Decide which system owns which number
At 2:05 p.m. a customer buys the last medium in a linen shirt at the register. At 2:09 another customer buys the same shirt on your website, because the site still says one is left. Now someone has to write an apology, refund the order, and hope the customer comes back. Nobody made a mistake on purpose. Two systems simply disagreed about one number.
This guide is for owners and managers of shops that sell in a store and online and are tired of oversells, phantom stock, and weekend recounts. TwoQM develops connected systems, automation, AI integrations, and custom software that keep point-of-sale, ecommerce, and inventory records telling the same story.
"Inventory" is really several numbers, and each needs one owner:
On hand: units physically in the building. One system—usually the POS or a dedicated inventory system—owns this. Everything else reads from it.
Committed: units promised to open orders, holds, or pickup shelves, not yet gone.
Available to sell: on hand minus committed minus any buffer you choose. This is what the website should show.
When the website keeps its own on-hand count and the POS keeps another, they will drift. Pick the source of truth and make the other system a reader, not a second writer.
Why the numbers drift
Most mismatches come from a short list of process gaps:
Sync lag. Some integrations push changes every few minutes or in batches. Fast-moving or low-quantity items oversell in that window.
SKU and variant mismatches. A shirt that is one product with sizes in the POS and three separate products online will never sync cleanly.
One-sided adjustments. Someone marks a damaged item out in the back office but not in the system the website reads.
Returns restocked in only one place. An online return goes back on the shelf but never back into the count, or the reverse.
Receiving entered late. Product is on the floor Tuesday and in the system Friday.
Bundles and kits. A gift set sold online should reduce three components; often it reduces none.
The National Retail Federation describes shrink as comparing a retailer's book inventory to the physical inventory on hand, and notes it includes non-theft losses such as administrative errors, damages, and process mistakes—losses retailers can often control with operational improvements. Most of the gaps above are process problems of exactly that kind.
Protect the last few units
Low quantities cause the most oversells. A few rules help:
A buffer for thin stock. If on hand is two or fewer, show zero online, or show "check store availability" instead.
Reserve pickup orders immediately. A buy-online-pickup-in-store order should commit the unit the moment it is placed, not when someone walks to the shelf.
Hold display and damaged units separately. A floor sample is not sellable stock; give it a location or status that keeps it out of "available."
Make every adjustment explain itself
An adjustment without a reason is a mystery next month. Require a short reason code—damaged, count correction, vendor return, display, suspected loss—and the person's name. Over a season, those codes tell you whether you have a receiving problem, a returns problem, or a training problem. Spot-check the items that oversell most often every week instead of waiting for the annual count.
An original example
Here is an original illustration, not a TwoQM client and not a measured result. A two-person boutique carries about 1,200 SKUs and sells on its own website. The owner keeps a spreadsheet of items that oversold in the last 60 days. Most entries share three causes: sizes listed as separate products online, returns from the website restocked without a scan, and gift sets that never reduce their components. None of that requires new hardware. It requires one product structure, one returns step, and a kit rule.
What to automate, and where AI helps
Ordinary business automation should carry the routine: push sales and returns to the source of truth as they happen, hide an item online when available hits zero, alert a named person when a count goes negative, and send a daily list of SKUs whose online and in-store numbers disagree.
AI helps with messy catalog work, with a person reviewing every change. It can compare a vendor's new product sheet against your catalog and flag likely duplicates or mismatched variants, or group a month of free-text adjustment notes into patterns. It should not change counts or merge products on its own. AI integrations at TwoQM start with the task, the data it may touch, and who approves the result.
Connect what you have before building more
If your POS and ecommerce platform offer a supported integration, configure it around your source of truth before adding anything new; TwoQM can connect the systems you already use. Custom software fits when the gap is specific to your store: a mismatch dashboard staff check each morning, transfers between two locations, kit rules your platform does not support, or a receiving screen that updates both channels at once. You can also see how we approach technology for retail and ecommerce.
Tell us about your workflow
Pull your last ten oversells or stock mismatches and note what caused each one. Then tell us about your workflow. TwoQM is based in Brownsville, Texas, and works with businesses nationwide, in English and Spanish. We will tell you whether a configuration change, a connection between systems, or a small custom tool fits—and what it will not do. An inquiry is not a quote, and it is not a claim that a system is already built.