A fractional AI ops department for e-commerce, SKUs reconciled, inventory honest, close on time.

Every DTC and marketplace brand between $5M and $50M annual GMV shows up to the audit with the same problem shape. Shopify carries the DTC storefront. Amazon Seller Central carries the Prime audience. Lazada or Shopee carries the SEA marketplace. Faire carries the wholesale buyer base. Each channel has its own SKU schema, its own inventory feed, its own returns policy, its own settlement timing. The ops manager is one person trying to keep all four reconciled against a single inventory truth and a single finance close. The close is always three to five days late. AOV reporting by channel is calculated three different ways. Returns are reconciled in a Monday morning panic that bleeds into Tuesday.

Behind the channels sits the fulfillment layer. The 3PL holds the bulk inventory. The brand warehouse holds the fast-mover SKUs. The drop-ship supplier in Shenzhen carries the long-tail. Multi-hub routing decides which warehouse fulfills which order based on stock position, shipping cost, and SLA. The router is a spreadsheet maintained by the ops manager. When the spreadsheet breaks (and it breaks every six weeks when a new SKU lands), the routing goes wrong, the 3PL ships out of stock, customer service gets the angry tickets, and the ops manager spends Saturday rebuilding the spreadsheet.

Behind the fulfillment layer sits the finance close. Supplier invoices arrive as PDFs in five different inboxes. Marketplace settlement reports land on different schedules with different reason codes for returns, refunds, FBA fees, advertising deductions, and FX adjustments. The accountant works from a spreadsheet that joins Shopify revenue to Amazon settlement to Lazada payouts to Faire wholesale invoices to supplier purchase orders. The close runs three to five days late every month. The CFO sees the gross margin number in week three instead of week one. The board hears about the SKU that lost margin two months ago.

The default fix is the four-hire ladder. An inventory analyst, a marketplace ops associate, a finance ops associate, and a senior ops lead by next year. Year-one loaded cost north of $300K. Most of them spend their first quarter learning the Shopify schema and the Amazon settlement file format. The function does not improve. The other path is to take the e-commerce ops function and run it as a fractional AI Ops Department tuned for e-commerce data shapes from day one. Shopify, Amazon, Lazada, Faire, 3PL feeds, brand warehouse, supplier invoice processing, finance close. Same scope as four hires, single monthly retainer.

The reason e-commerce ops is harder than it looks is that every channel models a SKU differently and the underlying inventory truth has to reconcile to all of them. Shopify uses one variant schema. Amazon uses ASINs with parent-child relationships. Lazada and Shopee model bundles separately. Faire treats wholesale packs as the atomic unit. The same product (a single moisturizer in a 50ml bottle, sold individually on Shopify and in packs of six on Faire) lives as four different SKUs across four channels. The 3PL knows it as one SKU with a case pack. The brand warehouse knows it as one SKU with a different case pack because they receive it in different cartons from the contract manufacturer.

When the ops manager publishes inventory to all four channels at once, the math has to flow correctly in both directions. The atomic unit in the warehouse times the case pack on the channel times the bundle multiplier on the marketplace equals the available-to-sell number the channel can show. When a Shopify order ships out, the inventory burn on the warehouse side needs to flow back to Amazon and Lazada and Faire before the next refresh window so two channels do not oversell the last six units. The current state of this in most $5M to $50M brands is a spreadsheet, two Zapier flows that break monthly, and an ops manager who checks inventory three times a day and ships an apology email when the math goes wrong.

A fractional AI Ops Department reads all four channel schemas, normalizes them against a single canonical SKU truth, and reconciles inventory in both directions on a continuous schedule. Shopify, Amazon, Lazada, Faire, and any new channel you add (TikTok Shop, Walmart Marketplace, Etsy for the long-tail) lands on the same schema with the same case pack and bundle logic. When the warehouse burns a unit, every channel sees the new available-to-sell number before the next refresh cycle. Returns flow back into the canonical inventory truth on the right SKU and the right case pack. The oversell apology email stops being a weekly event.

