name

wasted-spend

description

Quantify the TOTAL wasted spend in a Google Ads account and pinpoint the biggest leaks — money spent on search terms, keywords, ad groups, or campaigns that returned ~0 conversions or cost far more per conversion than target. Rolls the leaks up to one dollar number and a % of spend, then lists the top leaks by scope, each with the dollars wasted and a recommended action. Use this skill whenever the user asks how much they're wasting, where the money is leaking, "what spend is doing nothing", zero-conversion keywords/terms, spend with no return, or an account-level waste roll-up — even if they don't say "wasted spend". Quantifies and prioritizes waste ONLY; writing the negatives belongs to `negative-keywords`, reallocating the freed budget to `spend-allocation`, and classifying WHY a term is irrelevant to `search-term-classifier`.

Wasted Spend

Goal (job-to-be-done)

Answer the question the account owner is really asking when they say "am I wasting money?": how many dollars of this account's spend returned nothing (or cost far too much per conversion), and where are the biggest leaks? Google shows you cost and conversions per entity, but never a single "this is your wasted spend" number, and never ranks the leaks so you know which one to plug first. This skill reads cost + conversions at every grain (search term, keyword, ad group, campaign), flags each entity that spends above a floor while returning ~0 conversions or running at a CPA far above target, rolls the wasted dollars up to one account number + % of spend, and returns the top leaks in priority order.

Who: account owner / media buyer / PPC manager inheriting or auditing an account. When: the "how much are we wasting" item on the account-audit checklist, or a one-off leak hunt.
Decision it drives: which leaks to plug first — the single biggest dollar drains, at the level where you can act on them (add a negative, pause a keyword, restructure an ad group, cap a campaign).
The differentiator: it produces one honest account number, counted once. The naive version double-counts — a wasted search term is inside a wasted keyword is inside a wasted ad group is inside a campaign, so summing "waste" across scopes inflates the total 3–4×. This skill counts each wasted dollar at exactly one grain and uses the finer scopes only to pinpoint the leak, not to add it up again.

Scope

✅ Quantify + prioritize waste — cost with ~0 conversions, or CPA far above target, across search terms / keywords / ad groups / campaigns; rolled up to a total, a % of spend, and the top leaks by dollars.
❌ Writing the negatives (which term, which match type, which level) → complementary negative-keywords skill. This skill flags the term and its wasted dollars; it does not author the negative list.
❌ Reallocating the freed budget (where the recovered dollars should go) → spend-allocation.
❌ Classifying WHY a term is irrelevant (intent mismatch, wrong geo, competitor, etc.) → search-term-classifier. This skill is the account-level roll-up those search-term skills feed.

Components (read these references as needed)

Tools / data plan: references/tools.md — the four scope queries (one per grain), the fields, and the conversion-lag caveat.
Framework / rubric: references/framework.md — the brain: the leak test, the wasted-dollars formula, and the count-once rule that keeps the total honest.
Output schema: references/output.md — the JSON this skill emits.

Operate

Input: the reporting window, an optional target CPA (or target ROAS for a value-tracking account — if none is supplied, derive the account baseline CPA = account cost ÷ account conversions), and the account's known conversion-lag window (default ~7 days). Per grain: cost_micros, conversions, conversions_value segmented by search term / keyword / ad group / campaign — see references/tools.md.

Process: apply references/framework.md. For each entity above the spend floor, compute CPA (cost ÷ conversions) and the wasted dollars = cost − (conversions × target_cpa), bounded to [0, cost] — so a zero-conversion entity wastes 100% of its cost, and an over-CPA entity wastes only the overspend beyond target. Flag it as a leak when conversions are ~0 or CPA ≥ ~3× target. Then count the total once: sum wasted dollars at the keyword grain for Search/Shopping + at the campaign grain for the channels with no keyword breakdown (Performance Max, Display, Video, Demand Gen); the search-term and ad-group scopes are diagnostic lenses that pinpoint where inside the keyword/campaign the leak sits, never added to the total. Gate on volume (ignore sub-floor entities), and mark leaks whose spend falls inside the conversion-lag window as provisional — a recent zero-conversion term may just be waiting for conversions to land.

Emit the JSON in references/output.md:

synthesis — three strings: headline (the number — total wasted, its % of spend, and the single loudest leak), diagnosis (the shape of the waste — is it concentrated in a few terms, spread across a whole ad group, or a dead campaign), action (the highest-dollar plug, where / what / why, handing the mechanics to the sibling skill).
leaks[] — the top leaks, each tagged with its scope, its wasted_amount, why it's a leak, and the executable recommendation {where, what, why}.
rollup — total_wasted, wasted_pct_of_spend, the count-once components, the by_scope breakdown (canonical vs diagnostic-lens), and the top leaks by dollars.

A renderer (the orchestrator's formats/*) turns this JSON into the human document. Emit pure data — no emojis, tables, markdown, or colors in the output.

