agentleFS
scumunnascumunna/programmatic-skills/skills/client-deliverable-templates/SKILL.md

client-deliverable-templates

Skill1 starsChanged 5 months agoMITView on GitHub

Fill-in templates for client-facing programmatic deliverables, with the honesty rules baked in. Use when the user asks for a "media plan template", a "QBR deck" or "business review", a "client proposal", how to "explain this to a client" or put it in "plain English", a "client glossary", how to "deliver bad news to a client", or a "client report template". Provides a media plan, a QBR slide spine, a proposal outline, a plain-language glossary, and a bad-news framing pattern.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 5 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use these when the deliverable goes to a client.

    line 14
  2. Distinctive:

    No assist presented as a conversion: Keep post-click and post-view separate.

    line 78
  3. Distinctive:

    No correlation stated as a cause: If two things moved together, say so.

    line 82
  4. Distinctive:

    If you cannot source it, it does not go in.

    line 86
  5. Very specific:

    Plain language: Write for a non-technical reader.

    line 88
  6. Distinctive:

    No em dashes in client text: Use periods, commas, parentheses, or "and" and "but".

    line 90
  7. Distinctive:

    Calling an assist a conversion: Keep post-click and post-view separate, and lead with incrementality when a budget decision is at stake.

    line 135

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Client deliverable templates

Repeatable, fill-in templates for the things a client-facing programmatic team writes again and again: the media plan, the quarterly business review, the proposal, the glossary that translates jargon, and the message that delivers bad news. The templates carry the structure and the honesty rules so every deliverable is clear, sourced, and trustworthy by default, whatever DSP the work ran on.

Use these when the deliverable goes to a client. For the analysis that produces the numbers, see the reporting-by-campaign-goal skill and hand off to the reporting-analyst agent. To shape the narrative voice, use the client-communications-lead agent. Before anything ships, route it through the qa-scrutinizer agent, which checks for exactly the failures this skill is built to prevent.

When to use this skill

  • "Give me a media plan template" or "build a media plan for this campaign."
  • "I need a QBR deck" or "structure a quarterly business review."
  • "Write a client proposal" or "put together a pitch for this account."
  • "Explain this to a client" or "say this in plain English" or "drop the jargon."
  • "I need a client glossary" or "what does [term] mean for a client."
  • "How do I deliver bad news to a client" or "the campaign missed, how do I tell them."
  • "Give me a client report template."

Boundary: this skill provides the deliverable shell and the honesty bar. It does not decide which KPI belongs to which goal (that is reporting-by-campaign-goal), compute the metrics (that is programmatic-foundations), or produce the verified numbers and analysis (that is the reporting-analyst agent). Fill these templates with numbers that came from those, never with numbers invented to fit the template.

Quick reference

The user needsUse this templateWhat it gives them
A plan before a campaign runsassets/media-plan-template.mdObjective, KPI and targets, audience, channel mix, flighting, budget by funnel stage, measurement, assumptions, and a labeled forecast
To present a quarter to a clientassets/qbr-deck-outline.mdA fixed slide spine: cover, three-number summary, performance vs goals, what worked, what did not and what changed, market context, learnings, next-quarter plan and budget, appendix
To win new workassets/proposal-outline.mdProblem, approach, why us, scope, investment, outcomes as ranges, next steps
To translate a term for a clientassets/client-glossary.mdEach common term mapped to one plain client sentence
To deliver a miss or a problemassets/bad-news-framing.mdThe situation, what we saw, what we are doing, what we need from you, with a worked example

Pick the template by the deliverable, fill every bracket, and apply the honesty rules below to every line before it leaves the building.

Core process

  1. Confirm the numbers are real and sourced before you touch a template. Pull verified figures and the analysis from the reporting-analyst agent and the reporting-by-campaign-goal skill. A template makes a story look finished, which is exactly why it must never be filled with a number you cannot trace to a report.
  2. Pick the template that matches the deliverable from the quick-reference table, and keep its structure. The order is the value: a QBR always leads with the result, a bad-news note always leads with the situation and ends with the ask. Do not reorder the spine.
  3. Fill every bracket in plain language. Replace each placeholder with the client's actual objective, audience, and numbers. Swap any jargon for the plain sentence in assets/client-glossary.md, or paste those rows into an appendix. A blank field reads as an oversight, so write "to be confirmed" rather than leaving it empty.
  4. Apply the honesty rules to every line (see the next section). This is the step the qa-scrutinizer agent will check, so do it before, not after.
  5. State any forward-looking number as a range and label it an estimate. Forecasts and expected outcomes go in the clearly marked sections of the media plan and the proposal, given as conservative, expected, and stretch, with the assumptions shown. Never present a forecast as a commitment.
  6. Route the draft through review. Shape the voice with the client-communications-lead agent and send it to the qa-scrutinizer agent as the gate before it reaches the client.

