Automatic inventory updates one stock figure in every channel
A sale in the shop, a return from a marketplace, a delivery booked into the ERP, a correction after a stock count: each of those changes the number, and each reaches every channel within seconds. One source of truth, usually the ERP or a multichannel connector, and a workflow that pushes the figure out before somebody orders an item that left the shelf yesterday.
- 4 eventsthat is how many kinds of stock change one rule handles: a sale in any channel, a customer return, a delivery from a supplier and a correction after a stock count
- 14 daysthe statutory withdrawal window on distance selling across the EU, so returns are a constant stream of stock changes rather than an exception
- 2 to 6 weeksfrom the workshop to a sync running in production, depending on the number of channels and on whether your warehouse system has an API
The problem
Why does a customer buy something that is no longer on the shelf?
A shop on WooCommerce, Shopify or Magento, an Amazon account next to it, sometimes eBay and a price comparison feed, and alongside all of that wholesale or over the counter sales. Every one of those channels keeps its own counter. In plenty of companies the sync still works like this: somebody in the warehouse checks the shelf, types the number into the ERP, then fixes it on the marketplace, then in the shop. At a few dozen orders a day it holds together. In peak week and through December the gaps between systems grow by the hour.
The damage runs two ways. First: a customer pays for goods that physically are not there. Order cancelled, money refunded, a negative review on a marketplace where the account rating decides how often you get seen. The second one is quieter and costs the same: a delivery sits booked into the warehouse while the shop still shows zero, because nobody got round to entering the goods receipt. The stock is there, the sales are not.
In a study of more than 71,000 shoppers described in Harvard Business Review in 2004, 8.3% of the products people looked for were out of stock on average, and 31% of the shoppers who hit a gap bought elsewhere. That study looked at a physical shelf, but the mechanism online is the same: a customer who sees no stock does not wait. Online the gap is often artificial. The goods are in the warehouse and the channel shows zero, because the number never reached it.
- An order cancelled after payment, because the last unit sold twice
- A negative review over something neither the courier nor the product caused
- Stock sitting in the warehouse after a delivery while the channel still shows zero
- An extra person at a keyboard through peak season, retyping numbers between systems
Why now
What changed, so that this can be joined up in weeks instead of a quarter?
Every channel now has an interface that hands over a stock change within a second of the event: WooCommerce, Shopify, Magento and PrestaShop on the shop side, marketplace APIs and multichannel connectors such as BaseLinker on the other. The awkward class is the ERP that runs on a server in your own office instead of in the cloud, the kind Subiekt GT or Comarch ERP represent. There is no cloud endpoint to call, so stock is read through the product's own object layer or straight from its database, by a connector running inside your network, and that connector sets the refresh rhythm.
An automation platform, n8n or Make, picks the event up from the ERP or the WMS and pushes the new figure to every channel within seconds. In companies with no open API we build the bridge at the level of a CSV export or a Google Sheet that a script reads on a fixed cycle. That works with older, closed warehouse systems too, the ones that resist every other kind of integration.
The third change is the AI agent. A rule will carry a number from system to system, but it will not answer a warehouse manager asking which products sat below the minimum threshold, or where the warehouse and the marketplace drifted apart this week. BOND, our company AI agent, answers that in a plain sentence out of your own data in 5 to 10 seconds, with no logging into three systems and no comparing numbers by hand.
What changes
What does the warehouse get, and what does sales get?
One stock figure in every channel
A sale in any channel takes a unit off a shared counter, and the new number reaches the shop, the marketplaces and the rest of the channels within seconds. The source of truth is the ERP, the WMS or the multichannel connector, and which one is settled in the workshop.
Four kinds of event in one rule: sale, return, delivery, correction after a stock count
The last unit sells once
Peak week and December do not pull the numbers apart, because nobody is retyping them under time pressure. An order for goods that do not exist never gets created, so there is nothing to cancel and nothing to explain to the customer.
Updates in seconds from the event, and on systems with no API on a cycle you set yourself
A return goes back on stock with no manual step
A return booked in at the warehouse and judged fit to sell raises stock in the ERP and in every channel. With a 14-day statutory withdrawal window on distance selling across the EU, that is a daily stream of events, not something to handle when there is a free moment.
A return sits in the rule alongside a sale, not on a list of things to retype on Friday
A delivery is sellable on the day it arrives
A goods receipt note in the ERP raises availability in the shop and on the marketplaces without waiting for somebody to copy the figures over. Stock starts selling the same day it came through the door.
The goods receipt is the workflow trigger, not a reminder for a member of staff
An alert on a gap, not a complaint after the fact
The workflow compares stock between systems and reports the difference on Slack, in Teams or by email before a customer sees it. Error monitoring and alerts go into every rollout, not on request.
