figures from peer organizations and need a fair comparison with your own, with definitions lined up and adjustments shown.

Registry no.
RES-047
Collection
Research & learning
Laid out for
GPT
Score
10 out of 10
Blanks to fill
5
Added

Research & learningGPT

The prompt

Score 10 of 10, Excellent

Use it when You have gathered figures from peer organizations and need a fair comparison with your own, with definitions lined up and adjustments shown.

  • {{audience}}
  • {{use_case}}
  • {{min_peers}}
  • {{our_figures}}
  • {{peer_figures}}
Full prompt and why it works
Prompt
# Role
You are a benchmarking analyst. You compare an organization's figures with peer figures it has collected, make the comparison fair before drawing conclusions, and show every adjustment so others can check it.

# Goal
Benchmark our figures against the peer figures below, for {{audience}}, so that they can see where we lead and where we lag on comparable terms, and how much weight each comparison can carry, before using it in {{use_case}}. A good benchmark is one where every gap survives a check of definitions and periods.

# Steps
1. Definitions: for each metric, compare our definition with each peer source's definition, period and unit, and note whether higher or lower is better.
2. Adjust only where the inputs allow it, for example turning a quarterly figure into an annual one, or restating a figure per employee using a headcount we were given. Show each adjustment as a formula.
3. Compare our figure with the peer median and range for each comparable metric. Use the median rather than the mean, because one or two peers can skew an average.
4. Position: call us "in line" when our figure sits within the middle half of the peer figures, and "lead" or "lag" only outside it, according to which direction is better.

# Output
Use Markdown headings:
1. Definitions: a table with the columns Metric, Our definition, Peer definitions, Better if, Comparable (yes, after adjustment, no), Adjustment.
2. Benchmark: a table with the columns Metric, Our figure, Peer median, Peer range, Number of peers, Our position.
3. Where we lead and where we lag: up to six bullets, each with the size of the gap and a possible explanation, labeled as a hypothesis.
4. Caveats: few peers, old figures, self-reported data, and anything else that limits the comparison.
Keep the written parts under 500 words.

# Constraints
- Use only the figures below. Do not add industry averages or peer figures from memory.
- Do not compare metrics marked not comparable; list them with what would make them comparable.
- Do not name a peer that the source lists anonymously.

# Stop rules
If a metric has fewer than {{min_peers}} peer figures, show it but mark it "too few peers to judge". If a peer source's definition or period is missing, treat that comparison as not comparable and list what is missing. If the inputs do not say whether higher or lower is better for a metric, write [CHECK: direction] rather than assuming. For example, a caveat in the right form: "Peer figures for staff turnover are from a year earlier than ours." That example shows the form only.

# Data
Our figures, with definitions and period:
"""
{{our_figures}}
"""

Peer figures, with source, definitions and period:
"""
{{peer_figures}}
"""

Variables

{{audience}}
Who reads the benchmark, for example "our operations committee".
{{use_case}}
What it is for, for example "setting next year's targets" or "a board presentation".
{{min_peers}}
The fewest peer figures you will judge a metric on, for example 4.
{{our_figures}}
Your figures, each with its definition, unit and period.
{{peer_figures}}
The peer figures you collected, each with its source, definition, unit and period, anonymized if the source is.

Why it works

  • What good looks like. Definitions and periods are lined up before any comparison, and "lead" or "lag" is used only outside the middle half of the peers, so a gap is not a matter of definitions.
  • What to avoid. Only the figures you collected are used, with no industry averages from memory and no names for peers the source keeps anonymous.
  • If unsure. Metrics with too few peers are marked, missing definitions make a comparison not comparable, and an unknown direction becomes [CHECK: direction].

How to use it

  1. Press Fill in blanks and type your details into each blank. What you type stays in this browser.
  2. Copy the filled prompt into the AI tool you already use, such as ChatGPT, Claude or Gemini, or open it in the Grader or the Prompt Tester here.
  3. Before you rely on it, try it on three or four real examples. The Prompt Tester runs them all at once and shows which ones go wrong.
Use it from code

Get this prompt as JSON from the free REST API, with your own text in the blanks, or let an AI assistant fetch it through the MCP server. No key.

curl -s https://thinkhard.si/api/v1/prompts/benchmark-against-peers

With the MCP server added to Claude, ChatGPT, Cursor or VS Code, call get_prompt with {"prompt": "benchmark-against-peers"}, or just ask for it by name.