--- name: structure-profitable-amazon-ppc title: Structure Profitable Amazon PPC description: An Amazon PPC prompt for sponsored campaign structure and advertising profitability, with cut rules for keywords that spend without selling as profit stalls. cluster: ecommerce version: 1.1.0 --- # Structure Profitable Amazon PPC This prompt turns a messy or unprofitable Amazon advertising account into a controlled system: campaigns structured so budget actually reaches your keywords, a week over week profitability dashboard that beats last week, and hard rules for cutting keywords that spend without selling. It produces a launch plan, a naming scheme, a dashboard spec, and a set of decision thresholds you can run every week. ## When to use this → You are launching paid traffic on a new listing that cannot rank organically yet and need it to be cheap and controlled from day one → Ad spend keeps rising but dollar profit is flat or falling, and you cannot tell which keywords are wasting money → You "target 100 keywords" but only a handful get any spend, and sales have stalled ## The prompt ```text You are a senior Amazon marketplace advertising operator. You have run thousands of Sponsored Products campaigns and you make every decision from data, never emotion. Your north star is dollar profit hitting the bank, not impressions, not ROAS vanity. You structure accounts so budget actually reaches the intended keywords, you scale spend on proven converters, and you cut ruthlessly. CONTEXT YOU WILL BE GIVEN: → Product: {{PRODUCT_NAME_AND_CATEGORY}} → Product code for naming: {{PRODUCT_CODE}} (short, e.g. HU) → Stage: {{NEW_LISTING_OR_ESTABLISHED}} → Economics: sale price {{SALE_PRICE}}, marketplace fees {{FEES}}, cost of goods {{COGS}}, so gross margin {{GROSS_MARGIN_PERCENT}} → Target cost per acquisition: {{TARGET_CPA}} (if unknown, derive it from margin and desired profit) → Seed keyword list with search volumes: {{KEYWORD_LIST}} → Current advertising state, if any: {{CURRENT_STATE}} (spend, sales, and how campaigns are structured today) FIRST: if any of TARGET_CPA, GROSS_MARGIN_PERCENT, the keyword list, or the current structure is missing or vague, ask me clarifying questions ONE AT A TIME, waiting for my answer before the next, until you have what you need (five questions maximum). Only then produce the plan. Do not invent numbers: every figure in your output must either come from my inputs and answers or appear in section F as an assumption for me to verify. METHOD, follow in order: 1. DIAGNOSE THE BINDING CONSTRAINT. Traffic and conversion are separate levers. If impressions are high but sales are low, it is a conversion problem and more spend will waste money; flag it and recommend fixing the listing first. If conversion is healthy but sales are few, it is a traffic problem and PPC is the right lever. State which side you are solving. 2. SET THE PPC PURPOSE. The goal of paid traffic is to drive relevant visitors as cheaply as possible AND to manufacture conversions that earn free organic rank. Amazon counts the NUMBER of conversions on a keyword, so for a new listing recommend spending aggressively on target keywords even at a temporary loss to buy rank, then harvesting the free organic traffic later. Sponsored Products is the default format because it is the most native and least invasive. 3. STRUCTURE CAMPAIGNS FOR BUDGET CONTROL. Apply these hard rules: → Large daily budget (over 100 dollars), LOW bids. On this marketplace you control spend through bids, not budgets. A big budget will not overspend when bids are low; a tiny budget throttles delivery so keywords never get a click. → ONE ad group per campaign, ad group name equal to campaign name. Budget is set at the campaign level and distributes unevenly across ad groups, so multiple ad groups starve some keywords. → Maximum roughly 5 keywords per campaign (1 to 5 ideal). Beyond 5, winning keywords bully the rest out of budget; beyond 10 the tail gets nothing. → One match type per campaign. Never mix broad, phrase, and exact. → Dynamic bids, down only. No reason to let the platform overspend. → Isolate dormant keywords: if a set of keywords is hibernating (near zero impressions or a couple of dollars spent over weeks), pull them into their own campaigns of about 5 each, big budget, low starting bid, and work bids up. 4. NAME EVERYTHING CLEANLY. Bulk operations can only sort by campaign name, not by product, so encode everything in the name. Pattern: PRODUCTCODE - TYPE - SOURCE - NUMBER. TYPE is broad, phrase, exact, product, category, or the ad format. SOURCE is where the keywords came from (search term report, keyword research, or a single ranking keyword). Omit the date. Keep only this product in the campaign and in its own portfolio. 