An alert naming the product, the channel and the size of the gap, sent to a named person
Questions about stock asked in a plain sentence
Which products are below the minimum threshold, what will run out before the next delivery, where did the warehouse and the marketplace drift apart this week. The AI agent answers from the ERP, the connector and your spreadsheets in 5 to 10 seconds.
An answer from your own data, with no logging into three systems and no export to a spreadsheet
The proof
Do we run a shop on this data ourselves?
Yes. Our founders' shop runs on PrestaShop wired into a multichannel connector and an ERP, and 16 people use an AI agent on that data, the warehouse included. The measured number we have comes from advertising, not from stock: campaigns in that shop are steered by a rule built on sales and returns data, and the same data layer cut 302 products out of them in half a year, products that stopped earning once returns were counted. That number is not a measurement of stock drift or of cancelled orders. We show it because it is the only measured result from that shop we are allowed to publish, and we show it together with what it does NOT prove.
Built for ourselves. These are not client deployments but implementations inside the founders' own companies. We say so plainly, because proof you cannot verify is not proof.
SurfBottheir agentSurf Inc., premium clothing brand, an online shop and two physical stores
- 16
- people from the company work with the agent
- 21
- systems wired in permanently
- 89
- processes run without a human
- 323
- thousand lines of the agent’s code
The agent is wired into the order system, the ERP, the shop, the data warehouse and the ad platforms. Launched in March 2026, fully ramped in May. 16 people use it, the warehouse and the store staff among them.
8 of 21 systems it works in
PrestaShop- BaseLinker
- Comarch Optima
BigQuery
Google Ads
Meta Ads
GA4
Zalando
- Before
- From February to August 2025 every unit of ad budget came back as 3.53 in online revenue, and advertising ate 28.3% of that revenue.
- After
- Across the same months in 2026 the figure is 4.19, and advertising is down to 23.9% of revenue. The ad budget is 6.9% lower on sales 10.3% higher.
Return on ad spend 3.53 to 4.19, 18.5% higher, on a budget 6.9% lower
And in your company
What if our warehouse works differently?
Four sentences we hear most often, and what we answer to each.
Our ERP has no API, it is an old system.
Then we build the bridge at the level of a CSV export or a Google Sheet that a script reads on a fixed cycle, anything from a few minutes to an hour, depending on how fast stock moves at your company. The update is not in seconds, but it happens without a person and without retyping. We set the cycle in the workshop and write it into the document.
We already run a multichannel connector, it syncs stock for us.
Yes, between the shop and the marketplaces, and it does that well. The drift starts elsewhere: between the connector and the ERP or the WMS, on deliveries, returns, wholesale and over the counter sales, none of which the connector sees. If the workshop shows that all of your sales already run through it, we say there is nothing here to automate.
We also sell in store and wholesale, and none of that sits in any system.
That is the first subject in the workshop: settling the source of truth. Over the counter and B2B sales have to reach the ERP, on a receipt or a goods issue note, and only then travel to the online channels. Without that no sync helps, and we say so plainly before we build anything.
We are afraid an automation will zero out our marketplace listings mid season.
That is why the workflow starts in parallel, on copies of your data and on sample orders, not straight onto the live warehouse. An alert halts the update when a stock change crosses a threshold you set, say a drop from 40 units to zero in one move. We plan the switch outside the peak, not in peak week.
- Shop platforms
- WooCommerce, Shopify, Magento, PrestaShop, BigCommerce
- Marketplaces and connectors
- Amazon, eBay, marketplace APIs, BaseLinker
- ERP and warehouse
- SAP Business One, enova365, Fakturownia, on-premise ERP read through its object layer or database, WMS systems
- Systems with no API
- CSV export, a Google Sheet read by a script on a cycle
- Notifications
- Slack, Microsoft Teams, email
- Automation layer
- n8n, Make, custom integrations over the API
The events that change stock, and where we pick up the signal
Seven events that a stock sync is usually made of. Which of them go into the first rollout is settled in the workshop, because the order depends on where your numbers drift apart most often.
| Event | Where the signal comes from | What happens in the channels |
|---|---|---|
| Sale in the shop | a webhook from WooCommerce, Shopify, Magento, PrestaShop or another shop platform | stock drops by one unit in the ERP and on the marketplaces within seconds |
| Sale on a marketplace | the marketplace API or a multichannel connector | stock drops by one unit in the ERP and in the shop |
| Customer return | the return booked in at the ERP, the WMS or the connector | the unit goes back on sale in every channel, if it is fit to sell |
| Delivery from a supplier | a goods receipt note in the ERP | availability rises in the shop and on the marketplaces on the day it arrives |
| Correction after a stock count | the stock count document in the ERP | the channels get the new figure, the difference lands in a discrepancy report |
| Over the counter or B2B sale | a receipt or a goods issue note in the ERP | online stock drops by whatever left the shop floor or went to a wholesaler |
| A system with no API | a CSV export or a Google Sheet read by a script | an update on a fixed cycle, from a few minutes to an hour |
We start with the events that pull your stock apart most often, usually sales and returns. Deliveries, stock counts and offline sales come in a second step, once the first rule is running in production.
