Publicité
ERP IMPLEMENTATION
🇫🇷 Lire en français →

ERP Fit-Gap Workshops: How to Run Them Without Drifting into Custom Development

ERP fit-gap workshop method, 5-level decision grid, gap register and approval board to keep customizations down before the build starts.

ERP Fit-Gap Workshops: How to Run Them Without Drifting into Custom Development

The fit-gap is the point in an ERP project where the budget is really decided. Before it, everything is a sales promise and a pricing assumption. After it, everything is days consumed: every gap qualified as “custom development” becomes a line of effort, a testing risk and, above all, a debt that gets paid again at every version upgrade.

Yet these workshops are often run like requirements-gathering meetings: people listen, take notes and promise to look into it. The result is that customization swells and nobody ever decided it. This guide offers a practical facilitation method, with a decision grid, a register and an approval board, so that a gap is always a documented decision and never a drift.

Why the fit-gap decides the final cost of the project

A fit-gap compares the company’s processes with the capabilities of the software package and qualifies each difference. It is exactly where the functional scope turns into delivery effort. If you have already framed your scope with one of the 5 methods to prevent scope creep, the fit-gap is its natural extension: it applies that frame line by line.

Three mechanisms explain its weight on the final cost.

Customization is paid for three times. First at design and development, then at testing and training (a screen or calculation that does not exist in the vendor documentation must be documented and taught by you), and a third time at every update, when you have to retest and sometimes rewrite.

A gap left unaddressed upstream comes back more expensive downstream. A requirement discovered during acceptance testing is no longer up for discussion: the schedule is tight, users are anxious, and “let’s build it” is the fastest answer in the short term.

The fit-gap sets the level of transformation ambition. Every gap accepted as is represents a process the company chose not to change. Every gap rejected is a deliberate standardization choice.

Selection fit-gap vs. design fit-gap: don’t confuse them

The term covers two different exercises.

The selection fit-gap takes place before signature. It is used to compare vendors on business scenarios. It is deliberately broad and the level of detail stays limited: the goal is to tell options apart, not to specify.

The design fit-gap takes place after signature, with the selected vendor or integrator. It is detailed, process by process, and produces a list of costed gaps that feed the build schedule and budget. This is the one this article covers. The “yes, standard” answers given during selection are systematically rechecked here: a demo “yes” rarely covers every variant of your real-life case.

Preparing the workshops

A poorly prepared workshop produces a wish list. A well-prepared one produces decisions. The difference is made before participants sit down.

Process scope

Split the scope by end-to-end process, not by department or module. The classic breakdowns work well:

  • Order-to-Cash: from customer order to payment collection
  • Procure-to-Pay: from purchase requisition to supplier payment
  • Record-to-Report: from journal entry to close and reporting
  • Plan-to-Produce, Hire-to-Retire or Project-to-Profit depending on your business

A module-based split (“inventory workshop”, “sales workshop”) misses the gaps that arise at the boundaries: a discount negotiated in sales that must flow into accounting, a batch status that conditions invoicing. Plan one or several 3-hour workshops per process, depending on its complexity.

Who needs to be in the room

The makeup of the room determines the quality of decisions.

  • Key users of the process: the people who do the work, not their managers. They know the real exceptions.
  • A business decision-maker, able to accept a change of practice without checking with three other people.
  • The integrator or vendor, with a functional consultant who knows the standard inside out and an architect able to assess the feasibility of an extension.
  • The project manager, who keeps the register and the clock.
  • The sponsor, present at least at the opening and available the same day to decide: they arbitrate the conflicts the room cannot resolve.

Avoid rooms of more than 10 people: beyond that, debates scatter and the least confident key users go quiet, precisely the ones you need.

Prerequisite: demonstrate the standard before the workshop

Never ask a user to express a gap against a standard they have not seen. They will describe their current tool, not their need. Arrange a demonstration of the standard on realistic company data beforehand, ideally one session per process, a few days before the matching workshop.

Ask key users to bring three things: their real documents (order, invoice, delivery note as they actually circulate), their five most frequent exception cases, and a list of what they saw in the demo that bothers them.

