if (window.onLi) onLi(30580756); Skip to content
Operator’s Notes
Case file 01

The agency’s report gave four reasons. Only one held up against the data, and it was listed last. Finding where to look took one subtraction the report never ran: total sales minus ad sales.

−34%Total sales
−26%Ad sales
−46%Organic
What the agency said

Two strong months. Then September sales fell by about a third.

A month like that forces a decision. Whatever the owner settles on as the cause is what he spends the next month fixing.

He wanted two things: to know why sales fell, in a form he could check, and to know who on his side was going to fix it. The agency’s monthly report gave him neither, though it looked as if it did. He forwarded it to me and asked what I made of it.

STYLISED RECREATION · NOT THE ORIGINAL

September performance recap

[Account manager] · [Amazon PPC agency] to owner

Total sales were down in September versus August. The main drivers:

  1. Competitor pricing. Four competitors cut their prices during the month (a percentage quoted for each).checked against what?
  2. Lower search demand. Search volume for the main terms declined.the market, or our share of it?
  3. Seasonality. August is historically the strongest month.no data attached
  4. Campaign efficiency. Some higher-ACOS campaigns were paused to improve efficiency, which also reduced incremental sales.listed last. the one to check first.

Plan of action for next month:

  • Launch campaigns with the best-performing keywords and targets, to improve control and capture more sales.
  • Test category targeting.
  • Run a two-week test with more focus on Sponsored Brands and Sponsored Display.
  • Prioritize top-of-search placements for branded and high-potential keywords.
  • Keep optimizing toward the account ACOS target.

Attachment · Monthly_Dashboard.xlsx · Reasons and plan paraphrased · names, figures and dates removed

A reason you can check is a reason you can trust or reject. A reason you can’t check is just a story.

Why owners don’t check, and why it gets expensive

If you pay an agency, you’ve read a page like this. It sounds reasonable: competitors moved, demand dipped, it’s seasonal. What it doesn’t tell you is whether any of it is true, or whether the work you’re paying for is working.

Most owners read it and move on. Checking it yourself feels like doing the job you already pay someone else to do.

That’s where it gets expensive, because each reason points at a different fix. Believe the price story and the natural move is a price cut. Believe the demand story and you wait for the market to come back.

If the reason is wrong, another month of ad spend, fees and effort goes at the wrong problem. And the earliest you’d find out is the next monthly report, by which time that month is already paid for.

Checking is a smaller job than it looks. Each reason is a claim about the world, and claims about the world leave traces. Competitor prices leave a price history. Demand leaves a trace in Amazon’s own search data. So I pulled both and ran three checks, in order. Each one picked up the question the last one left open.

The three checks

Which explanations held up.

Check 1 of 3 · Competitor prices

Four price cuts were cited. Price history told a different story.

Didn’t hold upShow the evidenceHide the evidence

Of four price claims, one move never happened and one ran the other way. One was overstated roughly twofold, and one was true but beside the point.

September, start to end · illustrative

CompetitorReportHistorySo
Competitor A−20%0%never moved
Competitor B · click leader−10%+11%went up
Competitor C−17%−7%overstated ~2×
Competitor D−21%−12%still priced above the brand

The brand’s own price was flat all month. It took no price action at all.

Competitor B matters most, because it’s the niche’s click leader. The report had the right two prices in the wrong order: it started the month at its low and rose about 11%. The competitor that wins the most clicks got more expensive in the month the brand’s sales fell.

Why it mattersThe natural answer to a price story is a price cut, and a price cut comes off every unit you sell. Price doesn’t explain September the way it was presented, so on this evidence a cut would have answered moves that mostly didn’t happen as described.

The limit of this evidence

A careful reader should push back here, so I’ll do it first. Price-history tools show the lowest new-offer price. That isn’t always the Buy Box price, and a clipped coupon may not show up in it.

That blind spot can explain a gap of a few percent. It can’t explain a price that never moved, or one that moved the other way.

Check 2 of 3 · The subtraction

Total sales minus ad sales. The report never ran it.

Never runShow the evidenceHide the evidence

Competitor prices and seasonality would usually hit paid and organic roughly alike. But organic fell faster.

Monthly sales, Jul → Sep · illustrative · no scale

JulAugSep
total −34%ad sales −26%organic (total − ad) −46%

If competitor prices don’t explain the fall, the next question is where the lost sales went. This is the fastest check of the three, and it changes the most.

