The stock number everyone quotes and nobody trusts

Ask three people in a distribution business how many of an item are in stock and you can get three answers. The system says twelve. The shelf says nine. Sales already promised eight to a customer this morning. Everyone quotes the number with total confidence, and nobody actually trusts it, which is why there’s always someone who walks out to the racks to check before committing to anything that matters.

That gap between the number on the screen and the number on the shelf is where warehousing quietly bleeds money. You oversell stock you don’t have and disappoint a customer. You reorder stock you’re already sitting on and tie up cash. You count something as an asset that’s been damaged in a back corner for a year. Every downstream decision, sell it, promise it, buy it, cost it, runs on a figure that’s a little bit wrong, and the errors compound. Stock data looks like admin right up until you trace a lost customer or a cash crunch back to a number nobody could rely on.

Here’s how it plays out on an ordinary Tuesday. A sales rep at an ag-parts distributor outside Toowoomba takes a call from a good customer who needs eight of a particular pump before seeding. The system shows nine on hand, so the rep says yes on the spot, which is exactly what you want a rep to do. Except two of the nine were allocated to another order yesterday and one has been sitting damaged on a bottom rack since autumn. Now someone’s ringing the customer back to walk back a promise, purchasing is paying freight to expedite replacements, and the rep has learned to phone the warehouse before quoting, which slows every deal from here on. One wrong number, four people’s time, and a dent in the relationship that took years to build.

Decide who owns each fact before you connect anything

Most distribution businesses are quietly running several versions of stock truth at once. The warehouse has a physical count. Sales has a promise it made a customer. Purchasing has an inbound order and a lead time. Finance has a cost and a valuation. Each is right about its own slice and none of them agree, so the number everyone acts on is whatever the last person to look decided to believe.

Before you add any AI, and before you connect a single system, settle the boring question that fixes all of this: which system owns each fact. Quantity on hand, location, committed stock, reorder point, supplier, cost, customer allocation, each needs one authoritative home, with the others reading from it rather than keeping their own copy. Skip that step and integration doesn’t fix the confusion, it just spreads the wrong number to more places faster. Single source of truth isn’t a slogan here. It’s the difference between a connected system and a synchronised mess.

The private buffers are where the cash went

There’s a second cost hiding behind the untrusted number, and it’s usually bigger than the overselling. When nobody believes the figure, everyone builds their own insurance against it. Purchasing pads the safety stock, because being caught short once was painful. Sales quotes longer lead times than it needs to, because walking back a promise once was worse. The warehouse keeps an unofficial shelf of don’t-touch stock for the customers who shout loudest. None of these people are doing anything wrong. They’re each rationally protecting themselves from a number they’ve been burned by.

Add the buffers up, though, and you’re carrying weeks of extra stock across hundreds of lines, which is cash sitting on racks doing nothing, plus the shed space to hold it. For a mid-sized distributor that padding can quietly run into six figures of working capital. The fix isn’t a memo telling people to trust the system. It’s making the number right, at which point the buffers shrink on their own because nobody needs the insurance any more.

Integration replaces the retyping, not the thinking

Inventory has to talk to ecommerce, accounting, CRM, purchasing, freight and the warehouse floor, and when it doesn’t, staff spend their day exporting, copying, reconciling and fixing what didn’t match. That handling is slow, it’s where fresh errors get introduced, and it’s almost entirely avoidable.

API integration takes it away. A sales order reserves the stock the moment it’s placed. A supplier update shifts the expected arrival date on its own. A picked order updates finance and lets the customer know it’s on the way. A stock adjustment leaves a review record behind it. Nobody rekeys any of it, which means the number stops drifting every time a human touches it, and the figure on the screen starts matching the shelf because there’s only one figure. If you sell online as well, the store has to read from the same source, because an oversell to a stranger on the website costs you a refund and a review instead of an awkward phone call.

Put AI on the exceptions, not the arithmetic

The basic stock logic, reorder points, validations, approvals, permissions, belongs in ordinary software rules, where it’s predictable and you can trust it to behave the same way every time. That’s not where AI helps, and handing it those jobs just makes reliable things flaky.

Where AI is useful is around the exceptions and the mess. It can summarise a tangled supplier email thread, classify a stock issue, pull the details off an invoice, flag a movement that looks wrong, or assemble the daily exception report. Picture a supervisor starting the shift with a short list already built: orders at risk, discrepancies that need a look, supplier messages mentioning a delay, products moving outside their usual pattern. The supervisor still makes every call. They just start from a clear queue instead of digging through six systems to work out where the day’s problems are hiding.

Reports that drive a decision

Inventory reporting should help someone act, not just fill a screen. Which products are tying up your cash. Which ones sell fine but generate too many returns or service issues. Which suppliers keep missing dates. Which customers always need urgent handling that costs you. Which locations throw up the most picking errors. Answering any of that means pulling from several systems at once, which is why data analytics matters here as much as the warehouse screen does.

Count twenty SKUs before you believe anyone

If you want to know how bad your version of this is, the test costs an afternoon. Pick twenty SKUs, a mix of fast movers, slow movers and the awkward stuff that lives in more than one location. Walk the racks and count them, then compare to the system. Most businesses that haven’t invested in stock accuracy find somewhere between two and six of the twenty are wrong, and the interesting part isn’t the count, it’s the why. Trace each discrepancy back: a receiving error, damage nobody recorded, a transfer keyed twice, a pick against the wrong line, a return that never made it back on. Each cause points at a different fix, and none of them is a bigger stocktake.

Then, when you’re evaluating software or a builder, lead with the ownership question. Ask which system will own quantity on hand, and what happens when the website and the warehouse disagree at the same moment. Ask to see the audit trail on a stock adjustment. A vendor who answers those directly is worth another meeting. One who pivots to the dashboard demo hasn’t solved the problem you have.

Start with the question your staff keep working out by hand: what can we safely promise a customer today, what should we actually buy, what’s ageing in the racks, what’s missing, why did that order fall over. Connect the sources that one question needs, get them agreeing on a single number, and build out from there. For a Queensland warehouse or distributor, getting inventory right isn’t back-office housekeeping. It’s what protects your customer service, your purchasing discipline and your margin all at once. If there’s a stock figure in your business you don’t quite believe, tell us where it goes wrong and we’ll work out which system should own it.