Skip to content

Decides

Lead value math

Sets the most you can pay for a lead and a click from your own deal size, win rates and payback, and the budget a campaign needs to learn.

Other tools

Add this skill to Claude

Claude takes skills as a ZIP file. Your download has started; if it didn't, use the button below.

  1. 1In Claude, open Customize, then Skills.
  2. 2Choose +, then Upload a skill, and pick lead-value-math.zip.
  3. 3Ask for it by name, or let Claude use it when a task matches.

Skills need code execution turned on in Claude's settings. On a Team or Enterprise plan, an admin can share it with everyone.

Download the folder

Try it now, without installing

See the skill work on a made-up input before you add it. Nothing is installed, and you can swap in your own input.

Open in the Claude app

Opens a new chat in the Claude desktop app with the prompt filled in. Nothing is sent until you press enter.

Includes the whole SKILL.md and the sample input. Paste it into Claude, ChatGPT or any chat.

Influenced by
About Smart Bidding, on reading periods of 30 conversions, Budget and bid strategies, on the 30-conversion floor
Needs
Average contract value (ACV) · Win rates, lead to opportunity and opportunity to closed-won · The spend rule your P&L owner holds (a share of ACV, or payback months with gross margin)
Works with
Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md
Version
1.0.0 · Last checked · License MIT
Maintained by
Sabih AhmedPaid media
SkillPaid media

Lead value math

Example output

Set the ceilings for a channel
FigureValueCall
Max cost per lead$285Ceiling
Cost per lead today$200Under it
Campaigns the budget carries2No more

Sets the most you can pay for a lead and a click from your own deal size, win rates and payback, and the budget a campaign needs to learn.

The skill

The whole skill, in plain text.

Nothing is hidden. Read it, change it to fit your team, or take the whole folder.

Scripts
Optional Python scripts for the counting. It works without them and says which way it ran.
Network
Doesn't need the internet.

