Should you use AI to select an ERP?

Choosing an ERP system has long been one of the most strategic—and often most painful—decisions an organisation makes. Traditional selection projects take months, pull in large teams, and are vulnerable to biases, internal politics, and persuasive sales demos.

AI now promises a quicker, more objective alternative. It claims to streamline analysis, surface insights, and help organisations make decisions rooted in evidence rather than instinct. But is it truly ready to replace tried‑and‑tested human expertise?

This article explores where AI can help across the ERP selection lifecycle—and, just as importantly, where its limitations still matter.

Simply asking Copilot—or any AI tool—to recommend an ERP for your business is risky. Out‑of‑the‑box AI only knows what’s publicly available. It has no understanding of:

• Your culture, processes or data quality
• The “magic” your people bring every day
• What differentiates your organisation
• Your unique pain points and ambitions

AI also tends to over‑represent the well‑marketed “top ten” systems, overlooking niche or specialist solutions that might actually fit your organisation far better.

To give AI any meaningful chance of helping, you would need to feed it rich, detailed internal information: updated workflows, process maps, strategy documents, data definitions, and context only your people can provide.

And gathering that level of detail? Human experts still do it best.

No matter your approach, a successful ERP selection involves several core steps.

1. Understand Your Digital Strategy

Whether written down or not, your organisation already has an ambition for digitalisation. You need clarity on scope, timescales, priorities, and how ERP fits into that wider vision.

2. Gather and Prioritise Requirements

Often the most time‑consuming part of the project. Every department has needs—and politics. Requirements must be collected, rationalised, and prioritised (often using MoSCoW) with a focus on what truly differentiates your business.

And remember: basic functionality (purchase orders, chart of accounts, etc.) is a given in modern ERP systems.

3. Assess Your Requirements Against Vendor Capabilities

This is where bias and fatigue creep in. Vendors naturally over‑emphasise strengths. Evaluators get overwhelmed. Comparing responses consistently is tough—and critical.

Costs should also be assessed at this stage.

4. Shortlist for Demonstrations

Demos take at least a day each, so only invite systems that genuinely fit your whole organisation.

5. The Demonstrations

Great demos can inspire your team—but they can also create bias. A consistent agenda and structured scoring helps keep the evaluation fair and comparable.

6. The Decision

Often there is no clear winner. ERP capabilities have improved dramatically across the board, so additional workshops or deep dives are usually needed.

7. Negotiation

Licensing, scope, implementation approach, commercials—getting this right sets the tone for the next decade of your ERP journey.

AI has real potential across several stages—but only if fed accurate, complete, well‑structured data.

1. AI for Requirement Definition

AI could theoretically:

• Analyse your documents and workflows
• Identify gaps
• Group and prioritise needs
• Generate targeted vendor questions

This could produce a clean, deduplicated, business‑aligned requirements set.

But… if your documentation isn’t complete or up‑to‑date, AI will be working from a false reality. And you’ll still need to review everything carefully.

2. AI for Vendor and Product Shortlisting

AI tools could maintain dynamic profiles of ERP platforms, track industry changes, benchmark vendors, and match systems to your needs.

A query like:

“Which ERP supports multi‑entity consolidation, low‑code extensions and AI forecasting for mid‑market manufacturers?”

could generate an evidence‑based shortlist in minutes.

But… you must know which data sources your AI uses. If they’re incomplete, biased, or shallow, the shortlist will be too.

3. AI for Cost Evaluation and TCO Forecasting

Here, AI shines. It can model:

• Licensing and subscription scenarios
• Cloud vs on‑prem infrastructure
• Implementation options
• Multi‑year TCO across multiple growth or operating scenarios

If the data is available, the maths is straightforward—and genuinely useful.

4. AI‑Driven Vendor Evaluation and Scoring

AI could help by:

• Scoring responses consistently
• Flagging vague vendor promises
• Analysing demo content
• Normalising evaluator scoring to reduce bias

But… this is a sophisticated application of AI. Set‑up is heavy, and you will still need to interrogate results where nuance or context is missing.

At Gradient Transforming, we continue to advocate human led ERP selection. Our discovery sessions uncover what makes your organisation unique: the unwritten processes, personal expertise, workarounds, spreadsheets, and cultural dynamics that AI cannot interpret or evaluate.

We combine this understanding with our regular engagement with a wide range of ERP vendors. We help clients shortlist confidently, shape requirements, identify opportunities for improvement, and evaluate systems realistically.

Because ERP success isn’t just about technology—it’s about People, Process and Data, underpinned by the right system and the right partner.

Set up a call with our team to learn more about how we can support your transformation journey.