AI & Predictive Analytics

AI & Predictive Analytics

NexusRMS offers machine-learning-powered predictions to help you make smarter decisions about equipment purchasing, client retention, and revenue planning. This feature requires the Analytics & AI addon (£79/month).

Enabling AI Analytics

To activate AI features:

1. Ensure the Analytics & AI addon is active on your subscription (check Settings > Billing > Addons)
2. Navigate to Settings > Analytics to verify AI features are available
3. The AI Insights widget will automatically appear on your dashboard once the addon is enabled

If AI predictions show as unavailable, confirm your addon subscription is active and in good standing.

Equipment Demand Forecasting

The demand forecasting engine uses a Prophet ML model to predict future equipment needs. Forecasts are generated for three periods:

• Month — next 30 days of projected demand
• Quarter — next 90 days of projected demand
• Year — next 12 months of projected demand

Forecast results are cached for 24 hours to ensure fast loading. The model analyses the following factors:

• Historical utilisation patterns across your equipment catalogue
• Seasonal demand fluctuations (e.g., summer event peaks)
• Current project pipeline requirements
• Market trends derived from booking velocity

Purchase Recommendations

Based on demand forecasts, NexusRMS generates purchase recommendations displayed in a table with the following columns:

Column	Description
Equipment Item	The item name and category
Current Quantity	Number of units currently in your inventory
Forecasted Demand	Predicted units needed for the forecast period
Shortfall	Gap between current stock and forecasted demand
Recommended Purchase	Suggested quantity to purchase
Estimated Revenue Impact	Additional revenue if shortfall is fulfilled
Priority	High, Medium, or Low based on demand urgency
ROI (Months)	Estimated months to recoup the purchase investment

Client Churn Prediction

The ChurnRiskCalculator identifies clients who may be at risk of leaving. Each at-risk client is assigned a churn probability score (0–100%). Factors include:

• Declining booking frequency compared to historical average
• Decreasing project values over recent periods
• Reduced communication or engagement
• Late payments or outstanding invoices

At-risk clients appear in a dedicated list, sorted by churn probability. Use this to prioritise retention outreach before clients disengage.

Revenue Forecasting

The PredictiveRevenueCalculator projects future revenue based on your current pipeline, historical trends, and seasonal patterns. Revenue forecasts include:

• Projected revenue for the next month, quarter, and year
• Confidence intervals showing best-case and worst-case scenarios
• Trend analysis comparing forecast against the same period last year

AI Insights Dashboard Widget

The AiInsightsWidget on your main dashboard surfaces the most important predictive insights. Each insight displays:

• A plain-language summary of the prediction
• A confidence score shown as a percentage badge (e.g., 87% confidence)
• The data period and factors analysed
• Recommended actions based on the prediction

Confidence Badges & Model Accuracy

Each prediction displays a confidence badge indicating model reliability. Higher confidence scores mean the model has more historical data supporting the prediction. Scores below 60% are flagged as low confidence and should be interpreted cautiously.

Manual Model Retraining

If your business patterns have changed significantly (e.g., new market segment, major client acquisition), you can trigger a manual model retraining. Navigate to Settings > Analytics and select Retrain AI Models. Retraining uses all available historical data and typically completes within a few minutes.

How Predictions Improve Over Time

AI predictions become more accurate as NexusRMS accumulates more data. New accounts may see lower confidence scores initially. After 3–6 months of consistent usage, prediction accuracy typically improves substantially as the models learn your organisation’s specific patterns.