Skip to content
Tutorials Center
My learning

Topics

Lesson 33 of 36 ยท Selection and Case Studies

Case Studies and Lessons Learned

Learn from four composite ERP and CRM case studies: a manufacturer, a retailer, a services firm and a nonprofit, with what went right, what failed and a Python risk checklist.

Theory is easier to remember when it is attached to a story. This lesson presents four composite case studies. They are invented for teaching: they combine patterns that recur in real projects across industries, and they name no real company or product. After each story you will find the lessons and the questions to ask in your own project.

Case 1: A manufacturer replaces spreadsheets and an old ERP

Situation. A mid-sized maker of industrial pumps (about 400 employees, two plants) ran an ageing ERP supplemented by hundreds of spreadsheets for planning, costing and quality. Month-end close took 14 days and nobody trusted inventory values.

What they did. They set measurable goals (close in 6 days, inventory accuracy above 98 percent, on-time delivery above 95 percent), staffed the project with their best process owners part-time, chose phased by plant (pilot at the smaller plant first) and kept customization small by adopting standard processes except for a custom configurator that gave them a real competitive edge.

What went wrong. Bills of materials in the old system were full of errors, which only appeared during the first mock load. The data team had planned two trial migrations but needed five. The go-live moved by ten weeks. A key production planner, who had never been involved in design, rejected the new planning screens and kept a parallel spreadsheet for months.

Results. After the second plant went live the close dropped to 5 days. Inventory accuracy reached 97 percent in the first year and 99 percent later, after cycle counting became routine.

Lessons.

Case 2: A retailer launches CRM and loyalty on top of an existing ERP

Situation. A regional retail chain with 60 shops and an online store had separate customer lists in the shop till system, the web shop and the email tool. Marketing sent duplicate offers and could not see who the best customers were.

What they did. They implemented a cloud CRM with a customer data platform in front, matched records by email and loyalty card (master data management) and connected the till system through an event stream. They started with one measurable use case: win back lapsed customers with an RFM-based campaign (forecasting and analytics).

What went wrong. The consent status of customers was not migrated consistently, so the first campaign had to be paused for a compliance review. Shop staff were not told why they should ask for email addresses at the till, and the capture rate stayed low for six months.

Results. After consent data was fixed and shop staff received a simple script and a small incentive, the share of transactions linked to a known customer rose from 28 to 55 percent. The win-back campaign returned about three times its cost.

Lessons.

  • Treat consent and privacy as data requirements from day one.
  • Frontline behaviour decides data quality; explain the reason and make capture easy.
  • Start with one clear use case with a measurable return.

Case 3: A professional services firm picks the wrong system

Situation. A consulting firm with 250 staff needed project accounting, time recording and a CRM. The managing partner saw an impressive demo of a large, global ERP and signed quickly, without a scorecard.

What went wrong. The product was designed for manufacturing and distribution. Project-based billing needed heavy customization. The implementation partner had no experience in services. After 18 months and nearly twice the budget, consultants still entered time in a spreadsheet because the time-entry screens were slow and unfamiliar. The firm eventually replaced it with a purpose-built professional services product.

Lessons.

  • Run a structured selection with scripted demos of your own scenarios (selection criteria and scorecards).
  • Industry fit and partner experience matter more than brand size.
  • Usability for the daily user is a hard requirement, not a nicety.
  • A sunk cost is not a reason to continue: the firm saved money by stopping early after an honest review.

Case 4: A nonprofit adopts a low-code CRM

Situation. A charity with 40 staff and 3,000 volunteers tracked donors in spreadsheets and email. It had a small IT budget and one part-time administrator.

What they did. They chose a SaaS CRM with a nonprofit template, kept every process as standard as possible, used a certified partner for a 10-week setup, and trained volunteers with short videos. Reports were limited to six dashboards agreed with trustees. A yearly review decides about extensions.

What went wrong. Hardly anything: a slow start with email integration because the old domain settings were wrong, and a volunteer coordinator who needed extra coaching. The team admitted that small projects benefit most from restraint.

Results. Donation reminders became automatic, repeat giving rose, and trustees could see income by campaign without waiting for a manual report.

Lessons.

  • Small organizations succeed with standard, hosted solutions (cloud, SaaS and on-premise).
  • Restraint in scope and reports keeps the project cheap and finishable.

Common failure patterns

People and governance

Absent sponsor Decisions stall; scope grows.
Users left out First seen at UAT; then rejected.
Weak change plan Training only at the end.

Most failures are organizational.

1 / 2

A risk checklist you can run

The stand-in below scores a project against the failure patterns. Answer honestly: a "no" adds to the risk score. This is a conversation tool, not a prediction.

Python Project risk self-assessment (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.

Reading the numbers: schedule and budget overruns

Large IT projects commonly run over time and budget. The next stand-in shows how a modest slip in several areas compounds, and why a contingency reserve is realistic rather than pessimistic.

Python How small overruns compound

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.

The biggest single contributor here is the data work, a pattern found in nearly every migration story.

What successful projects share

  • Clear business goals and a sponsor who stays engaged.
  • A pragmatic scope with phased delivery and a pilot.
  • Standard processes, with few, justified customizations.
  • Data treated as a workstream with an owner.
  • Testing protected from schedule pressure and a real rollback plan.
  • Early, continuous user involvement and role-based training.
  • Honest reporting of bad news, and a willingness to say no-go.
  • Benefits tracked after go-live.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. In the manufacturer case, what caused the ten-week delay?

    Poor source data needed repeated cleansing and loading, a very common schedule driver.
  2. 2. What lesson does the services firm case teach about selection?

    A brand-name demo hid a poor fit with project-based billing.
  3. 3. Why did the retailer's first campaign pause?

    Consent and privacy status are essential data attributes and must be treated as requirements.
  4. 4. Why is it reasonable to plan contingency into schedules?

    Typical overruns in different phases add up, so realistic plans include a reserve.

Summary

  • Composite case studies show recurring patterns: data problems, weak user involvement, wrong fit, privacy gaps.
  • Success comes from clear goals, standard processes, protected testing and strong change work.
  • A risk checklist surfaces the gaps early; fix the highest-weight ones first.
  • Plan contingency; data work is the usual culprit for slips.
  • Stopping or re-planning early can be the best decision.

Next: Project: An Order-to-Cash Simulator