3 test cases

lead-value-math/SKILL.md

In the folder

  • lead-value-math/SKILL.md
  • lead-value-math/scripts/lead_math.py
---name: lead-value-mathdescription: >-  Sets the most a B2B SaaS company can pay for a customer, a lead, an opportunity and a click, from its own deal size,  win rates and spend rule (CAC share of ACV or payback months), then sets the smallest monthly budget a campaign  needs to learn and how many campaigns a budget can carry. Returns HEADROOM, OVER CEILING, UNDERFUNDED, CEILING SET,  NEEDS DECISION or BLOCKED with the rule that fired. Use it before setting bid targets, sizing a paid channel or  judging cost per lead. Not for sorting search terms (use search-terms-triage) or splitting budget across channels  (use channel-fit). Never uses a benchmark ratio or a borrowed cost per lead.license: MITcompatibility: >-  Works in Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md. Uses a connected CRM or ad account if the  agent has one, otherwise exports or pasted figures. Python 3 is optional, for the math.metadata:  author: SHAKEUP  version: "1.0.0"  stage: decide  plugin: paid-media  homepage: https://www.shakeuphq.com/skills/lead-value-math--- # Lead value math Turns deal size, win rates and the business's spend rule into the ceilings every paid campaign is held to. ## Purpose - One job: compute allowable CAC, max cost per lead (max CPL), max cost per opportunity, max CPC, the campaign floor and the judgment window, and say whether the account sits under its ceilings.- Stage: decide. It ends in a verdict.- It never uses a benchmark ratio, an industry cost per lead or a "typical" win rate. Every input is the reader's.- It never changes a bid, a target or a budget. It recommends; the reader applies. ## Use when / not | Situation | Use || --- | --- || Setting or checking a Target CPA, a cost cap or a max CPC | this skill || Asking "what can we afford per lead?" on any paid channel | this skill || Deciding how many campaigns a budget can carry | this skill || Sorting search terms into keep, negative or new ad group | `search-terms-triage`, which reads this skill's `max_cpl` || Choosing which channels to fund at all | `channel-fit` || Planning LinkedIn layers and formats | `linkedin-plan`, after this skill || Reporting pipeline per channel | `pipeline-report` | ## Inputs | Input | Required | Ways to provide, in order | If missing || --- | --- | --- | --- || ACV (average first-year contract value) | yes | `## Company` "Average deal size (ACV)"; a CRM closed-won report; the reader | ask once; then `BLOCKED` || Win rates: lead-to-won, or lead-to-opportunity and opportunity-to-won | yes | `## Targets`; a CRM report of paid leads over the trailing 12 months; the reader | ask once; then `BLOCKED` || Spend rule: CAC share of ACV, or payback months plus gross margin | yes (one) | `## Targets` lines "Target CAC share of ACV", "Target CAC payback", "Gross margin" if present; the P&L owner's answer | ask once, naming the P&L owner; then `BLOCKED`. Never default || Media share of CAC | no | the reader | if the spend rule is fully loaded CAC (salaries, tools), ask; otherwise treat it as media CAC and say so || Platform cost data: spend, clicks, leads (or CPC and conversion rate), with the period | no | a connected ad account; an export; pasted figures | ceilings only; verdict `CEILING SET` || Monthly non-brand budget for the channel | no | `## Channels`; the reader | floor computed, campaign count `not available` || Conversions a campaign needs a month to learn | no | Google Ads and Microsoft Ads: 30 (see step 6); any other platform: the reader or that platform's own documentation | floor `not available` for that platform || Sales cycle in days | no | `## Company` "Sales cycle" | judgment window `not available` || Are outcomes validated? (CRM stages reach the platform, or sales reviews leads monthly) | no | `## Accounts and data`; the reader | assume not validated and say so || Zero-conversion spend from a search terms run | no | `search-terms-triage` handoff `waste` | skip the recovered-pipeline estimate | ## Profile read From `company-profile.md`: `## Company` (ACV, sales cycle, pricing, business model), `## Targets` (target cost per lead, lead to opportunity rate, opportunity to closed-won rate, pipeline target, and any spend-rule lines), `## Channels` (monthly budgets), `## Accounts and data` (CRM, connected ad accounts), and every `## Learnings` entry for `lead-value-math` with status `noted`, `fact` or `accepted`. No profile: ask in one message for ACV, the 2 win rates (or lead-to-won), and the spend rule the P&L owner holds (share of ACV, or payback months and gross margin). Say that `company-profile` saves this next time. ## Steps 1. **Apply learnings.** A `fact` (a new win rate, a new spend rule) replaces the profile value. An `accepted` proposal changes a rule below. Name each one that fires. 2. **Normalize the inputs.** Same currency, rates as decimals, ACV as first-year value.    | Check | Rule |   | --- | --- |   | Only monthly price known | ACV = monthly price x 12 only after the reader confirms; label it `estimate` |   | Win rates for all sources, not paid | Use them, label `blended rate`, coverage `partial` (paid leads usually close at a different rate) |   | Rates from less than one full sales cycle of closed deals | Label `immature cohort`: late deals haven't closed yet, so the rate reads low |   | A rate above 100% or at 0 | Reject it and ask again |   | Cost data period | Spend, clicks and leads from the same campaigns and the same dates | 3. **Allowable CAC.**    | Route | Formula | Use when |   | --- | --- | --- |   | Share | ACV x CAC share | The P&L owner states a share of contract value |   | Payback | payback months x (ACV / 12) x gross margin | The P&L owner states a payback target |   | Both given | the lower of the 2 | Always: the stricter limit is the one the business holds the channel to. Show both |   | Fully loaded target | CAC x media share | The target covers salaries and tools, not only media | 4. **Ceilings.**   - Lead-to-won = lead-to-opportunity x opportunity-to-won, unless the reader gives lead-to-won directly.   - Max CPL = allowable CAC x lead-to-won.   - Max cost per opportunity = allowable CAC x opportunity-to-won.   - Operating target:      | Outcomes | Operating target | Why |     | --- | --- | --- |     | Validated (CRM stages reach the platform, or sales reviews leads monthly) | max CPL | The ceiling is measured on real buyers |     | Not validated | 60% to 70% of max CPL | The margin absorbs lead quality nobody has checked | 5. **Compare with what the account pays.**   - Actual CPL = spend / leads; without leads, CPC / conversion rate.   - Headroom = max CPL minus actual CPL, also as a share of max CPL.   - Max CPC = max CPL x conversion rate. Break-even conversion rate = CPC / max CPL.   - For a list of keywords or campaigns, score each row: it pays when CPC / its conversion rate is under max CPL. A row with no conversions is judged on spend against max CPL, not on a rate.   - Not validated and actual CPL above the operating band: recommend holding bids even when under the ceiling, because the headroom rests on unchecked leads. 6. **Campaign floor.** Floor = (conversions needed a month / conversion rate) x CPC. Campaigns the budget carries = non-brand budget / floor, rounded down.    | Platform | Conversions needed a month | Source |   | --- | --- | --- |   | Google Ads (smart bidding) | 30 | Google recommends judging smart bidding over periods with at least 30 conversions (About Smart Bidding) |   | Microsoft Ads (Max Conversions, Target CPA) | 30 | Microsoft says both stop optimizing below 30 conversions in any 30-day period (Budget and bid strategies, checked 2026-10-24) |   | Any other platform | the reader's figure, or the platform's documented one opened in this run | otherwise `not available` |    The conversion rate here is to the event bidding optimizes for, not to closed-won. Brand campaigns sit outside the count: separating brand protects measurement, not learning.   Budget below one floor: list the 4 moves in order. Merge into an adjacent campaign that shares the target; move the optimization event to a mid-funnel action with more volume; Maximize Conversions with no target; manual CPC. 7. **Judgment window.** 1.5 x the sales cycle in days, rounded up: the time before a campaign can be read on pipeline. Search-term relevance, conversion rate and CPL read within 2 weeks; cost per opportunity, deal size and win rate need the full window. 8. **Recovered value (optional).** With a waste figure: recovered leads = waste / actual CPL; recovered pipeline = recovered leads x lead-to-won x ACV. Label it `estimate`: it assumes the moved budget converts at today's rate. 9. **Run the math.** `python3 scripts/lead_math.py` with the inputs (flags in its docstring; `--validated` when outcomes are validated; `--platform google` or `microsoft` for the floor; `--rows file.csv` to score keywords). Without Python, apply steps 3 to 8 by hand, show each formula with its figures, and say you ran it by hand. ## Check and verdict | Verdict | Condition (first match wins) || --- | --- || `BLOCKED` | ACV, the win rates or a spend rule is missing after one question || `NEEDS DECISION` | The profile's "Target cost per lead" is above the computed max CPL: the business's own target is unaffordable by its own rule || `OVER CEILING` | Actual CPL, or actual cost per opportunity, is at or above its ceiling || `UNDERFUNDED` | The channel's non-brand budget is below one campaign floor || `CEILING SET` | Ceilings computed; no cost data to compare against || `HEADROOM` | Actual cost under the ceiling and the budget clears at least one floor | `OVER CEILING` lists the levers in order: the click price (negatives, qualifiers, match types), then the page (conversion rate), then lead quality (what you bid for). Raising the CAC share also raises the ceiling, but that is a finance decision, not a fix. Named stops: `BLOCKED` asks for the missing input once, naming who owns it. `NEEDS DECISION` shows the profile target beside the computed ceiling, with the inputs behind each, and asks which the P&L owner holds. No other stops. ## Learn Append to `## Learnings` in the profile when: - the P&L owner states or changes the spend rule: a `fact` (and propose adding it to `## Targets`);- a quarter's CRM data moves a win rate: a `fact` with the report and date;- after the judgment window, actual cost per opportunity comes back above or below the ceiling the plan set: a `fact`;- the reader overrides the verdict (for example keeps spending over the ceiling for a strategic account list): a `proposal` with their reason. Example: `- 2026-10-24 | lead-value-math | fact | Paid CAC may be up to 30% of first-year ACV | Why: CFO, budget review 2026-10-20 | Status: noted` ## Output ```Verdict: <HEADROOM | OVER CEILING | UNDERFUNDED | CEILING SET | NEEDS DECISION | BLOCKED> · Coverage: <complete | partial (what's weak) | blocked> · Math: <script | by hand>Rule that fired: <one line> | Figure | Value | Formula and inputs || Allowable CAC | ... | <route>, <inputs and their source> || Lead-to-won | ... | ... || Max CPL | ... | ... || Max cost per opportunity | ... | ... || Operating target | ... | <validated or not> || Actual CPL | ... | <period, source> || Max CPC / break-even rate | ... | ... || Campaign floor | ... | <conversions needed and its source> || Campaigns the budget carries | ... | ... || Judgment window | ... days | 1.5 x <sales cycle> |(every figure with its source, or "not available" and what would provide it) Rows: <keyword or campaign table with pays or loses, if given>What to do: <2 to 4 lines, levers in order; recommendations only>Learnings applied: <list, or none>``` Then the handoff block. ## Hand off ```handoffskill: lead-value-mathversion: 1.0.0verdict: HEADROOMcoverage: partial (outcomes not validated)for: <account or channel>outputs:  allowable_cac: 7200  max_cpl: 285.12  max_cost_per_opportunity: 1584  operating_target: 171 to 200  campaign_floor: 6000  campaigns_supported: 2  judgment_window_days: 93  currency: USDnext: [search-terms-triage, linkedin-plan, microsoft-ads-b2b]open: [validate outcomes before raising bids]``` `search-terms-triage` uses `max_cpl` as its spend threshold. `linkedin-plan`, `microsoft-ads-b2b`, `reddit-plan` and `chatgpt-ads-check` hold their plans to `max_cpl`. ## Failure modes | Problem | Do this || --- | --- || The reader asks for "a normal B2B ratio" or a benchmark CPL | Refuse the benchmark; ask who owns the P&L and what share or payback they hold. `BLOCKED` until answered || Platform conversions and CRM leads disagree by more than 15% | Use CRM leads for actual CPL and say so; the platform count is not describing real leads || Only one blended win rate exists | Use it, label it, coverage `partial`; propose a paid-only CRM report || The cost data mixes brand and non-brand | Ask for non-brand separately; brand converts at a different rate and hides the real CPL || The reader gives ACV as monthly recurring revenue | Ask; never multiply by 12 without confirmation || Conversion rate is 0 | Floor and max CPC are `not available`; judge spend against max CPL instead | ## Influenced by - SHAKEUP's B2B Google Ads playbook: allowable CAC from a spend rule the P&L owner states, max CPL from the lead-to-won rate, the campaign floor and the judgment window of 1.5 sales cycles.- Google's About Smart Bidding, for the 30-conversion reading period: https://support.google.com/google-ads/answer/7065882?hl=en- Microsoft Advertising's budget and bid strategies guide, for the 30-conversion floor on automated bidding: https://learn.microsoft.com/en-us/advertising/guides/budget-bid-strategies?view=bingads-13

