Doomed Before Signature: How ERP Projects Fail at Procurement

When an ERP project fails, the blame usually lands on go-live: the system that did not work, the data that did not migrate, the users who never adopted it. But most failed projects were set up to fail much earlier, during selection and procurement.

That matters because the period before signature is when the customer has the most leverage. Bidders are competing, terms are still open and no money is committed. Once the contract is signed, much of that leverage passes to the supplier, and changes are more likely to carry an additional cost. Five problems come up repeatedly, and each is far cheaper to fix before signature.

Features, not outcomes

Selection processes tend to become a comparison of features. RFP spreadsheets score hundreds of functional line items, and vendors run scripted demonstrations that show their product at its best.

But the business is not buying features. It is buying outcomes, such as a faster month-end close, accurate stock figures or fewer manual reconciliations, and a system can tick every box in the RFP without delivering them. Demonstrations can add to the problem: they run on the vendor’s own data, and what is shown may be standard functionality, a heavily configured version or something still on the roadmap.

The legal issue is reliance on vendor representations. ERP contracts will commonly contain an entire agreement clause, so anything not written into the contract does not count. Many also contain a non-reliance clause, in which the customer confirms it has not relied on anything said during the sales process. If a demonstration or an RFP answer turns out to be wrong, the customer’s options are usually limited.

The fix is to identify the handful of claims the decision rests on and get them into the contract, as warranties, in a schedule or in the statement of work. If the vendor will not commit in writing to something it said in the pitch, that tells you something.

The implementation partner

ERP vendors often rely on approved partners to design, configure and implement their systems. The quality of that partner, and the team it assigns, can matter as much as the software itself. The same product can succeed in one business and fail in another, and the difference is usually the quality of the design, the data migration, the testing and the team.

Yet businesses often spend months scrutinising the software and far less time on the implementer. Due diligence should cover four things: references from customers who implemented the same product at a similar scale, including what went wrong; the proposed delivery team, rather than only the people who led the pitch; any subcontracting or offshore delivery; and the implementer’s financial stability and standing with the vendor.

The contract should then lock in what the diligence found. Name key personnel, require approval before they are replaced and restrict subcontracting without consent. A strong delivery team is only worth something if the people assessed stay on the project throughout implementation.

Unrealistic assumptions

Budgets and timelines are often fixed before anyone has checked how well the system fits the business. The board approves a business case built on a headline price and a go-live date, both taken from the winning bid.

That price rests on assumptions in the implementer’s statement of work: the data will be clean, the business will adopt standard processes, key staff will be available, decisions will be made quickly and the number of integrations will be capped. In a competitive tender, every bidder is under pressure to make those assumptions as favourable as possible. When one proves wrong, the implementer can usually claim more time and money through change control. The fixed price is only fixed while the assumptions hold.

Proper scoping deals with this before signing. Check which critical processes the system handles as standard and which will need changes, extra development or workarounds, and price those gaps now. A focused paid discovery phase before committing to the full implementation is generally money very well spent. Read every assumption in the statement of work as carefully as the price, ask whether it is realistic and who controls it, and budget for the work the business will do itself, such as cleaning data and freeing up staff.

Gaps in responsibility

Three parties share the work. The vendor supplies the software and fixes its defects. The implementer designs and builds the solution, including configuration, integrations, data migration and testing. The customer provides decisions, data and people.

When something fails, the cause is often unclear. If invoices post to the wrong accounts, is that a software defect, a configuration error or bad data? Each party can usually point to another, and the project stalls while they argue. The vendor typically promises that the software works as documented. The implementer typically promises reasonable skill and care, which means doing a competent job, not delivering a result. Unless the contract says otherwise, no single party may be contractually responsible for the system working as a whole.

Allocating implementation risk is a pre-contract job. Set out who does what in a responsibilities matrix in the statement of work, including the customer’s own tasks. Make one party, usually the implementer, responsible for the solution working end to end, including chasing the vendor when a problem turns out to be in the software. If it will not, agree which risks it carries and which stay with the business, and price the split accordingly. Define the customer’s obligations precisely: what it must provide, by when and in what form. Vague obligations are easy to blame when the project slips.

Success not defined

Statements of work usually describe activities rather than results: workshops held, design documents delivered, configuration completed, test cycles run. Payments are often linked to the same milestones. Success then means the work was done, not that it worked.

Good requirements documentation starts with two separate documents. The specification describes how the system works: its functions, configuration, integrations and data flows. The customer requirements describe what the business needs it to achieve. “The system will perform all payroll functions for the company” describes what a system must do at a high level. “The system will calculate gross pay for each employee for the relevant pay period, including basic salary, overtime, bonuses, commission and other applicable variable payments” describes a more detailed outcome it must support. Vendors and implementers will usually commit to the specification, because they control it. Only the customer can write the requirements, and they need to be written before signing, in terms that can be measured and tested.

Put both the customer requirements and the specification in the contract, but make clear that the customer requirements take precedence over the specification. Before agreeing anything that affects those requirements, carry out a gap analysis to identify any shortfall between what the business needs and what the proposed solution will deliver. Any change to the requirements should then be signed off at the appropriate level within the customer’s business. Set acceptance criteria against the requirements, and tie milestone payments to the requirements being met rather than to activities being completed. Keep the list focused: a short set of critical requirements that are enforced is worth more than hundreds that are not.

Five questions for the board

Before signing, the board should be able to answer five questions:

  1. Which of the vendor’s claims are written into the contract?

  2. Have we checked the proposed delivery team, not just the implementation partner’s credentials, and secured a contractual commitment to keep that team on the project throughout implementation?

  3. Which assumptions is the price based on, and have we tested them?

  4. Who is responsible for the system working as a whole?

  5. How will we know the project has succeeded, and does the contract say so?

None of these questions requires the board to read the contract, but each needs an answer before anyone signs. If the board cannot answer the last one, nobody has promised the outcome.

If your business is selecting an ERP system or negotiating with a vendor or implementation partner, early legal input can help expose gaps before they become expensive project problems. Contact Mark Neale to discuss your requirements.

The content of this page is a summary of the law in force at the date of publication and is not exhaustive, nor does it contain definitive advice. Specialist legal advice should be sought in relation to any queries that may arise.

Related expertise

Get in touch

Contact us today

Whatever your legal needs, our wide ranging expertise is here to support you and your business, so let’s start your legal journey today and get you in touch with the right lawyer to get you started.

Telephone

Get in touch

For general enquiries, please complete this form and we will direct your message to the most appropriate person.