The financial reporting problem

A typical CFO dedicates 3 to 5 days per month producing financial reports. Not analyzing. Producing. Extracting data from the ERP, building tables in Excel, formatting charts in PowerPoint, writing commentary. It is mechanical work that consumes the company's scarcest resource: the time of the senior finance team.

And the result is usually frustrating for everyone. The executive committee receives a 40-page pack that nobody reads entirely. The Board of Directors asks for a 3-page summary that does not exist. The CEO wants to know "how we are doing" and does not want to wait for the month-end close to find out.

AI solves both problems simultaneously: it reduces report production time and improves the quality of financial communication. An automated report is not worse than a manual one. It is better, because it is more consistent, more timely, and can be personalized for each audience.

The three levels of financial reporting

• Operational: daily or weekly. For the finance team. Cash KPIs, collections, payments, daily sales. Must be automatic and real-time.
• Management: monthly. For the executive committee. P&L, balance sheet, cash flow, forecast, variances. Must be fast (D+3 maximum) and actionable.
• Governance: quarterly. For the Board of Directors, investors, regulators. Must be rigorous, complete, and narratively clear.

AI can partially or fully automate all three levels, but the strategy differs for each.

Executive dashboard for the CFO

The CFO dashboard is the most important visual management tool in the finance department. It is not a monthly report: it is a real-time screen showing the company's financial status at a glance.

The 12 KPIs every CFO dashboard needs

There are hundreds of possible financial metrics. A good dashboard shows 12 or fewer. The essentials:

1. Cash position: current treasury balance, 30-day trend, 30-day forecast.
2. Revenue vs budget: cumulative revenue vs budget, broken down by business line.
3. EBITDA actual vs forecast: month's EBITDA margin, YTD cumulative, and closing forecast.
4. DSO (Days Sales Outstanding): average collection period. Trend and comparison with target.
5. DPO (Days Payable Outstanding): average payment period. Balance with DSO.
6. Burn rate: for growth companies, how many months of runway remain at the current spending rate.
7. Working capital: net working capital evolution.
8. Net debt / EBITDA: leverage ratio, critical for bank covenants.
9. Headcount and personnel cost: FTEs and cost per employee vs budget.
10. Capex executed vs approved: investment plan execution.
11. Top 5 customers: revenue concentration and collection status.
12. Alerts: active alerts for covenants, debt maturities, significant variances.

Dashboard design

Design principles for an effective financial dashboard:

• One page: if it does not fit on one screen, it is not a dashboard; it is a report.
• Traffic lights: green (on target), yellow (attention), red (outside threshold). The CFO should be able to see the company's status in 5 seconds.
• Trend, not snapshot: each KPI should show its temporal evolution, not just the current value. A DSO of 65 days is concerning if it was 50. It is encouraging if it was 80.
• Drill-down: click on any metric to see the breakdown by business unit, customer, or period.
• Automatic refresh: data refreshes automatically from the ERP. No data is entered manually.

Financial KPI automation

KPI automation goes beyond calculating numbers. It includes definition, calculation, contextualization, and alerting.

Standardized definition

The first step (and the most ignored) is defining exactly what each KPI measures. "EBITDA margin" can be calculated 5 different ways depending on what you include as extraordinary. AI can help you document each KPI definition precisely and consistently.

For each KPI you need to document:

• Exact formula with the GAAP/IFRS accounts involved.
• Calculation frequency (daily, weekly, monthly).
• Data source (which ERP table, which system).
• Validation responsible.
• Alert thresholds (green/yellow/red).
• Sector benchmark.

Predictive KPIs

Traditional KPIs are retrospective: they tell you what already happened. AI enables predictive KPIs:

• Projected cash position at 30 days: based on expected collections, scheduled payments, and historical patterns.
• Month-end revenue forecast: based on sales pipeline data and historical conversion rates.
• Projected DSO: based on current receivables aging and each customer's historical behavior.

Board reporting with AI

The board report is the most important financial document the CFO produces. And the most time-consuming, because it requires rigor, clarity, and a level of synthesis that is not easy to achieve.