How it thinks

Inputs in, a verdict out, and a stop only where you decide.

What we built

A borrowed cost per lead benchmark says nothing about your business, so every ceiling here starts from your own contract value, close rates and spend rule. We added the working margin for leads nobody has checked, and the campaign floor, because a budget split into starving campaigns never learns.

It reads your company profile if you have one, or asks for only what it needs. Then it works through its steps, checks its own work and ends in a verdict.

The steps

  1. 01

    Read your learnings first

    A new win rate or spend rule your team recorded replaces the profile value before any math runs.

  2. 02

    Set what a customer may cost

    Allowable CAC comes from your ACV and the share or payback your P&L owner holds. When both are given, the stricter one wins.

  3. 03

    Turn it into ceilings

    Your lead-to-won rate gives the most a lead can cost, and your close rate gives the most an opportunity can cost.

  4. 04

    Leave a margin when leads aren't checked

    If sales doesn't confirm lead quality, it sets the working target at 60% to 70% of the ceiling, so unchecked leads can't eat the headroom.

  5. 05

    Compare with what you pay today

    Spend divided by leads shows where the account sits, and each keyword or campaign gets a pass or fail line.

  6. 06

    Size the campaign floor

    It works out what one campaign must spend to reach the conversions its bidding learns from, and how many campaigns your budget can carry.

  7. 07

    Make the call

    It returns one verdict with the rule that fired, the levers in order and a handoff block the next skill reads.