The ads report is the agency’s scoreboard, and organic sales don’t appear in it at all. Read only that, and the bigger loss is invisible. Split total sales into paid and organic and the picture changes.

A broad outside explanation would normally be expected to pressure both. When organic falls materially faster, that divergence is a reason to look inside the account before accepting the outside explanation as sufficient.

Why it mattersOrganic sales are the ones the brand doesn’t pay for click by click. When they fall faster than ad sales, the business comes out of the month more dependent on paid traffic, not less.

What the pauses do and don’t explain

Some of the drop is easy to explain. Campaigns were paused and spend came down, and that clearly accounts for part of the fall in ad sales. By itself, it doesn’t explain why organic sales fell even faster.

In fairness, this was a fall from a two-month peak, not a collapse. The question is why the unpaid side fell harder.

Check 3 of 3 · Search demand

On the main search, the brand’s share fell from 7.8% to 4.9%. Available impressions dipped in September but ended above where they started.

Not ruled outShow the evidenceHide the evidence

Across the reviewed weeks, Aug 1 – Sep 26, 2026, the brand’s share on the main search fell from 7.8% to 4.9%, while estimated available impressions ended above their starting level.

Main search term · weekly impressions · week of Aug 1 vs week of Sep 26, 2026 · end weeks only; impressions dipped during September

~415k ~545k 7.8% 4.9% Week of Aug 1Week of Sep 26
available impressions (estimate)the brand’s share

Check 2 left one question: why did the unpaid side fall harder? The report’s answer was demand. It said search volume for the main terms declined.

Available impressions also declined during September, so weaker demand cannot be ruled out as part of that month’s decline. The share loss is on the brand’s side of the auction either way. Either way, the brand was winning a smaller share of what was there.

Why it mattersSo demand may have played a part in September. The share loss is the part you can act on. It’s the better news, even if it doesn’t feel like it: you can’t control category demand, but you can change how the account competes for the demand that exists.

Not checkable

Seasonality: plausible, but the report offered no data for it.

UntestedShow moreHide

Not wrong, just unchecked.

Listed last

Paused campaigns: the one reason inside the account held.

HeldShow moreHide

The price story didn’t hold up. Weaker demand can’t be ruled out, and seasonality couldn’t be checked. The one inside the account held, and it was the one listed last. That leaves the useful question: why a few reasonable pauses seem to have cost so much more than expected.

Would your last report hold up to the same checks?

The Amazon Growth Diagnostic runs them across your whole account before anything else gets changed, and tells you what’s constraining growth and what to do first. View-only access. One document back.

Underneath the report

The report was one page. The problem ran through the whole brand.

Every cut looked sensible on its own. Reason four, the one the report listed last, said some higher-ACOS campaigns were paused. That part was true. But the report treated the pauses as one line item among four. They were really a window into how the account was run.

On the priority searches I inspected, routing wasn’t controlled enough for a single bid, budget or negative decision to land predictably.

The evidence: each money search could be reached through 9 to 14 campaigns, and across the account 270 search terms appeared in three or more campaigns, carrying 68.6% of search-term spend.

The overlap isn’t the problem in itself. The problem is control. If the same important search can be served through many routes, the account can’t reliably steer money toward that search, or away from it.

One money term · nine campaigns · illustrative

Money term

Nine routes are drawn because one key search could be reached through nine campaigns. Which ones were cut is illustrative.

Add up enough sensible local cuts and a term can lose the auction.

The ads stop showing where the buyers are, and nothing in the reporting showed the term being evaluated as a whole. On one key search, share fell from 24% to 7.4% in a week, and to 3.6% the week after. The first drop came in the week estimated available impressions were highest; available impressions fell in the following weeks.

That key search · the brand’s share · two weeks

24%Early Sep
7.4%One week later
3.6%The week after

Not a slow drift. Most of the fall, 24% to 7.4%, came in a single week in early September, the week estimated available impressions were highest. They fell in the weeks after.

Why this matters beyond one bad month

So when ACOS looked high, campaigns were paused and budgets cut one at a time. Each decision was defensible inside the ad account. What didn’t appear in the reporting was what those decisions did together: to share on the searches that matter, to organic rank, to total sales.

Available impressions on that search fell in the weeks after, too. Even so, the brand had gone from a real presence on that search to a sliver of it.

This case doesn’t show which cuts did what, and the routes drawn above are illustrative. The point is the mechanism, and it isn’t specific to this brand. When an account lacks control over where its money goes, and the reporting doesn’t show the effect across paid, organic and share, the ads can look efficient while the brand gives up position. And the report can read as good news while it happens.

