Amazon seller analytics tools fall into eight categories, and seven of them share the same gap: none of them reconcile the settlement. They measure rank, bids, price, reviews, stock, or estimated margin from a data source that is upstream of the money. The settlement, the two-week statement where Amazon lists every fee it charged and every refund it issued, is the only document that says what a sale was worth after Amazon took its share, and the tools that read it are in the accounting category. Pick tools by category with that gap in mind, and the eighth category, AI on reconciled data, makes sense as the layer that reads what the accounting tools produce.
Vendor names are left out on purpose. Amazon's own reports and tools are named where relevant.
What they measure: search volume estimates, keyword rank by ASIN, competing listings, listing quality scores, and in some cases estimated category sales.
Data source: scraped search results, sampled traffic models, and, for brand owners, Amazon's Brand Analytics dashboards (Search Query Performance, Top Search Terms), which Amazon's sell.amazon.com guide describes as query-level impression, click, and purchase shares.
What they cannot reconcile: anything after the click. A keyword suite can show that a listing ranks fourth for a query with high volume. It has no view of the referral fee, the fulfillment fee band, the return rate, or the landed cost of the product that ranks there, so it cannot say whether winning that query is worth it.
What they measure: spend, clicks, impressions, attributed sales, ACoS, and bid performance by campaign, ad group, keyword, and target.
Data source: the Amazon Ads console and its reports. advertising.amazon.com defines ACoS as ad spend divided by ad revenue, times 100, and states that ACoS has to sit below profit margin to hold profit.
What they cannot reconcile: the margin that ACoS is supposed to sit below. A PPC manager knows spend and attributed sales. It does not know the SKU's FIFO cost, its current fee band, or its return rate, so the "target ACoS" a seller types into it is a guess unless the books supplied it. It also cannot tie the ad invoice to the settlement period it affected.
What they measure: your price against competing offers on the same listing, featured offer share, and price change history.
Data source: the listing's offer data and your own Business Reports (featured offer percentage, sessions, conversion).
What they cannot reconcile: the floor. A repricer will hold a minimum price you set. It has no way to compute that minimum from landed cost plus current fees plus expected returns, so a floor set in January is wrong by June if a PO layer cost more or the SKU moved into a heavier fee band.
What they measure: review counts, star ratings, review velocity, seller feedback, and request-a-review activity.
Data source: listing pages, Seller Central's feedback and review request features, and the Voice of the Customer dashboard.
What they cannot reconcile: the cost of the problem the reviews describe. A tool can show a spike in one-star reviews mentioning "arrived broken." It cannot show the refunds, the reimbursements, and the fee reversals that spike produced in the settlement, or what the return shipping cost.
What they measure: days of cover, sell-through, forecast demand, reorder points, and recommended PO quantities.
Data source: Seller Central inventory reports, FBA inventory, sales history, and, if entered, supplier lead times. Amazon's own Inventory Performance Index and restock recommendations live in Seller Central.
What they cannot reconcile: what the inventory is worth on the books. A planner counts units. It does not carry FIFO layers with freight allocated, so the cash tied up in a SKU, the storage fee the units incur, and the aged-inventory surcharge they are approaching are outside its arithmetic unless a second system feeds them in.
What they measure: estimated profit by SKU, by day, and by period: sales less estimated fees, less ad spend, less a COGS figure the seller typed in.
Data source: order data and fee estimates from Amazon's APIs, ad spend from the ads console, and a manual COGS entry per SKU.
What they cannot reconcile: the estimate against the settlement, and the typed COGS against the actual PO. A dashboard's fee line is Amazon's estimated fee at order time, not the fee Amazon charged when it settled, and the two diverge on fee band changes, promotions, and reimbursements. The COGS is a single number per SKU, which is wrong the moment a second PO lands at a different price. Dashboards are fast and useful for direction; they are not the books.
What they measure: the settlement itself. Each transaction typed and posted: orders, refunds, referral fees, fulfillment fees, storage, reimbursements, adjustments, and the payout matched to the bank deposit. With inventory, FIFO cost per unit and landed cost allocation.
Data source: Amazon settlement reports, the accounting system (QuickBooks or Xero), and purchase orders.
What they cannot reconcile: nothing on the money side; that is their job. What they do not do is the upstream analytics: rank, bids, reviews, competitor prices. ConnectBooks sits here. The Amazon integration posts settlements at the transaction level, the accounting layer keeps the books on FIFO cost with landed freight, and the profit reports show contribution by SKU and by channel on settled numbers.
What they measure: whatever question the seller asks about profit, margin, ads, fees, inventory, and cash, answered from category seven's output: what changed, over which periods, on which products, why, and what to do next.
Data source: the reconciled ledger. Not a CSV paste, not a dashboard estimate.
What they cannot reconcile: anything category seven did not. An AI on reconciled data inherits every gap in the books. Missing landed cost, an untagged carrier bill, a settlement posted as one lump: each becomes a confident wrong answer. It also has no view of categories one through five, so it cannot tell you why a keyword rank fell or which competitor took the featured offer.
Illustrative numbers. A seller sells a bamboo cutting board at $24.99 in Home and Kitchen, where sell.amazon.com's pricing page lists the referral fee at 15 percent.
Six tools, and the five upstream ones are working from a product with a 27 percent margin that has a 21 percent margin. The repricer's $19.99 floor, set on the old numbers, now clears cost plus fees by $3.39 before ads (referral at 15 percent of $19.99 is $3.00, plus $5.75 fulfillment, plus $7.85 COGS, equals $16.60) and by $0.99 after the $2.40 of ad spend, before a single return. None of the five upstream tools can see the change, because none of them read the settlement.
Each category answers one question well and cannot answer the one next to it. The reconciled ledger is where the answers meet: it holds the settled fee, the FIFO cost, the refund, and the ad invoice against the SKU, which is why the comparison of Seller Central reports against an AI analyst puts the money questions on the accounting side. Crunch, the analytics AI inside ConnectBooks, reads that ledger and answers the seller's question in plain language: why margin on the cutting board fell six points, which products were profitable last quarter and are not now, what pausing a campaign did to profit. It cannot read the keyword rank or the repricer's floor. Bring those, ask the question with the settled margin in hand, and set the floor at the number the books support.
Running an e-commerce business comes with plenty of challenges, but ConnectBooks is here to make your life easier. With real-time insights, seamless integrations, and detailed tracking of your profitability and inventory, you can stay ahead of the game. Whether you’re selling on Amazon, Shopify, Walmart, TikTok or eBay, ConnectBooks helps you manage your finances with 100% accuracy and confidence, so you can focus on growing your business.
Ready to level up? Start making smarter, data-driven decisions every step of the way. Try ConnectBooks Free Today or Schedule a Demo
