Skip to content
Tutorials Center
My learning

Topics

Lesson 31 of 36 ยท Selection and Case Studies

Selecting an ERP or CRM System

Run a structured ERP or CRM selection: requirements, long list, RFI and RFP, demos with scripts, weighted scorecards, reference checks and contract points, with a Python scorecard.

Choosing an ERP or CRM is a decision a company lives with for ten years or more. It is easy to be swayed by a polished demo or the lowest price, and hard to undo a poor choice. A structured selection process replaces opinion with evidence: clear requirements, a fair comparison, and a decision people can defend later.

This lesson covers the steps from requirements to contract. The products themselves are not available on this site, so the comparison tools are Python stand-ins and the figures are illustrative.

The selection process

Define

Goals Business problems and measurable targets.
Requirements Prioritized must, should, could.
Team Sponsor, process owners, IT, finance.

No requirements, no fair comparison.

1 / 3

1. Goals and requirements

Start from business goals ("reduce month-end close from 12 to 5 days"; "give field sales a mobile CRM"), not from product names. Turn them into prioritized requirements, as in requirements and fit-gap analysis. Split them into mandatory (a no means elimination), important (scored) and nice to have. Include non-functional needs: expected users and transaction volumes, countries, languages, currencies, tax rules, security, integration points, deployment model (see cloud, SaaS and on-premise).

2. Long list, RFI, RFP

  • Long list: use analyst reports, peer recommendations, industry associations and partner networks. Fit matters: industry-specific products often beat general ones; size matters too (a tool built for global groups may be too heavy for a 50-person firm).
  • RFI (request for information): a short questionnaire: size and references in your industry, deployment models, regions, partner network, pricing model. Use it to reduce the long list.
  • RFP (request for proposal): a detailed document with your requirements, scenarios, volumes and a standard response format so answers can be compared. Ask vendors to state per requirement: standard, configuration, extension, custom, or not supported, and to give a price for every item. Ask for a fixed-price or capped implementation estimate with assumptions.

3. Scripted demonstrations

Never accept a generic demo. Give each shortlisted vendor the same demo script: three to six end-to-end scenarios from your real business (for example a make-to-order sale with a partial shipment and a credit note) and some of your sample data. Let your process owners watch and score each scenario right away on a standard form. Ask to see things live, not slides, and ask what is standard versus add-on.

4. Reference checks and due diligence

Speak to at least three customers of similar size, industry and deployment, ideally ones the vendor did not hand-pick. Ask: how long and how much compared with the plan? What surprised you? How is support? How painful are upgrades? Also check the vendor's financial stability, product roadmap, user community, the implementation partner's experience and the availability of skilled consultants in your region.

A weighted scorecard

Combine all evidence into one scorecard: criteria with weights agreed before seeing the vendors, and scores from the whole evaluation team.

Python Weighted vendor scorecard (stand-in)

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Run a sensitivity check like the one above. If the winner changes when weights move a little, the choice is fragile and you need more evidence on the deciding criteria. If the winner stays on top, you can decide with confidence.

Mandatory criteria as a filter

Mandatory requirements work as a gate before scoring: a product that cannot meet a legal requirement (for example local tax reporting in a country where you operate) is out, however good the rest looks.

Python Gate first, then score

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Vendor B scores highest but fails a mandatory requirement, so the gate removes it. Without a gate, a high total can hide a deal-breaker.

Contract and negotiation points

  • Scope and price: licences or subscription metrics (named user, concurrent, usage), price protection for renewals, discounts for volume, rates for extra consulting.
  • Implementation: fixed price vs time and materials, milestones with payment linked to acceptance, named key consultants, and what happens when the schedule slips.
  • Service levels: uptime, support response by severity, penalties or credits.
  • Data: ownership, export formats and fees, return or deletion at exit, residency and privacy terms.
  • Change and exit: upgrade policy, end-of-life commitments, termination rights, escrow (for on-premise source code where relevant).

Common selection mistakes

  • Choosing the system before defining the problem.
  • Letting one executive's favourite win without a scorecard.
  • Asking for too many requirements, so every vendor answers "yes" to everything and the answers carry no information.
  • Comparing licence cost only; ignore the implementation and the running cost at your peril (see total cost of ownership and ROI).
  • Not involving the people who will use the system daily.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What is the main purpose of a scripted demo?

    The same script for every vendor makes the demos comparable and shows real fit.
  2. 2. When should scorecard weights be agreed?

    Fixing the weights first prevents tuning them to favour a preferred vendor.
  3. 3. What does a sensitivity check on a scorecard reveal?

    If small weight changes flip the winner, the decision is fragile and needs more evidence.
  4. 4. Why use mandatory criteria as a gate?

    A missing mandatory capability eliminates a candidate no matter how well it scores elsewhere.

Summary

  • Selection follows goals, requirements, long list, RFI, RFP, scripted demos, references and negotiation.
  • Use mandatory gates plus a weighted scorecard fixed in advance.
  • Test the ranking with a sensitivity check.
  • Reference checks and partner quality matter as much as the product.
  • Negotiate exit and data terms before signing, not after.

Next: Total Cost of Ownership and ROI