The compounding effect is what funds the next channel expansion. A $10M GMV brand that opens TikTok Shop while still running the inventory spreadsheet manually will lose two weeks of an ops manager getting the SKU schema right and another two weeks of customer service tickets when the schema breaks. The same brand running on a SKU-normalized ops layer adds the channel in three days, with the inventory math correct from the first order. That is the operational maturity the brand is supposed to have at the next stage and almost none of them have at the right time.

The fulfillment layer is where mid-stage e-commerce brands quietly leak margin. A growing DTC brand at $10M GMV runs three fulfillment hubs by default. The 3PL on the East Coast holds the bulk inventory and serves the Prime-equivalent SLA on Shopify. The brand warehouse on the West Coast holds the fast-mover SKUs the founder personally hand-packed in year one and never moved out because the unit economics work. The drop-ship supplier in Shenzhen handles the long-tail SKUs that do not justify a domestic stock position. Each order needs to route to the hub that minimizes shipping cost while hitting the SLA window the channel promised.

The routing logic in most $5M to $50M brands is a Google Sheet that maps SKU plus zip code plus channel SLA to a hub assignment. The ops manager maintains it. When a new SKU lands the manager spends an hour adding it to the sheet. When a hub stock position shifts the manager updates the sheet again. When a marketplace introduces a new SLA window (FBA two-day, Lazada same-day in Singapore, Shopee next-day in Kuala Lumpur) the entire sheet gets a refactor. Every six weeks the sheet breaks in a way that ships an order out of stock from the wrong hub and customer service spends Wednesday in apology mode.

A fractional AI Ops Department holds the routing logic as a continuously reconciled function instead of a spreadsheet. The agents read live inventory positions from every hub, live SLA expectations from every channel, live shipping cost tables from every carrier, and route each order to the optimal hub at the moment of order placement. When a new SKU lands the routing logic picks it up from the canonical SKU truth and the next refresh cycle. When a hub goes out of stock the router rebalances across the other two without an ops manager touching the sheet. When the marketplace introduces a new SLA window the agents update the routing rules from the marketplace API change log.

The compounding effect for a $10M to $50M brand is somewhere between two and four points of contribution margin recovered. The margin sits in three places. Lower shipping costs because every order ships from the optimal hub. Fewer overstock writedowns because the router rebalances before a hub overstocks. Fewer customer service tickets because the SLA window gets hit consistently. Two to four points on $20M GMV is $400K to $800K of margin a year, which pays for the retainer many times over and leaves a real budget for the next product launch.

The ops manager at a $10M to $50M DTC brand is one of the most consequential people in the business and one of the most under-supported. They own the channel mix, the fulfillment routing, the close, the inventory truth, the supplier relationships, and the customer experience downstream of all of it. They are the person customer service escalates to, the person finance escalates to, the person the founder calls at 9pm on a Tuesday when the FBA settlement report does not match. In most brands they are also the person spending Saturday morning rebuilding a spreadsheet because a Shopify variant got remapped and the Amazon ASIN parent-child broke.

Get the spreadsheet out of their life and the same person can operate a brand twice the size. The same ops manager who was holding a $10M GMV brand together by hand can run the ops function at a $30M GMV brand without breaking sweat, because the canonical SKU truth, the multi-hub routing, the returns reconciliation, the finance close, and the supplier invoice processing are running in the background. Their time goes to the work the brand actually needs from them. Channel strategy, supplier renegotiation, SKU rationalization, the launch of the next product line, the conversation with the founder about whether the next geo expansion is worth the inventory exposure.

The savings show up in three places. Ops manager retention at the senior level, which is the single most under-budgeted line item in mid-stage e-commerce. Faster close, which means faster decisions, which means faster reaction to a SKU losing margin. And the operational headroom to add the next channel or the next geo without hiring three people first, which is what funds the next stage of growth.

There is one more line that does not fit on a spreadsheet. The ops manager who stops working Saturdays stops drafting their resignation letter in their head. Senior operators leave funded brands when the work they signed up for gets crowded out by spreadsheet maintenance and oversell apology emails. Eighteen months in, the conversation starts. The Saturday spreadsheet rebuild is one of the loudest signals that conversation is on the calendar.