BoQ Automation Implementation: What You Need Before You Start – turian

BoQ Automation Implementation:
What You Need Before You Start

Most mid-market companies that delay implementing BoQ automation don't delay because they decided against it. They delay because they're not sure what "ready" looks like, and in the absence of a clear answer, the project stays on the to-do list.

This guide gives you a concrete answer. It covers what you actually need before implementation starts, what you can prepare quickly, what takes more time, and what you do not need at all, including several things vendors sometimes suggest are prerequisites but aren't.

The short version: if your product catalog exists somewhere accessible, your ERP has an API or you have a shared inbox, and you can name the person who will own exceptions, you are further along than you think.

The five areas

What This Checklist Covers

There are five areas to work through before a BoQ automation implementation starts. None of these require months of preparation. Some take an afternoon. A few require a conversation with your IT contact or ERP provider. One or two require an internal decision.

01

Product catalog readiness

02

Historical BoQ data

03

ERP or system access

04

Exception handling rules

05

Internal process ownership

01

Area

Product Catalog Readiness

What you need: your product catalog in a format the system can read, not necessarily clean or perfectly structured, but accessible.

BoQ automation works by matching incoming line items against your internal product catalog. The matching quality depends on what catalog data is available. This does not mean the catalog needs to be perfect before implementation starts. It means it needs to exist in a readable form.

  • Your product catalog is accessible in at least one of: your ERP system, an exported spreadsheet, a PIM system, or a supplier catalog file
  • Each product entry has at minimum: an internal SKU or article number, a product description, and a unit of measure
  • If you have customer-specific cross-references (mapping a customer's article number to your internal SKU), those records exist somewhere, even if they are incomplete or in a spreadsheet rather than the ERP
  • You know which product categories are in your active catalog versus discontinued or delisted: the system should match against live products, not the full historical archive
What you do not need A perfectly clean, deduplicated, fully normalized catalog. Gaps and inconsistencies in the catalog will be surfaced by the matching process, which is part of the value of implementation. You will learn more about your catalog quality from the first weeks of live operation than from a pre-implementation cleanup audit.
If your catalog is fragmented Partly in the ERP, partly in a spreadsheet, partly in a supplier feed: identify which source is most complete and most current. Implementation starts with that source. Additional sources can be added in phase two.
02

Area

Historical BoQ Data

What you need: a sample of real BoQs your team has processed, ideally from the last six to twelve months.

Historical BoQs are used during implementation to test and validate the matching logic against your real order mix before going live. They are not required to train a model: turian's matching is LLM-based and does not require labeled training data. But they are useful for two things: configuring the exception handling rules, and giving both your team and the implementation team confidence that the system handles your actual document mix before it touches live orders.

  • You can locate and export a sample of 20 to 50 BoQs from the last six to twelve months, in whatever format they arrived (GAEB, Excel, PDF)
  • The sample covers your main customer types and product categories, not just the easiest cases
  • You have access to the resolved quotes for those BoQs: knowing what the correct internal SKU was for each matched line item allows the system to be validated against a known-good outcome
  • If BoQs are stored in email rather than a document management system, you can export a representative set without spending days searching inboxes
What you do not need A complete archive of every BoQ ever received, or a cleaned and labeled dataset. Twenty to fifty real examples is enough to validate the matching logic against your environment. The system improves further in live operation.
If you have very few historical BoQs Because your volume is recent or because documents weren't retained: the implementation can proceed without them. The first weeks of live operation become the validation period, with human review on a higher proportion of outputs until confidence is established.
03

Area

ERP or System Access

What you need: either API access to your ERP, or an email integration point, and someone at your company who can confirm which is available.

This is the step that most often causes unnecessary anxiety. "We don't have our ERP set up for API access" is a common concern. In most cases, the ERP does have API access: it just hasn't been used before, and nobody has confirmed the specifics.

  • You know which ERP you use and which version: SAP Business One, SAP S/4HANA, Microsoft Dynamics 365, Infor, ProAlpha, or another system
  • You have an IT contact or ERP administrator who can confirm whether the ERP's API is enabled and reachable
  • You can create or have created a service account for the integration with defined read and write permissions: read access to product master, customer master, and pricing; write access to quote or order records
  • Alternatively: you have a shared inbox or email forwarding address where BoQs are received, and you can configure forwarding or access for the integration
What you do not need A middleware layer, an API gateway, or a custom integration built in advance. The integration is part of the implementation: you provide access credentials and permissions, the implementation team builds the connection.
If your IT team is skeptical or unavailable The minimum viable integration path is email-based. BoQs forwarded to a configured address are read and processed without ERP write access initially. ERP integration can be added in phase two once the matching and processing logic is validated.
BoQ Automation Implementation: Exception Rules, Ownership & Readiness Checklist – turian
04

Area

Exception Handling Rules

What you need: a set of decisions about which situations a human must review, and who reviews them.

