Circes connects to Seller Central and the Amazon Ads API, reads the account daily, and returns a queue of recommendations for review. Below is what it covers, what it proposes, and what it will never do on its own.
Bid and budget changes are proposed against each SKU’s own profitability rather than against a single account-wide target.
Terms that convert are proposed for promotion; terms that spend without converting are proposed as negatives.
A guided flow builds a campaign structure for a chosen SKU and objective — launch, scale, profitability, brand defense or discovery. The proposed structure is reviewed before anything is created in the ad account.
Spend, ad sales, organic sales and the standard efficiency metrics, with the paid/organic revenue split and daily trends.
Circes flags campaigns bidding on terms the brand already ranks for organically.
Term-level performance for the account’s own catalogue, used for campaign structure and for the listing work below.
On-hand, reserved and inbound units across FBA and AWD, with days of cover for each SKU.
Each SKU gets a projected stockout date, an order-by date and a suggested order quantity, so the purchase order goes in before the lead time runs out.
Purchase orders are recorded with their cost inputs, so margin reporting reflects what stock actually cost you.
Advertising and inventory are read together, so a bid increase on a SKU that cannot cover the demand is held back rather than queued.
Returns are tracked per SKU and netted into contribution margin, so profitability reflects what was kept rather than what shipped.
SKUs whose cover has fallen below their lead time are raised on the daily brief with the date they run out.
Circes audits each listing in the catalogue against the account’s own search-term performance rather than generic keyword tools. It reports what is converting, which terms the copy is underweighting, and the specific edits that would close the gap.
Terms and segments converting above the catalogue average, so a rewrite does not damage what is working.
High-volume or high-converting terms that the title, bullets and backend keywords under-serve.
Title, bullet, image and attribute problems are graded so the highest-value fix is obvious rather than buried in a list of everything that could be improved.
Circes drafts the proposed title, bullet and keyword changes. You read them side by side with the current copy and accept, edit or reject. Nothing is published to Amazon without a person approving it.
The search intent a SKU attracts, compared against the intent the copy targets — which is often where a conversion problem starts.
Best-seller rank tracked per SKU and category, used to tell a listing problem apart from a demand shift.
This is advisory, and deliberately so. Circes reviews price against landed cost, fee structure, conversion rate and competitive position, and tells you where a price is leaving margin behind or costing conversion. Circes is not a repricer. It does not change your prices automatically, and it holds no automated pricing rules against your catalogue. Every price change is made by you.
Contribution per unit after Amazon fees, landed cost and cost of goods, kept current as those inputs change.
How price moves have affected conversion and volume for your catalogue, and what that means for advertising efficiency.
When a fee or cost change erodes a SKU’s margin, Circes flags it so spend is not left running against out-of-date assumptions.
Circes does not compete for the Buy Box on price, does not run scheduled repricing, and does not hold min/max price rules. Pricing recommendations appear in the queue for a person to act on.
What sold, what it earned after costs, an account health score combining margin, ad efficiency, inventory and sync health, and the decisions waiting on you.
Revenue, fees, COGS, advertising and operating expenses resolved into profit, with cash flow and a forecast against goals.
Applied recommendations are measured against what happened afterwards, so an estimate can be checked against its result.
Questions answered from the account’s own data, with the working shown and the figures checked against the source before they are reported.
Authorised through Amazon’s own consent flow. Sync status and history are visible, and cost data can be imported where Amazon does not provide it.
Multiple brands under one login, with per-manager permissions and reporting that goes out under your own brand.
Every function above produces a proposal, not an action. Proposals are checked against limits you set, then queued for a person to approve, edit or dismiss.
An account runs in local only, where nothing is sent to Amazon at all, or in push to Amazon, where approved changes are applied through the API.
Configurable limits on how much can change and how often. A proposal that breaches your limits is discarded before it reaches the queue.
Profitability, efficiency and stock-cover thresholds, plus terms you never want bid on. Set per account and enforced as a hard stop rather than a preference.
Each recommendation carries the data it was derived from and how it was verified. Every action taken is logged with its author and timestamp.
Approved Amazon SP-API Developer · Member, Amazon Ads Partner Network