The 5-level decision grid

The fundamental mistake is thinking in binary terms: “it fits the standard” or “we have to build”. Between the two lies a continuum, and most gaps are settled in the first levels. Put the following grid in front of the room and apply it in order: you only move to the next level if the previous one is ruled out for a written reason.

LevelAnswerExample requirementRelative upfront costUpgrade impact
1. StandardThe process adapts to the ERPApprove an order through the native flowVery low (training)None
2. ConfigurationSetting provided by the vendorPayment terms, chart of accounts, numbering rulesLowLow: settings are preserved by the update mechanism
3. Advanced configurationWorkflow, custom fields, business rules using the intended toolsTwo-level approval depending on an amount, customer classification fieldMediumLow to medium: retest needed, but within the supported framework
4. Side-by-side extensionApplication or service attached via API, BTP or connector, without touching the coreSupplier portal, external commission calculationMedium to highMedium: depends on interface stability
5. Custom developmentCode that modifies or duplicates core behaviorRewrite of an inventory valuation calculationHighHigh: to retest and potentially rewrite at every version

Cost levels are relative and specific to each vendor and each project: they indicate an increasing order of magnitude, not a price list.

Two important notes on this grid.

Level 3 is the most underused. Modern ERPs offer many ways to adapt without coding: workflows, additional fields, business rules. To learn what these tools really allow and where their limits lie, see our guide on low-code to customize Odoo, SAP and Dynamics without developers.

Level 4 is the answer of “clean core” architectures. At SAP, the strategy is to keep the core close to standard and to run extensions alongside it, on the dedicated technical layer. If you are on S/4HANA, read our analysis of the clean core strategy and BTP extensions. The principle holds for any SaaS ERP with frequent releases: the more the core is modified, the more every release becomes a regression-testing project.

Running a 3-hour workshop

Here is a typical agenda, to adapt to the density of the process.

DurationSegmentObjective
15 minOpening by the sponsorRestate the rules: we start from the standard, every exception must be justified
20 minStandard recapWalk through the flow live in the ERP, step by step
90 minWalking the process stepsFor each step, note the gaps and enter them in the register
30 minQualificationApply the 5-level grid, propose a provisional decision
15 minContested gapsIdentify what goes up to the approval board
10 minClosingReread the register, confirm decisions and actions

Keep one simple rule: the workshop only decides gaps that are settled at levels 1 to 3. Anything heading toward levels 4 and 5 is qualified, estimated, then submitted to the board (see below), not approved in the room.

The “why × 3” rule

A user almost always expresses a solution (“I need an extra field on the purchase order”), rarely a need. To get back to the process, ask “why” up to three times.

  • “I need a project code field on the order.” Why?
  • “To find the purchases for a given site.” Why?
  • “Because the controller asks me for the cost per site every month.”

The real need is cost analysis per site. The ERP may well provide it natively through analytical dimensions, which makes the field unnecessary. If, after three “whys”, the need is still a real need, it is now phrased as an expected outcome, which lets you qualify it with the grid.

Handling “we’ve always done it this way”

This phrase is not an argument, but it is not always baseless: it sometimes hides a real constraint (regulatory, contractual, security) the user has not made explicit. Three reflexes help.

  • Ask about the origin: “What would happen if we stopped doing it?” If nobody can answer, it is a habit. If the answer mentions a customer, an auditor or a law, it is a constraint to document.
  • Separate effort from value: a practice that costs several people time for a benefit nobody measures is a candidate for retirement.
  • Escalate to the sponsor the cases where the user and the integrator remain in disagreement. Do not seek consensus at any price: a clear decision beats a vague compromise.

The gap register

Everything decided in a workshop must end up in a single register, kept by the project manager. Decisions are documented, not passed on orally: six months later, nobody remembers why a screen was built or who asked for it.