Voice (don't copy the rules, link them): write every narrative line per _framework/writing.md — a question heading the data answers yes/no; the metric+delta carried as data, never spelled out in prose; first sentence answers the heading, then names the driver; one bridge line to the next section. Plain language for a non-technical owner — "this search term spent $840 and got nothing", not "high-CPA query with zero attributed conversions".

Example (illustrative — NOT rules)

The number: the account wasted $11,400 (18% of spend) last 30 days — counted once, no double-counting across scopes.
The loudest leak (search term): the query "free crm" spent $840 across two campaigns with 0 conversions → the whole $840 is wasted; hand it to negative-keywords to add as a negative (this skill won't pick the match type or level).
A hidden over-CPA leak (keyword): the keyword "project software" spent $2,300 for 3 conversions — a $767 CPA at a $120 target → ~$1,940 wasted (the overspend beyond target); pause or cut the bid, don't reallocate yet (that's spend-allocation).
A dead campaign (campaign): "Display — Prospecting" spent $3,100 for 1 conversion → counted at campaign grain because Display has no keyword breakdown; review the whole campaign, not a keyword.
Provisional, not counted firm: "new pricing tool" spent $310 with 0 conversions but all of it in the last 5 days → inside the conversion-lag window; flagged provisional, wait before cutting.

Wasted Spend

Goal (job-to-be-done)

Who: account owner / media buyer / PPC manager inheriting or auditing an account. When: the "how much are we wasting" item on the account-audit checklist, or a one-off leak hunt.

Decision it drives: which leaks to plug first — the single biggest dollar drains, at the level where you can act on them (add a negative, pause a keyword, restructure an ad group, cap a campaign).

The differentiator: it produces one honest account number, counted once. The naive version double-counts — a wasted search term is inside a wasted keyword is inside a wasted ad group is inside a campaign, so summing "waste" across scopes inflates the total 3–4×. This skill counts each wasted dollar at exactly one grain and uses the finer scopes only to pinpoint the leak, not to add it up again.

Scope

✅ Quantify + prioritize waste — cost with ~0 conversions, or CPA far above target, across search terms / keywords / ad groups / campaigns; rolled up to a total, a % of spend, and the top leaks by dollars.

❌ Writing the negatives (which term, which match type, which level) → complementary negative-keywords skill. This skill flags the term and its wasted dollars; it does not author the negative list.

❌ Reallocating the freed budget (where the recovered dollars should go) → spend-allocation.

❌ Classifying WHY a term is irrelevant (intent mismatch, wrong geo, competitor, etc.) → search-term-classifier. This skill is the account-level roll-up those search-term skills feed.

Components (read these references as needed)

Tools / data plan: references/tools.md — the four scope queries (one per grain), the fields, and the conversion-lag caveat.

Framework / rubric: references/framework.md — the brain: the leak test, the wasted-dollars formula, and the count-once rule that keeps the total honest.

Output schema: references/output.md — the JSON this skill emits.

Operate

synthesis — three strings: headline (the number — total wasted, its % of spend, and the single loudest leak), diagnosis (the shape of the waste — is it concentrated in a few terms, spread across a whole ad group, or a dead campaign), action (the highest-dollar plug, where / what / why, handing the mechanics to the sibling skill).

leaks[] — the top leaks, each tagged with its scope, its wasted_amount, why it's a leak, and the executable recommendation {where, what, why}.

rollup — total_wasted, wasted_pct_of_spend, the count-once components, the by_scope breakdown (canonical vs diagnostic-lens), and the top leaks by dollars.

A renderer (the orchestrator's formats/*) turns this JSON into the human document. Emit pure data — no emojis, tables, markdown, or colors in the output.

Voice (don't copy the rules, link them): write every narrative line per _framework/writing.md — a question heading the data answers yes/no; the metric+delta carried as data, never spelled out in prose; first sentence answers the heading, then names the driver; one bridge line to the next section. Plain language for a non-technical owner — "this search term spent $840 and got nothing", not "high-CPA query with zero attributed conversions".

Example (illustrative — NOT rules)

The number: the account wasted $11,400 (18% of spend) last 30 days — counted once, no double-counting across scopes.

The loudest leak (search term): the query "free crm" spent $840 across two campaigns with 0 conversions → the whole $840 is wasted; hand it to negative-keywords to add as a negative (this skill won't pick the match type or level).

A hidden over-CPA leak (keyword): the keyword "project software" spent $2,300 for 3 conversions — a $767 CPA at a $120 target → ~$1,940 wasted (the overspend beyond target); pause or cut the bid, don't reallocate yet (that's spend-allocation).

A dead campaign (campaign): "Display — Prospecting" spent $3,100 for 1 conversion → counted at campaign grain because Display has no keyword breakdown; review the whole campaign, not a keyword.

Provisional, not counted firm: "new pricing tool" spent $310 with 0 conversions but all of it in the last 5 days → inside the conversion-lag window; flagged provisional, wait before cutting.