Exception handling rules are the instructions that tell the system when to proceed automatically and when to stop and ask a human. You do not need to define every possible exception before implementation: you will discover most of them in the first weeks of live operation. But you need to define the threshold cases before go-live.

  • Price deviation threshold defined

    If the price the system finds in the ERP differs from a price on the incoming BoQ, at what percentage difference should it flag for human review? A practical starting point: flag any deviation above 3% for standard catalog items, tighter for custom or configured products.

  • Unmatched item handling decided

    Options: flag and hold the entire order, flag the unmatched lines and continue with matched lines, or route the full document to a named person.

  • Out-of-scope item handling decided

    Should items outside your product range be automatically marked as "not applicable" and excluded from the quote, or flagged for a human to confirm exclusion?

  • New customer handling decided

    Should BoQs from customers not yet in your ERP be processed automatically or held for manual review pending account creation?

  • Submission deadline flagging decided

    If a BoQ has a tender deadline, should the system flag urgency? Who sees that flag?

What you do not need A complete decision tree covering every edge case. The five decisions above cover the vast majority of exception scenarios for a standard mid-market BoQ workflow. Additional rules can be configured after go-live based on what the exception log shows.
05

Area

Internal Process Ownership

What you need: one named person who owns the implementation on your side, and clarity on who handles the exception queue after go-live.

This is the prerequisite that most implementation guides skip. Technical setup is straightforward. Process ownership is where implementations stall.

  • Implementation owner named

    One named person at your company who is the primary contact during implementation, who has authority to make configuration decisions, and who will sign off on go-live readiness. A Vertriebsleiter or operations manager is the right person.

  • Exception queue owner named

    Who reviews and resolves exceptions after go-live? This should be a specific person or a defined rotation, not "the team." Assign it before the first exception arrives.

  • ERP contact identified

    Someone who can provide API credentials, confirm field mappings, and test the ERP write connection. One hour of their time during implementation is usually sufficient.

  • Go-live definition agreed in writing

    Which order types, which customer groups, what exception review process. Define this before implementation starts.

  • Feedback loop process defined

    How will your exception queue owner communicate configuration changes back to the system? Weekly review of exception patterns, a shared log, or a direct line to the implementation team.

What you do not need A dedicated AI project manager, a change management consultant, or an internal IT team with automation experience. Mid-market implementations run smoothly with one internal owner who has authority to make decisions and half a day per week available during the implementation period.

The full checklist

The Full Readiness Checklist
at a Glance

Product catalog

  • Catalog accessible in ERP, spreadsheet, or PIM
  • Each product has SKU, description, and unit of measure
  • Customer cross-references exist somewhere, even if incomplete
  • Active catalog is distinguishable from discontinued products

Historical BoQ data

  • 20 to 50 real BoQs from the last six to twelve months available
  • Sample covers main customer types and product categories
  • Resolved quotes available for the sample where possible

ERP or system access

  • ERP name and version confirmed
  • IT contact identified who can confirm API availability
  • Service account with defined read/write permissions created, or
  • Shared inbox available as an alternative integration point

Exception handling rules

  • Price deviation threshold defined
  • Unmatched item handling decided
  • Out-of-scope item handling decided
  • New customer handling decided
  • Submission deadline flagging decided

Internal process ownership

  • Implementation owner named
  • Exception queue owner named
  • ERP contact identified
  • Go-live definition agreed in writing
  • Feedback loop process defined

Defer these

What You Can Leave
Until After Go-Live

A few things that vendors sometimes present as prerequisites are actually better addressed during or after live operation.

Full catalog cleanup

A pre-implementation catalog audit takes weeks and delays go-live. The matching process surfaces catalog gaps more effectively than a manual audit: implement first, clean as you go.

Complete customer cross-reference table

Unmapped customer codes are surfaced in live operation as exceptions. Build the cross-reference incrementally based on what actually appears rather than trying to anticipate every mapping in advance.

Edge case exception rules

You will not know your real exception patterns until the system has processed a few hundred real BoQs. Configure the five threshold rules above before go-live; add the rest based on what the exception log shows in the first four weeks.

Full ERP field mapping

Start with the core fields: order header, line items, quantities, prices. Additional ERP fields can be mapped in phase two once the core workflow is stable.

Next step

Book a 30-Minute BoQ
Implementation Scoping Call

If you have worked through this checklist and have most items confirmed, you are ready to start. If two or three items are unclear, that is a normal starting point: the implementation scoping call exists to work through exactly these questions.

Book an implementation scoping call