28 Aug 2026 · Margin desk · 5 min read
The fee tier that turned six SKUs negative
An agency ran ads against loss-making inventory for five weeks. Every dashboard they owned looked healthy the whole time.
Decision capacity · per analyst
Amazon agency software that adds clients without adding analysts
Set the dials to your agency's shape — clients, ASINs, hours per review — and read the number the routine work is costing you.
The bottleneck at 30 clients isn't talent. It's decision capacity.
Set the dials to your shape. The output is the number you take to your own leadership.
Agency worksheet
Clients you could add without hiring
12
61
hours freed / week
$190k
analyst cost redirected
198
decisions / week
Model assumes agents absorb 55% of routine review time. Your own numbers replace these during the Decision Audit.
Why agencies stall, and at what
Agencies do not stall at thirty clients because they run out of talent. They stall because every additional account adds a fixed weekly cost that no amount of seniority removes: somebody has to look at it, decide whether anything changed, and write down what they found.
That is decision capacity, and it is the constraint the calculator above is measuring. Not headcount, not hours worked, not tooling spend — the number of accounts one analyst can hold in their head at a review cadence the client will accept.
The three inputs, and what they mean
Set them to your agency rather than to a flattering number. The output is only as useful as the hours figure is honest.
Accounts managed
Live accounts across the whole agency, not per analyst. The model divides by the hours figure rather than assuming a ratio.
Typical: 12 to 40 for an agency with two to five account managers
Hours per account per week
Routine review only: reading the account, checking spend and inventory, spotting what moved, and preparing the weekly report. Strategy calls, creative and onboarding are deliberately excluded.
Typical: 3 to 8, rising with catalogue size rather than with spend
Loaded hourly cost
Salary plus employer costs, tooling and a share of overhead, divided by productive hours. It is roughly 1.4 times the bare hourly salary for most agencies.
Typical: $50 to $95 for a mid-level Amazon account manager
What the 55% is actually made of
An absorption rate is a claim, and a claim you cannot itemise is a guess. Here is the itemisation — six agents, six pieces of the weekly routine, and what each one still hands back to a person.
Margin
Re-deriving true contribution per ASIN after fees, storage and returns
The task most often skipped entirely, because it cannot be read off any single report.
Spend
Search-term and placement review, bid ceilings against break-even
Absorbed almost completely. The judgement calls escalate; the sweep does not.
Supply
Days-of-cover checks, restock dates, inbound reconciliation
Absorbed, except the purchase decision itself, which stays with a human.
Recovery
Ledger reconciliation and reimbursement case preparation
Drafted, never filed. Filing needs your approval on every account.
Health
Suppression and listing-status monitoring across the catalogue
Absorbed. Content fixes are drafted and wait for a person.
Client
Weekly report assembly, per brand, white-labelled
Drafted and branded. The commentary and the send stay yours.
What this model does not claim
What the model does not claim: that the freed hours convert to revenue automatically. They convert to capacity. Whether that becomes more clients, deeper work on existing ones, or a shorter week is a decision about your agency, not about the software.
It also assumes routine review is genuinely routine. If your analysts spend their five hours on strategy rather than on reading dashboards, the 55% is too high for you and the honest answer is a smaller number. The way to find your own figure is the Decision Audit: seven days of your real account, with the decisions counted, replaces every assumption on this page.
The 55% absorption rate is our own modelling assumption, not a measured average across customers. It is stated here so you can argue with it.
Questions agencies ask about the model
Including the one worth asking first, which is where the 55% comes from.
Does this replace my account managers?
No, and an agency that buys it for that reason will be disappointed. It removes the part of the week that is reading and reconciling, which is the part that scales badly and that nobody was hired to do. The judgement, the client relationship and every client-facing send stay with your people.
Where does the 55% come from?
From itemising the routine review tasks above and estimating how much of each an agent can complete unattended. It is a modelling assumption, not a measured customer average, and it is the number to challenge first if the output looks too good.
What about clients who want a human on every decision?
Keep those accounts in Recommend, or in Watch. The hours saved drop, because a person still reviews every staged decision, but the reading and reconciling time still goes — and that was always the larger half.
How long before the capacity is real?
Coverage across all seven agents takes about a week of account history. The reporting time goes in the first week; the review time goes as your team starts trusting the staged decisions, which in practice is the fortnight most agencies spend in Watch.
From the blog
What the freed hours were being spent on, and how much rope to give the agents once they are free.
Connect one account, read-only. See seven days of what the agents would have caught.
That is the Decision Audit. No configuration, no write access, no call required to start.
Manifest — what lands in your inbox
01Every decision the seven agents would have surfaced, timestamped.
02The money attached to the ones you missed.
03The reasoning chain for each, so you can check our work.
04A read-only audit log of everything we touched.