The honesty rules (what qa-scrutinizer checks)

These are non-negotiable and apply to every template in this skill. The qa-scrutinizer agent checks client communication for precisely these, so build them in from the start.

  • No overpromising. State targets as targets and forecasts as estimates with ranges and assumptions, never as guarantees. Do not commit to a number the evidence does not support.
  • No assist presented as a conversion. Keep post-click and post-view separate. A view-through assist is not a conversion, and last-click credit is not proof of incremental impact. When a budget decision rides on it, lead with incrementality, not last-click. See reporting-by-campaign-goal and the platform measurement skill.
  • No correlation stated as a cause. If two things moved together, say so. Do not claim one caused the other unless an experiment or a clean attribution path supports it. If the cause is uncertain, say it is uncertain and say what you are doing to find it.
  • Every number is sourced. Each figure in a deliverable traces to a named report or study. If you cannot source it, it does not go in. This includes the numbers in a forecast, which trace to the stated assumptions.
  • Plain language. Write for a non-technical reader. Translate every term with assets/client-glossary.md. No internal jargon without a plain-language gloss.
  • No em dashes in client text. Use periods, commas, parentheses, or "and" and "but". This holds across every template and every filled-in deliverable.

Reference material

The five templates live in assets/ and are meant to be copied and filled, not read as theory:

  • assets/media-plan-template.md: the pre-flight plan. Objective, KPI and targets, audience strategy, channel and platform mix, flight and flighting calendar, budget split by funnel stage, measurement plan, assumptions, and a forecast section that is explicitly labeled an estimate. Use it when planning a campaign for a client.
  • assets/qbr-deck-outline.md: the quarterly business review slide spine. A fixed order from cover through executive summary (three numbers), performance versus goals, what worked, what did not and what we changed, market and competitive context, learnings, next-quarter plan and budget, and an appendix. Use it to present a quarter.
  • assets/proposal-outline.md: the new-business proposal. Problem, approach, why us, scope, investment, expected outcomes stated as ranges, and next steps. Use it to pitch or scope an engagement.
  • assets/client-glossary.md: the two-column plain-language glossary, each jargon term mapped to one client sentence. The client-facing counterpart to the practitioner glossary in programmatic-foundations. Use it to translate any term in any deliverable.
  • assets/bad-news-framing.md: the situation, what we saw, what we are doing, and what we need from you pattern, with a worked example. Use it for any miss, risk, or breakage.

Templates and examples

A worked example ships inside assets/bad-news-framing.md: a retargeting campaign whose recorded cost per order blew past target because a site change broke conversion tracking for several days. It shows the four-part pattern in full, leads with the miss in the first sentence, sources every number, separates the tracking break from the real performance, states the expected recovery as a range, and ends with a concrete ask, without apologizing three times or blaming the client.

For a media plan, the forecast table is filled as conservative, expected, and stretch columns for each outcome (reach, primary KPI, and efficiency), under a heading that says in plain words that these are estimates, not guarantees. For a QBR, the executive summary is exactly three numbers, each shown against its target with one sentence of so-what.

Common pitfalls

  • Filling a template with numbers that are not sourced. The polish of a template hides the gap. Pull every figure from the reporting-analyst agent and name its source.
  • Presenting a forecast as a promise. A forecast is an estimate on stated assumptions. Give it as a range and label it, or a client will hold you to the midpoint.
  • Calling an assist a conversion. Keep post-click and post-view separate, and lead with incrementality when a budget decision is at stake.
  • Stating a correlation as a cause. Two metrics moving together is not proof. Say "we saw" unless an experiment supports "we caused".
  • Leaving jargon in client text. Translate it with assets/client-glossary.md. If the client has to ask what a word means, the deliverable failed at its one job.
  • Reordering the QBR spine or the bad-news pattern. The order is the message. Result first in a QBR, situation first and ask last in bad news.
  • Skipping the review gate. Shape the voice with the client-communications-lead agent and pass every client-facing draft through the qa-scrutinizer agent before it ships.

Sources

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.