What you need

  • Required

    Average contract value (ACV)

    Your company profile → A closed-won report from your CRM → Your answer

  • Required

    Win rates, lead to opportunity and opportunity to closed-won

    Your company profile → A CRM report of paid leads over the last 12 months → Your answer

  • Required

    The spend rule your P&L owner holds (a share of ACV, or payback months with gross margin)

    Your company profile → Your finance lead's answer

  • Optional

    Spend, clicks and leads for the channel

    A connected ad account → An export you upload → Figures you paste

  • Optional

    Monthly non-brand budget

    Your company profile → Your answer

What it can return

  • HEADROOM

    You pay less per lead than the ceiling, and the budget clears at least one campaign floor.

  • OVER CEILING

    A lead or an opportunity costs more than your own economics allow.

  • UNDERFUNDED

    The budget can't carry even one campaign at its learning floor.

  • CEILING SET

    The ceilings are ready, but there is no cost data to compare yet.

  • NEEDS DECISION

    The cost per lead target in your profile is above what your own spend rule allows.

  • BLOCKED

    The contract value, the win rates or the spend rule is missing.

Where it stops for you

  • When the spend rule, win rates or contract value is missing after one question
  • When your profile's target cost per lead is higher than the math allows

Example output

What you get back.

A run for a made-up company. Your output depends on your inputs.

Here's a run for a made-up B2B SaaS company that sells software on a $24,000 average contract. Finance allows 30% of first-year contract value per customer, and leads aren't yet checked against the CRM.

Verdict

HEADROOM

Coverage: partial (outcomes not validated)

FigureValueWhere it comes from
Allowable CAC$7,200$24,000 x 30%, the finance lead's rule
Lead-to-won3.96%18% x 22%, from the profile
Max cost per lead$285$7,200 x 3.96%
Max cost per opportunity$1,584$7,200 x 22%
Working target$171 to $20060% to 70% of the ceiling, since leads aren't checked yet
Cost per lead today$200$54,000 / 270 leads, last 90 days
Campaign floor$6,000 a month30 conversions / 6.25% x $12.50 a click
Campaigns the budget carries2$16,000 / $6,000, rounded down
Judgment window93 days1.5 x the 62-day sales cycle
What to do

hold bids where they are. The account sits under its ceiling but at the top of the working target, and that headroom rests on leads nobody has checked. Send CRM stages back to the platform first, then decide whether to bid harder.

The handoff block the next skill reads
skill: lead-value-math
version: 1.0.0
verdict: HEADROOM
coverage: partial (outcomes not validated)
for: Google Ads, non-brand
outputs:
  allowable_cac: 7200
  max_cpl: 285.12
  max_cost_per_opportunity: 1584
  operating_target: 171 to 200
  campaign_floor: 6000
  campaigns_supported: 2
  judgment_window_days: 93
  currency: USD
next: [search-terms-triage, linkedin-plan, microsoft-ads-b2b]
open: [validate outcomes before raising bids]

What good looks like

  • Every ceiling shows the formula and where each input came from
  • No benchmark ratio, win rate or cost per lead appears anywhere
  • A working margin when lead quality isn't checked yet
  • A campaign count that leaves every campaign above its learning floor
  • The levers named in order when the account is over its ceiling

Install

Use it in the tool you already have.

Pick your tool. Each tab has the shortest path we know of.

  1. 01
  2. 02
    In Claude, open Customize, then Skills.
  3. 03
    Choose +, then Upload a skill, and pick the ZIP. Code execution must be on in your settings.
  4. 04
    Ask for it by name, or let Claude use it when a task matches.

Questions

Before you add it.

Why won't it use an industry cost per lead?

Because 2 companies with the same contract value can afford ceilings several times apart, once their close rates and spend rules differ. A benchmark hides that. The skill asks for your own figures instead.

What if we don't have a CAC target?

It stops and asks who owns the P&L. A share of contract value or a payback target in months, with gross margin, is enough to start.

Does it work for LinkedIn and Microsoft Ads too?

Yes. The ceilings hold for every channel. Its campaign floor needs the conversions each platform learns from, which Google and Microsoft document. For other platforms it asks for that figure.

What does the script do?

It runs every formula, scores a list of keywords or campaigns against the ceiling and returns the verdict. It's optional. Without Python, the skill shows each formula with your figures instead.