Before we rebuild an account, we read every target it has ever run. The plan you sketch before the audit is only a template, and templates leak money. This account held 552 PPC targets over its lifetime, with 346 enabled and 129 paused, and we read all of them before we touched a single bid. We run the audit before the restructure, never after it, because the target type economics in the reports tell you where the money actually lives and where it only looks like it lives.
We have spent ten years in Amazon e-commerce and we manage 15 to 20 accounts at a time, so we watch the same pattern repeat. A small set of targets carries almost all sales, while the long tail holds bids nobody has challenged in a while. In this account the concentration was severe, because phrase match alone produced $84,647.95 of roughly $114K in total PPC sales, which means one match type carried the whole account.
Those 271 phrase targets ran at 36.91 percent ACOS. That mix of heavy volume and middling efficiency is what phrase is built for. Phrase does the discovery by wrapping a root term and showing which real searches convert. But discovery is not the destination, so the winners inside those 271 targets get mined into exact match and single keyword campaigns where we control the bid, the budget, and the placement, instead of leaving them to sit inside a 36.91 percent blend.
The number that makes people blink is exact. Exact ran at 74.87 percent ACOS across 53 targets and only $5,274.21 in sales. Many assume exact is the efficient tier by default. It is only efficient when the exact terms come from evidence, and these did not. Someone guessed their way into exact, so the account paid premium placement for terms that had never proven themselves anywhere. Broad match told the same story at 55.60 percent ACOS across 21 targets and $6,313.10, because broad without steady harvesting and negatives drifts toward whatever Amazon decides to serve.
The auto campaigns did quiet, solid work, and their 84 targets earned every dollar they spent. Close match ran 41 targets at $11,173.73 in sales and 22.42 percent ACOS, loose match ran 33 targets at $6,721.11 and 15.67 percent, substitutes ran 8 targets at $1,232.64 and 13.47 percent, and complements ran 2 targets at $79.98 and 0.70 percent. Auto match types earn their keep because Amazon targets off its own conversion data, so the account collects sales it did not have to research.
Product targeting held the cheapest growth and almost nobody had built it. ASIN targets produced $11,470.71 at 19.14 percent ACOS across 109 targets, and category targets produced $1,577.47 at 17.37 percent across 12 more, so the two most efficient manual tactics in the account were also the most underbuilt at 122 total targets. The single Sponsored Display target ran $1,448.59 at 43.06 percent, which reads like an experiment someone launched and then forgot. When the best economics sit in the smallest section, the restructure writes itself.
Every ASIN earns three campaigns
The rule we carried out of this audit is now our default. Every ASIN in an account gets three campaigns before anything else: an auto campaign for close and loose match, a phrase campaign around its core terms, and an ASIN targeting campaign aimed at the competitors it can realistically beat. Those three cover discovery, harvesting, and conquest at the three price points this audit exposed, so every additional campaign in the account has to justify its existence against them.
This is why we keep flagship accounts under 15 percent TACoS, and it has little to do with bidding harder than the next person. We have managed more than $10M in ad spend, and the lesson that survives all of it is that we do not move money until we know where it already lives. Reading all 552 targets is slower than opening a bulk sheet and cutting everything above target ACOS, but that slow read showed us that phrase was a ours and not a bleeder, that exact was running on guesses, and that the cheapest growth in the account had barely been built.
← Back to all posts