An automated order-to-production workflow connecting sales orders, Fishbowl manufacturing and Reckon accounting for a New Zealand manufacturer

Quick answer: order-to-production automation means an incoming sales order raises the matching work order in Fishbowl against the right bill of materials, the financials post to Reckon reconciled, the freight books itself with the right carrier, and management watches it all on a live dashboard. Nobody retypes anything. This article walks through the full reference architecture we build from, stage by stage, for New Zealand manufacturers running Fishbowl and Reckon.

What's in this article

The Real Problem: Your Systems Are Fine, the Gaps Between Them Are Not

A typical New Zealand manufacturer running Fishbowl and Reckon has already made good software decisions. Fishbowl is genuinely strong at inventory, bills of materials and manufacturing work orders. Reckon is a solid accounting backbone, and the two officially integrate. The carriers all have booking portals that work.

The cost is not in any of those systems. It is in the handovers between them. An order arrives by email and someone keys it into Fishbowl. Someone checks stock and raises the work order. When the job ships, someone logs into a carrier portal and retypes the address that already exists in two other systems. At month end, someone assembles a spreadsheet from Fishbowl exports and Reckon reports so the directors can see how the month actually went, three weeks after it went that way.

Every handover is a person acting as an integration. People are expensive integrations, and they make mistakes exactly when it is busiest.

The fix is not a new ERP. Rip-and-replace projects are how manufacturers lose a year. The fix is automating the handovers on the systems you already own, one at a time, in a chain that looks like this: sales order in, Fishbowl work order raised, Reckon updated, freight dispatched, dashboard live. Here is each stage in detail.

Stage 1: Sales Order Intake, Without the Keying

Order data flowing between business systems automatically instead of being retyped

Orders reach a manufacturer in messy ways: emailed purchase orders as PDFs, a web store, phone calls, repeat schedules from key accounts. Today, someone reads each one and types it into Fishbowl. That is the first place hours go, and the first place a transposed quantity or a wrong SKU enters the chain.

The automated version watches every intake channel. An emailed PO is read, and the line items, quantities and delivery details are extracted and validated against your live catalogue and the customer's account terms. Anything that passes validation is created as a sales order in Fishbowl through its API. Anything that does not, an unknown part number, a price that does not match the agreed rate, a credit-hold customer, is flagged to a person with the reason attached, before it can become a production problem.

The keying disappears, but so does something subtler: the delay. An order that arrives at 4:55pm on Friday is in the system at 4:56pm, not on Monday morning.

Stage 2: Fishbowl Raises the Work Order Against the BOM

This is the stage people usually assume needs custom software inside Fishbowl. It does not, because Fishbowl Advanced ships with a REST API served by the Fishbowl server on your own network. It exposes sales orders, manufacture orders, work orders, parts and inventory as JSON endpoints, alongside CSV import and export routes and a data-query layer for reads. In plain terms: everything a person does in the Fishbowl client to get an order into production can be done by software, against the same data, with the same rules.

i

Why this matters for batch manufacturers: in Fishbowl, manufacture orders organise the work orders that consume your bill of materials. Driving them through the API means a confirmed sales order can commit components, raise the work order against the correct BOM revision, and surface a component shortage as an alert at order time, not as a discovery on the factory floor two days into the run.

The workflow logic lives outside Fishbowl, in an orchestration layer, we typically build on n8n with AI handling the unstructured parts. Fishbowl stays completely standard: no plugins, no database hacks, upgrades unaffected. The orchestrator is just an extremely reliable operator that never gets busy, works nights, and logs everything it does.

Stage 3: The Financial Sync Into Reckon

Fishbowl ships a native Reckon integration for posting inventory and cost-of-goods data into your accounts, and the right move is almost always to keep it as the backbone rather than rebuild it. Reckon stays the source of financial truth. What is missing in most setups is everything around that sync: the checking, the exceptions, and the traffic that never goes through Fishbowl at all.

So the automation layer takes the jobs a bookkeeper quietly does by hand. It verifies that every dispatched order actually produced its invoice. It matches supplier bills for raw materials against what goods-in received. It attaches freight charges to the orders that incurred them, which is the difference between knowing your margin and guessing it. And it flags the exceptions, the credit note that does not match a return, the invoice that never generated, to a person with the context already assembled.

We run this exact pattern in production today: our nightly sync agent for PestAway turns every completed job in their field system into a reconciled Xero invoice, ready to approve, at a rate of thousands of invoices with no manual entry. Swap the field system for Fishbowl and Xero for Reckon and the architecture is the same.