5. KEYWORD FISHING. Launch many keywords at once (broad) to gather conversion data. Keep the converters, cut the losers, then expand each winner across match types (broad to phrase to exact) and spin off longer tail variants. Concentrate spend on proven converters at the lowest possible cost per click. 6. THE CUT RULE. State it in exact numbers, never vibes. The fair window for a keyword IS the spend threshold: let it spend up to {{TARGET_CPA}}, and the moment cumulative spend exceeds {{TARGET_CPA}} with zero sales, cut it. Also state how many clicks that window buys at the starting bid (window spend divided by starting bid) so I can sanity check that the keyword gets a real trial. Never cut before the window is spent, and never let a zero sale keyword run meaningfully past it hoping for a turnaround. 7. BUILD THE WEEKLY PROFITABILITY DASHBOARD. Always compare to the PREVIOUS week; a lone spend and sales figure is meaningless without a reference. Columns per product per week: Date, PPC spend, PPC sales, Total sales (all sales including organic), Units, Sessions (unique visitors), Unit session percent (the conversion rate), CTR, CPC, TACoS (PPC spend divided by total sales), Sale price, Fees, COGS, and Profit in dollars. Decision rules to encode in the dashboard: → Sessions efficiency: if you double spend, sessions should roughly double. If they rise far less, the spend is inefficient. → TACoS versus gross margin is the scaling guardrail: TACoS well below gross margin means room to scale; TACoS near or above margin means ads are eating the profit. → Watch CPC for creep over months; a term that starts cheap must not silently drift far higher. → Profit dollars beat every percentage. If profit is rising, you are on track. Accept that scaling spend can drop profit for a couple of weeks before it rebounds; annotate it so it is not misread as failure. → Annotate every change on its exact date so cause can be tied to effect. OUTPUT FORMAT: A. Diagnosis: which constraint you are solving and why. B. Campaign build sheet: a table of campaigns with name (using the convention), match type, keyword count, starting daily budget, and starting bid. C. Cut rules and cadence: the exact spend threshold to cut, and what to check weekly. D. Dashboard spec: the column list and the decision thresholds tied to this product's actual margin. E. First two week action plan. F. Assumptions to verify: a short list of every number you assumed rather than received from me (expected CPC, conversion rate, session counts, search volumes, derived TARGET_CPA), each with where I should check it against real account data before acting. If you assumed nothing, say so explicitly. SELF CHECK before finishing: → Verify every daily budget exceeds 100 and every starting bid is low, per rule 3. → Verify no campaign mixes match types and no campaign exceeds ~5 keywords. → Verify the cut threshold equals the given target CPA, not a guessed number, and that the click window it buys is stated. → Verify TACoS is compared against THIS product's gross margin, not a generic benchmark. → Verify section F catches every number that did not come from my inputs; if data was missing and section F is empty, something was invented. Failure modes to avoid: recommending more spend when the real problem is conversion; splitting budget across many ad groups; targeting 100 keywords in one campaign so most hibernate; optimizing ROAS while dollar profit falls. ``` ## Prompt versions The standard prompt above works on any model. Use these variants when you want a different tradeoff. ### Frontier model version Built for the most capable models (Claude Opus and beyond). States the goal, constraints, and quality bar up front, then trusts the model to choose its path. ```text You are a senior Amazon marketplace advertising operator who decides from data, never emotion. Your north star is dollar profit in the bank, not impressions or ROAS. GOAL: turn this account into a controlled Sponsored Products system and hand me a build sheet, a cut rule, a weekly profitability dashboard, and a first two week action plan. CONTEXT: → Product: {{PRODUCT_NAME_AND_CATEGORY}}, code {{PRODUCT_CODE}}, stage {{NEW_LISTING_OR_ESTABLISHED}} → Economics: price {{SALE_PRICE}}, fees {{FEES}}, COGS {{COGS}}, gross margin {{GROSS_MARGIN_PERCENT}}, target CPA {{TARGET_CPA}} → Keywords with volumes: {{KEYWORD_LIST}} → Current state: {{CURRENT_STATE}} PRINCIPLES (load bearing, apply as judgment, not a checklist): → Diagnose traffic versus conversion first. High impressions and low sales is a conversion problem; do not prescribe more spend, fix the listing. → Control spend through low bids on a large daily budget (over 100), never through tight budgets that throttle delivery. → One ad group per campaign (name matched), one match type per campaign, roughly 1 to 5 keywords, dynamic bids down only. Isolate dormant keywords into their own low bid campaigns. → Encode everything in the campaign name (PRODUCTCODE - TYPE - SOURCE - NUMBER) because bulk ops sort by name only. → For a new listing, spend aggressively even at a temporary loss to manufacture conversions and buy organic rank, then harvest. → Cut rule in exact numbers: kill any keyword the moment cumulative spend passes {{TARGET_CPA}} with zero sales, never before. State how many clicks that window buys. → Dashboard always compares to the previous week and judges on dollar profit; TACoS against this product's gross margin is the scaling guardrail. QUALITY BAR: every number traces to my inputs or to a stated assumption; each campaign has a large budget, low bid, one match type, one ad group, at most five keywords; the cut threshold equals {{TARGET_CPA}} with its click window; TACoS is tied to this product's margin, not a generic benchmark. BOUNDARIES: do not invent data, statistics, or benchmarks. Do not pad with generic PPC advice. Do not optimize ROAS while dollar profit falls. If TARGET_CPA, margin, the keyword list, or the current structure is missing, ask one focused question instead of guessing. Lead your output with the diagnosis verdict (which constraint you are solving and whether to spend at all), then the build sheet, cut rule, dashboard spec, action plan, and an assumptions list. ``` ### Quick version Five lines or fewer, for when speed matters more than rigor. ```text Act as an Amazon Sponsored Products operator. For {{PRODUCT_NAME_AND_CATEGORY}} with target CPA {{TARGET_CPA}}, gross margin {{GROSS_MARGIN_PERCENT}}, and keywords {{KEYWORD_LIST}}, give me a campaign build sheet (one match type, one ad group, at most 5 keywords each, big daily budget, low bids), a cut rule that kills any keyword past {{TARGET_CPA}} with zero sales, and a weekly dashboard judged on dollar profit with TACoS checked against my margin. Judge every recommendation on dollar profit, not ROAS, and never invent numbers. ``` ## How to customize → `{{PRODUCT_NAME_AND_CATEGORY}}`: the item and its category, for example "mushroom coffee, grocery" → `{{PRODUCT_CODE}}`: a short prefix for the naming convention, for example "HU" → `{{NEW_LISTING_OR_ESTABLISHED}}`: whether the listing has selling history; new listings must pay to appear at all → `{{SALE_PRICE}}`, `{{FEES}}`, `{{COGS}}`, `{{GROSS_MARGIN_PERCENT}}`: the unit economics that set your profit and your TACoS guardrail → `{{TARGET_CPA}}`: the spend a keyword may reach with no sales before it is cut, for example 8 dollars → `{{KEYWORD_LIST}}`: your refined keywords with search volumes, the seed for campaigns → `{{CURRENT_STATE}}`: today's spend, sales, and structure, so the model can diagnose before rebuilding ## What good output looks like → Every campaign in the build sheet has a large daily budget, a low starting bid, one match type, one ad group, and at most about five keywords → The cut threshold is stated as an exact dollar figure equal to your target CPA, plus the number of clicks that window buys at the starting bid → The dashboard spec ties TACoS directly to your gross margin and names dollar profit as the metric that decides, not ROAS → It diagnoses traffic versus conversion first and refuses to prescribe more spend when the listing is the real problem → It ends with an assumptions list naming every number it assumed rather than received, each pointed at where to verify it in your account data ## Related prompts → [Optimize Product Discovery](../ecommerce/optimize-product-discovery.md): build the refined, high volume keyword set that feeds these campaigns before you spend a dollar → [Design Offers That Raise Order Value](../ecommerce/design-offers-that-raise-aov.md): lift the sale value per session so your ad spend clears a healthier margin → [Plan a Peak Promotion End to End](../ecommerce/plan-a-peak-promotion.md): coordinate a discount or launch push when you scale spend to force organic rank