A report can show a better ad metric while the owner is left with the harder question: did the business get better?

The next-month plan

Same account. Same month. Two plans.

The plan at the bottom of the report decides whether September stays one bad month or becomes the pattern.

The Amazon PPC agency’s plan

Manages the ad account

  1. Launch campaigns with the best-performing keywords and targets, to improve control and capture more sales.
  2. Test category targeting.
  3. Run a two-week test with more focus on Sponsored Brands and Sponsored Display.
  4. Prioritize top-of-search placements for branded and high-potential keywords.
  5. Keep optimizing toward the account ACOS target.

Five tactics, several of them individually reasonable. None says what it should change in total sales, share or rank, and the last one keeps the same scoreboard: account ACOS.

Operator’s Notes plan

Operates the brand’s growth

  1. Restructure PPC to regain control of traffic and capital.
  2. Build the measurement loop across paid, organic and competitors.
  3. Improve click-through and conversion, so more of the right traffic becomes sales.

Three decisions: the control system, the scoreboard, and the asset the traffic lands on. Each is judged on share, rank and total sales.

The tactics changed. The scoreboard didn’t.

What the plan measures, and what it repeats

Read the agency’s plan for what it measures. It’s built around campaigns, placements and ACOS. Total sales, organic performance, rank and market share aren’t part of the scoreboard it lays out. That describes the plan as presented to the owner. It says nothing about what the people behind it care about.

Then read it for what it repeats. September ran on one operating logic: decide at the campaign level, judge through ACOS, and don’t measure what those decisions do to total sales, organic rank or share. The next-month plan keeps that logic and adds more activity inside it.

To be clear, this case doesn’t show that optimizing to ACOS caused September. What the plan does is leave the same blind spot in place. The report itself described the trade: campaigns paused to improve efficiency also reduced sales. If the scoreboard stays ACOS, the account can improve the number the ads manager owns while the business loses ground on the numbers the owner lives on.

An example from the plan: branded top-of-search

Defending branded searches can be strategically useful. Branded traffic also tends to convert well because the shopper already knows the brand. If those paid clicks replace purchases the brand would have won organically anyway, they can make account ACOS look better without creating equivalent extra sales.

Without measuring incrementality, share and organic impact, a cleaner ACOS can hide the underlying problem rather than solve it. Nothing in this case shows that happened here. The point is that the plan had no way to tell.

Sit in the owner’s chair for one more month

He could cut price to answer competitor moves that mostly didn’t happen as described. He could approve more campaigns in an account that was already hard to steer. He could watch ACOS improve while total sales, organic position and share keep slipping, and read it as progress.

He could spend more without knowing whether any of it creates extra sales. And he would still be the one checking the work he pays for. None of those is certain. Each is possible, and each is another month of fees and spend on the wrong problem.

None of this needs a bad agency or a careless manager. It only needs a setup where the whole number is no one’s job.

Why it happens

Why a capable team writes the first plan.

I don’t think anybody here was careless. A capable person inside an account-management system is pulled toward the metrics and tasks they own: bids, campaigns, ACOS.

One real senior Amazon PPC manager job description

15+Amazon accounts to own
40–45working hours a week
2.7–3.0hours per account per week, at 15 accounts

Before meetings, reporting, Slack and email, training, launches and internal work. One job description from the category, not a claim about this agency or any other.

Who owns the whole number?

Keeping the bids moving is one job. Looking at total sales, organic rank, share, conversion, price, the listing and the competition together is another, and it needs time the box above doesn’t leave much of.

Ads can sit with one specialist and the listing with another. Each can do the job well while no one owns what happens between them. When that happens, the checking comes back to you.

The report is a symptom. The question underneath it is who owns the whole number: total sales, organic rank and share. Your PPC manager may be capable. The question is whether their time, scope and scoreboard are built around the same whole-business outcome you care about.

When no one else holds the whole picture, the owner does, by default. That’s the most expensive seat in the business, spent checking work it already pays for.

The prescription

What I would do if this were my brand.

Regain control first. Then go on the offense. Every decision is judged on the whole number: total sales, organic rank and share.

Decision 1 · control

Restructure PPC

Fix how PPC acquires and routes traffic. The product already sells. The constraint isn’t “nobody wants this.” It’s that PPC isn’t run as a system for directing traffic and capital toward the searches that grow the brand.