How we do it
Five steps from retyping by hand to one stock figure
Each step ends with something you can judge before you decide on the next one.
Workshop
We sit down with the warehouse and sales teams and map the path goods take from delivery to dispatch. We check how many channels you run, which ERP or WMS sits behind the warehouse, whether it has an API and where the gaps get created. We leave with the source of truth for stock settled.
Rule design
The list of events that change stock, the order they are handled in, the alert thresholds and the exceptions: a return that cannot be resold, stock reserved for B2B, a bundle sold out of several line items. The tool choice: a multichannel connector, n8n, Make or an integration over the API into your ERP. Always the simplest path to the goal, not the most elaborate one.
Build and testing on copies
We build the workflow and test it on copies of your data and on sample orders, before anything touches the live warehouse. We replay situations from your last season: two sales of the last unit in the same minute, a return 14 days after the order, a delivery booked in after hours.
Switch and training
After the tests we go to production with error monitoring and alerts on gaps. The warehouse and sales teams get the documentation and learn to react to alerts, change thresholds and add a channel. We do not leave a black box that only the builder understands.
Measurement after 30 and 90 days
We compare the number of gaps, the orders cancelled for lack of stock and the hours spent retyping against the period before the rollout. We tune the alert thresholds to what actually comes in, rather than to what we predicted in the workshop.
Before and after
How the day changes for a warehouse manager
Before
- Every morning they compare stock in the ERP with stock on the marketplaces and fix by hand whatever has drifted
- They hear about a gap from customer service, once an order has to be cancelled and the money refunded
- A delivery is booked into the ERP and the shop still shows zero, because nobody got round to entering it
- In peak week they drop everything else and retype numbers between systems until the evening
After
- Stock corrects itself after every sale, return and delivery, and they only watch the alerts
- They know about a gap from a notification, before anybody orders the item that is missing
- Goods from a receipt note sell in every channel on the day they arrive
- Through the season they do the work that needs a person: packing, planning deliveries, handling exceptions
The first sync is visible on the day of the switch. We compare the number of gaps and cancelled orders after 30 and 90 days against the period before the rollout.
For the person signing this off
What the shop owner or the operations director gets out of it
Cancelled orders and negative reviews stop being a Monday meeting subject, because the number the customer sees is the number the warehouse sees. Peak season stops needing an extra person at a keyboard, and you are not explaining to a customer why they paid for something that was not there.
The second thing matters more when you plan. You know how many orders you cancelled for lack of stock and how much stock sat invisible to sales. Those are numbers nobody at your company collects today, and they decide whether to add another sales channel and how much to order before the season.
- One source of truth for stock, written down and agreed, instead of three counters
- An alert on a gap instead of a customer complaint and a negative review
- Tested on copies of your data and rolled out in parallel, the live warehouse never stops
- Documentation and training, so the team changes thresholds and adds channels with no ticket to us
What it costs
The workshop has a published price, the rollout is quoted after it
The process workshop costs EUR 3,500 net per week. On a stock sync one week is usually enough. Two, when there are more than three channels or part of the sales happens outside the system and has to be brought into the ERP first. We quote the rollout after the workshop, once it is clear how many channels are in play, which ERP sits behind them and whether it has an API.
You pay the cost of running the workflow directly to the tool vendor, n8n, Make or the connector, or you pay nothing at all when it runs on your own infrastructure. No commission from us. The same goes for language models when you add the AI agent: model queries cost the vendor EUR 0.005 to 0.07 each.
Process workshop
EUR 3,500 net
per week, on a stock sync usually one to two weeks
A self-contained piece of work. It ends with a map of how goods move, the source of truth for stock settled, and a plan you can execute without us.
- Map of how goods move from delivery to dispatch, and of the points where stock drifts apart
- The source of truth settled and the list of events that change stock
- Tool recommendation: a multichannel connector, n8n, Make or the API into your ERP
- Metrics to measure after 30 days: gaps, cancellations, hours spent retyping
- Most popular
Workshop and sync rollout
Quoted after the workshop
scope follows from the number of channels and from whether the ERP has an API
The option most companies choose. A working stock sync within 2 to 6 weeks of the workshop, tested on copies of your own data.