Board pack structure

A typical financial board pack has this structure:

1. Executive summary (1 page): the 3-5 things the Board member needs to know. Most important first.
2. P&L (2 pages): income statement with comparison vs budget and vs prior year. Commentary on major variances.
3. Balance sheet and cash flow (1-2 pages): financial position and cash movements.
4. Forecast (1 page): year-end closing forecast. Scenarios if there is significant uncertainty.
5. Operational KPIs (1 page): non-financial metrics that impact future results.
6. Risks and opportunities (1 page): top risks with probability, impact, and mitigation.
7. Decision proposals (if applicable): investments, financing, restructurings requiring Board approval.

How AI assists in board reporting

• Draft generation: from the closing data, AI generates the complete first draft of the board pack. The CFO reviews, adjusts, and approves.
• Financial narrative: AI transforms number tables into explanatory text. "Revenue grew 8.3% year-over-year, driven by the new digital business line (+340K EUR) which offset the decline in the traditional segment (-120K EUR)."
• Consistency: style, formats, and terminology are consistent between quarters. AI maintains a template the CFO has approved.
• Personalization: if the Chairman requests a special focus on cash flow and the independent director wants more detail on risks, AI can generate adapted versions.

Financial data visualization

Financial visualization has its own rules. A colorful bar chart that works for marketing may be completely inappropriate for a financial report.

The 5 essential financial charts

1. Waterfall (bridge): for explaining EBITDA, revenue, or cash variation. The most powerful chart for financial narratives.
2. Actual vs Budget (double bars): for comparing actual vs budget by month or line.
3. Trending (time line): for showing KPI evolution over time, with a target area.
4. Mix (donut or stacked bar): for showing the composition of revenue, costs, or assets by category.
5. Variance (horizontal bars): for ranking variances by business unit, product, or cost center.

Automated narrative generation

Financial narrative is the text layer that transforms numbers into actionable information. "Revenue was 5.2M EUR" is data. "Revenue grew 8% year-over-year, exceeding budget by 200K EUR, driven by the launch of product X which contributed 400K EUR in its first quarter" is information.

Types of automatable financial narrative

• Variance commentary: for each P&L line, a paragraph explaining the variance vs budget and vs prior year. The biggest time consumer for the controller and the most automatable.
• Executive summary: the 5 most important things of the month in one paragraph.
• Trend analysis: narrative on the evolution of key KPIs over recent quarters.
• Risk commentary: description of key financial risks with estimated probability and impact.
• Outlook: perspective for the rest of the fiscal year based on the current forecast.

Self-service analytics for finance

The holy grail of financial reporting is that end users (CEO, area directors, business managers) can query financial data without depending on the controller. AI makes this possible through natural language interfaces.

Natural language queries

Imagine the commercial director can ask: "how many sales have we closed in the northern region this quarter compared to the prior one?" and get the answer in 10 seconds, without requesting a report from the controller. Tools like Power BI with Copilot, ThoughtSpot, and similar already enable this.

Necessary guardrails

Financial self-service needs controls:

• Permissions: not everyone can see all data. A division director should not see the personnel costs of another division.
• Definitions: when someone asks about "sales," the answer must use the official definition (net sales, gross sales, billed sales, collected sales?).
• Audit: logging of what queries each user makes, to detect misuse or unmet needs.
• Disclaimer: data is provisional until the official close. Self-service provides operational data, not audited figures.

AI tools for reporting

Dashboards

• Power BI with Copilot: the corporate standard. Integrated AI for natural language queries, automatic insight generation, and narratives. From 10 EUR/user/month.
• Tableau with Einstein: powerful visualization, with AI for pattern and anomaly detection. Enterprise pricing.
• Looker (Google): ideal for companies using BigQuery. Integrated AI for data exploration.
• ThoughtSpot: leader in self-service analytics with natural language search. Enterprise pricing.

Report generation

• Claude + Python: for technical teams. Upload data, AI generates the complete report with narrative and charts. Cost: 20 USD/month.
• Narrative Science / Automated Insights: platforms specializing in automated financial narrative generation from data.
• Datarails: FP&A with automated reporting from Excel. From 1,000 USD/month.

Ejercicio practico

Puntos clave