What sits under it
  • 270 search terms ran through three or more campaigns, 68.6% of search-term spend. Define one intended primary route for each priority search, and enforce it where appropriate.
  • 111% ACOS on one competitor-product target, in the same campaign as a target at 26%. Review the loss-maker on its own economics.
  • 61/63 observed days the best large campaign reached at least 90% of today’s daily budget. Budget history wasn’t available, so first confirm whether it was capped back then.
  • 497 search terms for newer versions outside the stated fit range drew spend, for 14 orders. If the product doesn’t fit them, negate those searches, once fitment is confirmed.
Decision 2 · control

Measure the whole system

Know whether a change worked: track share, rank and conversion, not just ACOS. Without a baseline, no change can be judged. The loop is simple: change, observe, compare against the baseline and competitors, learn, reallocate.

What sits under it
  • 24→3.6% share on one key search in two weeks, and share didn’t appear in the reporting.
  • 61.9% of the niche’s clicks sit in a sub-category the brand isn’t listed in. A category correction may improve relevance, and should make rank and peer comparisons more meaningful if Amazon accepts it.
  • 4 things to track from day one: search-query share, organic rank on priority searches, click share against the niche leaders, and conversion.
Decision 3 · offense

Win the click, then the sale

Improve the listing so more of the right shoppers click and buy. Once traffic is controlled and measured, the listing decides how much of it turns into sales.

28.4%Niche leader
5.8%The brand
What sits under it
  • 3 text badges on the main image don’t read at thumbnail size. In my judgment the whole image stack needs reworking, main image first, and tested rather than assumed.
  • 33.4% of the niche’s return mentions cite compatibility. It’s a major objection in this niche, so the listing should answer it clearly and early.
  • A strength to build on: 40.8% of the niche’s negative reviews complain about the problem this product is built to solve.

Structure buys control, not an instant drop in ACOS. The work is an experiment, measured against a baseline recorded first. It isn’t a forecast. If PPC is restructured, measurement is in place and the listing improves, yet sales, organic position and share stay flat, the diagnosis gets revisited, not defended.

Request a Diagnostic QuoteSame method, your data.
Run it yourself

The check you can run tonight.

You don’t need me for the first pass. Each of these takes minutes, and the first needs only numbers you already have. The whole account is a bigger job.

  1. 1Total sales minus ad sales, month by month.

    That’s your organic line. If it falls materially faster than ad sales, the decline isn’t confined to paid media. Look inside the account before you accept a broad outside explanation.

  2. 2Price history against the claims, by date.

    When the report blames competitor pricing, pull those exact products. Did the price move? Which way, and by how much?

  3. 3The market against your share.

    When the report blames demand, look at the term’s total market, not just your impressions. If the market is flat or growing and your share is falling, look inside the account before you blame demand.

If you’d rather I ran it

The Amazon Growth Diagnostic

Find out what’s constraining growth, and which few decisions to make first. The same checks you just read, run across your whole account and reduced to the decisions that matter.

Nothing is paid until you’ve seen the quote.

Next: a short form about your account · quoted to your account · view-only access · the written findings are yours to keep

Only want a second opinion on your PPC? The free PPC Second Read looks at the ads only →

What you get

  • The few decisions that matter, in order: what to do first, and what not to touch.
  • What’s constraining growth, with the evidence for each finding.
  • The story you’ve been told, checked against the data, claim by claim.
  • A baseline to judge every change against: share, organic rank and total sales, not just ACOS.

See an excerpt of the summary you’d receive →

What it isn’t, and your side

If the explanation you were given holds up, I’ll tell you that. It isn’t a replacement for your agency. You don’t have to fire anyone or switch anything. The findings are yours to use however you want: your current team can act on them, you can, or you can ask me to own the implementation.

Your side: grant view-only access, then read one document in a sitting. Pulling, reconciling and checking the data is on me.

Who it’s for: established Amazon brands already paying an agency or PPC manager, where growth has stalled and you’re not sure the explanation is true. You want it done for you, not taught. Not for owners who want to learn to run it themselves. A coach is the better buy for that.

“What sets him apart is his depth of understanding: he doesn’t apply a one-size-fits-all approach, but tailors strategy to each product and category based on what actually works in that space.”
Aiden C. · Founder, 8-figure Amazon brand · USA · from my agency years

I’m Konstantin. I spent five years running Amazon PPC on the agency side, then built and ran my own brand with my own money. When it’s your money, you stop reading reports and start reading data.

Not ready?

Get the next Case File.

Real Amazon case studies, growth experiments, and practical insights on PPC, conversion and what’s actually driving business performance.