Cash Forecasting: Priorities & Context¶
Date: 2026-02-17 Domain: Cash Forecasting
Priority Ranking¶
Based on customer validation, frequency, and unsolved gaps — here's the order of cash forecasting jobs to tackle:
Top Priority (Confirmed — high urgency)¶
1. Explain and demonstrate forecast reliability to stakeholders - Sources: ON (4 sessions), Sonder, Personio (Confirmed) - Current status: Partial — Variance Analysis exists but no accuracy trends, no model performance dashboard, no plain-English explanations - Why it matters: "Can I trust it?" is the universal question. Until treasurers can show their CFO why the forecast is right (or wrong), they won't rely on it. This is the single biggest barrier to adoption and expansion. - Unsolved outcomes: Forecast version comparison, accuracy trends over time, model transparency
2. Produce a short-term cash forecast I can trust and act on - Sources: ON (5 sessions), Sonder, Levi's, Personio (Confirmed) - Current status: Partial — core forecasting works, but AP/AR integration missing, no "push button" weekly generation - Why it matters: The foundational job. Every treasurer wants to push a button and get a weekly forecast. Data gathering is the enemy; analysis is the value. Levi's: "My treasurer just wants a button... push a button. It spits out everything for the week." - Unsolved outcomes: AP/AR data integration, push-button forecast generation, data source transparency
High Priority (Confirmed — important gaps)¶
3. Configure forecasting models based on data quality and account characteristics - Sources: ON (2 sessions) (Confirmed) - Current status: Partial — ML on/off toggle exists, but no data quality guidance, no per-account model recommendations - Why it matters: Customers with 100+ accounts need control over which models run where. ON: "For this account, please deactivate machine learning. Or for this account, please make sure that ML definitely includes ARP, whereas for this one our ARP is shit." - On roadmap: Configure forecast models
4. Optimize use of operational cash - Sources: Personio (2 sessions), Sonder (Confirmed) - Current status: Partial — forecasting helps reduce buffers but no integrated investment-availability calculation - Why it matters: The downstream payoff of good forecasting. Sonder: "If you have too much you're losing out on interest income." Better forecast accuracy → smaller buffers → more cash deployed to investments. - Unsolved outcomes: Investment maturity integration, "available for investment" calculation
5. Forecast cash flows entity-by-entity across decentralized group structure - Sources: Euroports, Volvo Cars (Confirmed) - Current status: Mostly solved — entity-level forecasting works, batch uploads work, but no forecast collection workflow or forecast-vs-forecast comparison - Why it matters: Multi-entity complexity is one of the top sales themes. Volvo: "When it's on the Excel, anything can go wrong with it." 40+ entities, each forecasting differently.
6. Improve forecast accuracy by handling outliers - Sources: ON, Sonder (Confirmed) - Current status: Partial — one-off items feature exists, but no automatic outlier detection - Why it matters: One large atypical payment can skew ML models for weeks. Users must manually identify anomalies today.
Emerging — Needs Validation¶
| # | Potential Job | Source | Key Question |
|---|---|---|---|
| 7 | Apply judgment to stakeholder inputs | Sonder | How common is the "unreliable input" problem? |
| 8 | Detect missing items in entity forecasts proactively | Volvo | Would this create alert fatigue? |
| 9 | Incorporate external signals into revenue forecasting | Live Events | Niche or broadly applicable? |
| 10 | Understand AR/AP movements driving working capital | Euroports | Overlap with bridging domain? |
| 11 | Distribute monthly collections across working days | Personio | How many customers have monthly-only data? |
| 12 | Maintain coherent forecast across the organization | Sonder | Overlap with bridging domain? |
| 13 | Track influential customers for AR forecasting accuracy | Levi's | How many customers have customer-level AR data? |
Next Steps¶
- Forecast reliability (#1) is the unlock — without trust, nothing else matters. Invest in accuracy dashboards, version comparison, and plain-English explanations
- AP/AR integration (#2) is the "push button" blocker — treasurers want live data flowing in, not manual uploads
- Validate emerging signals — especially #8 (proactive detection) and #10 (AR/AP movements) which overlap with other domains