- Everything in the workshop
- A workflow in n8n, Make or over the API, wired into the ERP, the shop, the connector and the marketplaces
- Four events handled: sale, return, delivery, correction after a stock count
- A CSV or Google Sheets bridge for systems with no API
- Error monitoring, alerts on gaps, the switch after testing
- Training for the warehouse and sales teams on reacting to alerts
Sync and an AI agent for the warehouse
Quoted after the workshop
for teams that ask about stock a dozen times a day
Everything in the rollout plus BOND, which answers questions about stock and gaps in a plain sentence, in 5 to 10 seconds.
- Everything in the sync rollout
- AI agent wired into the ERP, the connector and your spreadsheets: stock below threshold, gaps, turnover
- A shortage forecast from turnover and delivery dates, before stock runs out on the shelf
- Access for the warehouse, sales and the owner, with no login to three systems
If the workshop shows that your multichannel connector already handles every channel and the drift comes from one wholesale line kept by hand, we say so and show you how to bring it into the system. The workshop ends with a document that stands without us. The work then ends at the workshop, not at a rollout you do not need.
Frequently asked questions
What is automatic inventory updating?
It is a workflow that picks up every stock change, a sale in any channel, a customer return, a delivery from a supplier and a correction after a stock count, and pushes the new figure to the shop, the marketplaces, the multichannel connector and the ERP within seconds. One place is the source of truth and the rest take the number from it. A person steps in only on an alert about a gap.
How long does a stock sync take to implement?
One week of workshop and 2 to 6 weeks of build, depending on the number of channels and on whether your ERP has an API. A shop plus one marketplace through a connector is the lower bound. Three channels, an ERP with no API and over the counter sales that first have to reach the system is the upper one. Before a peak season we count backwards, so that the switch falls outside the peak.
We already run a multichannel connector. Is that not enough?
Between the shop and the marketplaces it usually is, and we will tell you so. The drift starts between the connector and the ERP or the WMS, on deliveries, returns, wholesale and over the counter sales, none of which the connector sees. The automation closes that loop: an event in the ERP reaches the connector, and a sale from the connector goes back to the ERP without retyping.
What if our ERP or warehouse system has no API?
We build the bridge at the level of a CSV export or a Google Sheet that a script reads on a fixed cycle, from a few minutes to an hour. It works with older, closed systems too. The update is not in seconds then, but it happens without a person and without retyping. We pick the cycle in the workshop to match how fast stock moves at your company.
What happens when stock drifts apart in spite of the automation?
The workflow compares stock between systems and on a difference sends an alert to Slack, to Teams or by email, naming the product, the channel and the size of the gap. The source of truth settled in the workshop wins, usually the ERP or the WMS. On a change larger than the agreed threshold the update stops and waits for a person, so that an error in one system cannot zero out stock in all of them.
How do you handle returns in a stock sync?
A return booked in at the warehouse is an event like a sale, only with a plus sign. A unit judged fit to sell goes back on stock in the ERP and in every channel, a damaged one goes onto a separate stock outside sales. With a 14-day statutory withdrawal window on distance selling across the EU, that is a daily stream of events, which is why returns go into the first rollout, not the second.
Does the sync cover over the counter and wholesale sales?
Yes, as long as those sales reach the ERP on a receipt or a goods issue note. Online stock then drops by whatever left the shop floor or went to a wholesaler. If over the counter sales sit in no system at all today, that is the first subject in the workshop, because without a source of truth no sync will help you.
How does the AI agent work with stock levels?
BOND is wired into the ERP, the multichannel connector and your spreadsheets, and answers in a plain sentence in 5 to 10 seconds: which products are below the minimum threshold, what will run out before the next delivery, where the warehouse and the marketplace drifted apart this week. It answers from your data, not from the model. The agent takes around 2 months to roll out and we add it after the sync, not instead of it.
Let us find where your stock drifts apart today
The free consultation takes 30 minutes. You tell us how many channels you sell through, which ERP or WMS sits behind the warehouse and where the numbers drift most often between the shop, the marketplaces and the shelf. We show you which events can be joined up straight away and which need the sales brought into a system first. You do not have to give access to anything at this stage. You write to Pawel, the founder, not to a reception desk.
We run at most two rollouts at a time, so that someone is actually sitting on each of them.
Book a free consultationPS. If you cancelled an order last season because the goods were not on the shelf, or the goods were there and the channel showed zero, this is exactly that problem. A number retyped by hand between three systems always drifts apart, the only question is when. The workshop starts with settling one source of truth.