Stage 4: Freight That Books Itself, and Reconciles Itself

Packed orders ready for dispatch with automatically generated consignment documentation

Freight is the most automatable job in the building, because the carriers have done half the work already. Mainfreight publishes an API that creates consignments in its transport management system, generates labels, returns freight estimates and provides track-and-trace. Aggregators like GoSweetSpot and Starshipit put multiple NZ carriers behind one API, so one integration covers couriers and line-haul both.

The automated flow: when a work order completes and the job is packed, the system weighs the options for that lane, weight and service level against your negotiated rates, books the consignment with the winning carrier, prints the labels, and sends the customer their tracking link. No portal logins, no retyped addresses.

For chemical manufacturers there is a second half to this: dangerous goods paperwork. DG declarations and safety data sheet references are assembled from product data you have already signed off, rather than by hand for every consignment. The system never invents a UN number or a classification. It fills documents from your approved data, and anything missing stops the consignment and asks a person.

Then the loop most manufacturers never close: reconciliation. The carrier's weekly invoice is matched line by line against the rates quoted at booking. Overcharges, wrong service levels and phantom surcharges surface automatically. Freight is usually one of the largest unexamined cost lines in a manufacturing business, unexamined only because checking it by hand was never worth anyone's day.

Stage 5: The Executive Dashboard That Ends Month-End Surprises

Everything above generates data as a by-product, and that is where the final stage comes from. Fishbowl knows orders, stock, WIP and output. Reckon knows revenue, costs and margin. The freight layer knows cost per consignment. Pulled together into one live KPI dashboard, the executive view stops being a month-end reconstruction and becomes a window: orders in this week against the same week last year, WIP right now, margin by product line, freight cost per order by carrier, on-time dispatch rate.

The difference is not prettier charts. It is timing. A bad week found in week three of the month costs you three weeks of the same bad week. A bad week found on Wednesday costs you until Wednesday.

Where These Patterns Are Already Running in NZ

We should be straight about what this article is: a reference architecture, not a writeup of a single site. The stages above are the patterns we assemble it from, and each one is running in production for New Zealand businesses today. The system-to-accounting sync runs nightly for PestAway. The live stock intelligence pattern runs for STS Electrical, reconciling van-by-van stock from job data all week with the reorder email drafted by Tuesday morning. The order-to-follow-up automation runs for an NZ printer supplies retailer, ending hours a day of retyped order data.

Fishbowl and Reckon are, from an integration point of view, better-behaved than most of the systems in those builds. They have documented APIs, an official integration between them, and a large installed base among NZ manufacturers, which is exactly why we standardised on this architecture for manufacturing work.

Rolling It Out Without Touching Production

Manufacturers are rightly allergic to anyone experimenting near production. So nothing we build goes live against the factory on day one. New automations run in shadow mode first: the system does the work in parallel while your team keeps doing it the current way, and we compare outputs until the numbers match, week after week. That is how our stock reconciliation builds settle in before anyone relies on a reorder threshold, and it is how an order-intake automation earns the right to create real sales orders.

  1. One gap first. Usually order intake or freight reconciliation, because both have a measurable before-and-after inside a month.
  2. Shadow mode. The automation runs in parallel and is audited against the manual process until it is boringly correct.
  3. Human sign-off stays. A person approves anything consequential, an order over a threshold, a DG consignment, a reorder, until you decide otherwise.
  4. Everything is logged. Every automated action carries the order ID and can be traced, audited and reversed.
  5. Then extend. The next gap reuses the same connections, so the second workflow is cheaper than the first.

Common Questions

Do we need Fishbowl Advanced for this?
The REST API ships with Fishbowl Advanced, so that is the cleanest path. On other configurations, Fishbowl's CSV import and export layer covers most of the same flows, just with a file-based rhythm rather than a live one. We confirm exactly what your install exposes during scoping, before anything is quoted.

Does this replace Fishbowl's built-in Reckon integration?
No. The native sync stays as the accounting backbone. The automation wraps around it: verifying it, handling the exceptions, and moving the data the native sync does not carry, like freight cost allocation and intake-channel orders.

What if an order is ambiguous or wrong?
It stops and asks. The design rule across every stage is the same: the system acts on validated data and escalates everything else to a person, with the context attached. An automation that guesses at a quantity or a UN number is worse than no automation, so ours are built not to.

How long does the first workflow take?
Most first workflows are live, through shadow mode, within three to six weeks of go-ahead, depending on how many systems are in the chain. The second is faster, because the plumbing already exists.

If you run Fishbowl and Reckon and want to know which of these five stages would pay for itself first at your volumes, that is exactly what we work out in a free 30-minute audit, and you keep the roadmap either way. Start with our manufacturing automation service for New Zealand page, or book the audit directly.