The Amazon PPC Audit

Part 1

The philosophy of the audit

Most accounts are not under-optimized. They are un-audited. Small changes pile up for months — a campaign duplicated, a budget bumped, a target copied — until nobody remembers why anything exists. An audit is how you get the account back to a state you can reason about.

That is first-principles thinking applied to ads: break the problem down to fundamental truths, then rebuild the solution from the ground up. Don't follow SOPs blindly. Don't apply rules without context. Don't assume what worked for one account works for another — even if the other account is yours from last year.

Run the audit in layers, not as an annual event. A fifteen-minute weekly pulse: spend vs. plan, ACOS deltas on your five biggest campaigns, budget caps hit before noon. A thirty-to-sixty-minute monthly pass: steps 3 through 8 — bids, search terms, structure, budgets. The full ten-step audit runs quarterly, or before any major planning cycle. Each layer catches problems at a different size — weekly finds them while they are still cheap, quarterly finds the patterns you have stopped being able to see.

Part 2

The ACOS formula — foundation of everything

Every diagnosis in this guide reduces to one formula. Internalize it and half of PPC management becomes arithmetic:

Read it as a machine: your bids set CPC. Your listing and targeting set CVR. Your product sets AOV. ACOS is just the output. When ACOS moves, one of those inputs moved first — and the audit exists to find out which.

A worked example. Your money keyword runs at 22% ACOS against a 30% target — comfortable. Three weeks later it reads 31%. Nothing was changed in the account. Decompose it: CPC went from $0.90 to $0.94 (competition or a modifier drift) — a 4% worsening. CVR went from 14% to 11.5% — an 18% worsening. The dominant lever is conversion, not cost. Cutting bids would shave the symptom and starve a healthy keyword; the fix is on the listing — a new competitor photo, a price move, two recent negative reviews sitting on page one. The formula tells you where the problem lives before you spend a rupee being wrong about it.

Part 3

The 10-step audit framework

Run these in order. Each step feeds the next.

Step 1 — Total sales performance

Combine Business Reports (total sales) with the Advertised Product Report (ad sales and spend). You are looking for four patterns: ASINs with high total ACOS (scale back), ASINs under-advertised but generating sales (support them), ASINs with no ads but meaningful sales (new opportunity), and ASINs draining budget at poor ROAS (reduce or pause).

Then check budget distribution across the catalogue. The ideal state: each product's ad-spend share roughly aligns with its sales contribution.

Step 2 — Campaign placements

Three placements, three temperaments: Top of Search typically converts best; Rest of Search sits in the middle; Product Pages often carry the lowest CVR and the highest ACOS. Pull the placement report and look for spend-share that contradicts performance.

Step 3 — Bid management

Every keyword falls into exactly one of four categories, and each category has its own bid formula (see the quick-reference at the end). The audit's job is to sort keywords into categories — not to nudge bids by feel.

Step 4 — Search term mining

Pull the search term report and mine it for four things: tactic allocations to fix, branded spend leaking into non-brand campaigns, irrelevant terms to negate, and converting terms to harvest into exact. N-gram analysis (see our n-gram guide) turns this from a scroll into a system.

The branded-leak check deserves its own five minutes. Search every campaign's terms for your own brand name. Brand terms convert at rates non-brand traffic never touches — which means every rupee of brand spend inside a non-brand campaign makes that campaign look better than it is and hides the real non-brand economics. Pull brand terms into their own defense campaigns, then re-read every non-brand campaign's numbers. Accounts routinely discover their best-performing campaign was mostly branded traffic wearing a costume.

Step 5 — Tactic allocation

Step 6 — Campaign structure

Check naming conventions (can you tell what a campaign does from its name alone?), segmentation by goal (research vs. scaling vs. brand defense), and whether the structure matches the catalogue size. Structure is memory — if you cannot navigate it, you cannot optimize it.

Step 7 — Targeting mix

In a mature account: 50%+ of spend on exact match (the scaling engine), 30–40% on broad/phrase (discovery), 10–20% on auto (continuous research). A young account runs discovery-heavy and shifts toward exact as terms graduate.

Step 8 — Budget management

Step 9 — Ad type utilization

Step 10 — Dayparting opportunities

Check whether bids reflect when your buyers actually convert. The philosophy: optimize by revenue per click, not by raw sales volume — high volume at thin RPC is activity, not profit. (Full method in our dayparting guide.)

Part 4

Troubleshooting when performance drops

When ACOS spikes or sales slide, work the diagnostic framework — in this order:

Then take the three-lever view, straight from the formula: if CPC rose — check recent bid changes, "up and down" dynamic bidding, inflated placement modifiers, or new competition; fix with the RPC formula. If CVR fell — check product changes (price, images, reviews, Buy Box loss), seasonality, drifting search terms, or a placement shift away from Top of Search; note that a falling organic rank compounds the problem. If AOV dropped — check your price and bundle structure before touching a single bid.

Part 5

The negative spiral — and how to break it

The most dangerous pattern in Amazon PPC is the spiral: lower CVR → worse placement → lower CVR → worse placement — each turn making the next one cheaper to fall into. Accounts rarely die from one bad decision; they spiral.

It usually starts innocently. A competitor undercuts your price, so your CVR dips from 14% to 11%. At the same bid you now win fewer auctions, so your ad slips from Top of Search to page two — where CVR is half again lower. Sales drop, so you cut bids to protect ACOS, which pushes you further down the page. Two months later the account looks broken and nobody can point to the decision that broke it — because there wasn't one. There were six small ones, each locally reasonable.

Part 6

Common audit mistakes

Part 7

Quick-reference formulas

The arithmetic that runs the whole framework:

This is the exact audit sequence we run on client accounts — the same order, the same formulas, the same discipline. If you want it run on yours, the first diagnosis is free.