Recommended columns:

  • ID: stable identifier (for example OTC-012)
  • Process and step concerned
  • Description of the need, rephrased after the “why × 3”
  • MoSCoW priority: Must, Should, Could, Won’t
  • Grid level chosen (1 to 5) and decision
  • Estimated effort in days, with the estimator’s name
  • Maintenance and upgrade risk: low, medium, high
  • Requester and arbitrator (name, date)
  • Status: to qualify, decided in workshop, submitted to board, approved, rejected

Example of five fictional rows, for illustration only:

IDProcessNeedMoSCoWDecisionEffortUpgrade risk
OTC-004Order-to-CashMulti-tier discount by customerMustLevel 3: native pricing rules3 dLow
P2P-011Procure-to-PayDual approval of purchases above a thresholdMustLevel 3: standard workflow2 dLow
R2R-007Record-to-ReportMonthly export in the group formatShouldLevel 4: extension via API6 dMedium
OTC-019Order-to-CashCustom commission calculation for 2 agentsCouldRejected: handled outside the ERP0 dNone
LOG-023LogisticsParcel label with carrier-specific layoutMustLevel 5: development, submitted to board8 dHigh

Note row OTC-019: a healthy register contains rejections, with their reasons. If every gap is accepted, it is not a register, it is an order book.

Arbitrating: the custom-development approval board

Every level 5 gap (and the heaviest level 4 ones) goes before a small board. This board can be a dedicated segment of the weekly anti-drift steering committee rather than one more body.

Composition: the sponsor (who decides), the business owner of the process, the CIO or architect, the project manager, and the integrator in an advisory role.

Effort threshold. Set a threshold above which a development must be justified in writing and approved by the sponsor. The amount depends on project size: what matters is that it exists, that it is known before the workshops start, and that it also applies to cumulative totals (ten small customizations of 2 days weigh as much as one big one).

Maintenance cost over the lifetime. For every customization, ask the question nobody asks: what will it cost over the lifetime of the ERP? The initial cost is only the visible part. Add the retest at every version upgrade, the scarce skill to retain, the documentation to keep current, and the risk of blocking an update. On a SaaS ERP with frequent releases, this cost is recurring and not negotiable with the vendor: the release cycle is imposed on you.

Questions to ask the board every time:

  1. Is there an alternative at levels 1 to 4 that the workshop did not explore?
  2. What happens, concretely, if we do not build this? Who is penalized, and by how much?
  3. Can it be postponed until after go-live, once the standard is used in real conditions?
  4. Who bears the maintenance, and with what annual budget?

A good reflex: accept customization more readily “after go-live” than “before go-live”. Users then discover that many needs expressed during design evaporate on contact with the standard.

The 5 common mistakes and how to counter them

1. Running the workshop without having shown the standard. Users describe the existing setup, and the gap is measured against a tool they know, not against the ERP. Countermeasure: a mandatory demo before each workshop, on realistic data.

2. Leaving the integrator as the sole judge of qualification. Their business model may favor development, even when they act in good faith. Countermeasure: have levels 4 and 5 validated by the board, with a written estimate and a justification of why lower levels are not possible.

3. Making verbal decisions. An “we’ll see” in a workshop turns into a promise six weeks later. Countermeasure: no decision exists until it is in the register, reread at the end of the session.

4. Not forcing conflicts to be settled. When two departments defend incompatible practices, compromise often creates a customization that satisfies everyone and simplifies nothing. Countermeasure: systematic escalation to the sponsor, with an agreed response time.

5. Accepting a customization without measuring its maintenance cost. The development is priced, never the life of the development. Countermeasure: a mandatory “upgrade risk” column in the register, and an annual maintenance budget line attached to every level 5.

Fit-gap phase exit checklist

Do not close the phase until all these items are ticked.

  • Every process in scope has had at least one workshop, with a prior demonstration of the standard
  • Every gap is in the register, with ID, MoSCoW priority, grid level and decision
  • Level 4 and 5 decisions have been approved by the board, with the arbitrator’s name and the date
  • Every customization has an estimated effort, an owner and a documented maintenance risk
  • Rejections and post-go-live deferrals are documented with their reasons
  • The build budget and schedule are updated from the register, not from the initial proposal
  • Key users have reviewed and signed off the decisions for their process
  • The sponsor has approved the final list of customizations and the associated maintenance budget

Conclusion

A successful fit-gap is measured not by the number of needs satisfied, but by the quality of the decisions made and their traceability. Start from the standard, qualify every gap on a 5-level grid, keep a register, have the sponsor settle conflicts: this discipline calls for facilitation, not technology.

To go further, estimate your trade-offs with our interactive ERP tools, and download our ERP evaluation scorecard to keep an objective basis for comparing the options in your project.