This is a summary of our conversation. Luke and I recently sat down for our first CFO Q&A, tackling five of the questions we hear most often from finance leaders about data and analytics projects. These are the questions that come up again and again in real conversations with clients and in consults.
1. Why isn't an ERP system enough for reporting?
ERP platforms are brittle — every company is unique, so customization is almost always required, and every new business rule or process change means another round of costly developer edits. They're rigid (new visual = more visual code), not built for history (most ERPs report as-at-today, so you can't reliably answer "what was this number a while ago?"), and cross-system reporting is hard once you add logistics, HR, or payroll.
Pulling straight from the ERP into Power BI isn't the fix either: logic gets duplicated across reports (numbers quietly drift apart), manual extraction is a hidden risk (unclear provenance once someone else takes over), it strains the ERP's resources (refreshes slow the system or get throttled), and Power BI itself can time out on complex queries or large volumes right as you're trying to scale.
2. How do you know your current setup is holding you back?
Red flags to watch for:
- Too many man-hours on reporting — e.g., a full week before month-end, four people, five days, just reconciling numbers.
- Reports only come out monthly or quarterly.
- Small new business rules require the team to "go off and research" how the current logic even works.
- Nobody can quickly answer "where does this number come from?" — a simple reconciliation question turns into a research project.
- The team hesitates at the idea of a new system, worried about interdependencies and things breaking.
3. How do you show ROI to justify the investment to the board?
- Start with the cost of manual reporting — capture time spent cleaning/maintaining data, multiply by hourly rate and ~52 weeks a year. The number is usually bigger than expected.
- Quantify the cost of slow reactions — time saved is time spent catching problems (underused equipment, creeping costs) before they escalate.
- Look for revenue leakage — e.g., missed discount deadlines on invoices because data wasn't available in time to act.
- Connect it to top-line performance — faster access to sales/campaign data means quicker course-correction and profitability, not just efficiency.
4. How do you avoid repeating past implementation mistakes?
- Understand the business before you build anything — learn what reports are actually used, which numbers are typically wrong, and where the recurring frustrations are.
- Build modularly so it scales, capturing logs, run times, and system info as you go — otherwise you get a "black box" that fails silently.
- Engage business users early and often — show an MVP before building the whole thing, since preferences can be surprising.
- Put governance in place — assign clear ownership of data, IT security, and data quality, ideally under a data steward.
5. How much would a project like this cost, and how long would it take?
- Requirements gathering: 4–8 weeks per system.
- Bringing data into the platform: 2–4 months per system.
- Data modeling and cleaning: ~3 months, building a centralized model (often a star schema).
- Reporting: 3–6 months for a typical 10–20 report environment.
For a mid-sized environment with three or four source systems (12–18 months): $150,000–$320,000 with an external consulting company, or $475,000–$650,000 with an internal team over 18–24 months. Internal teams tend to take longer and cost more since they're juggling other priorities — and more people doesn't mean faster, since the work is highly interdependent.