DashboardSettingsHelp

Inaccurate Stockout Predictions

Fix issues with velocity calculations and improve prediction accuracy

Troubleshooting
7 min read

Troubleshooting: Inaccurate Predictions

If Stock Sentry's stockout predictions seem off, here's how to improve accuracy.

Common Causes

1. Promotional Spikes

Issue: Sale/promotion causes temporary surge in sales

Impact: Velocity artificially inflated, predictions too aggressive

Solution:

  • Temporarily increase alert thresholds during promotions
  • Use longer lookback period (30 days) to smooth spikes
  • Return to normal settings after promotion

2. Seasonal Products

Issue: Sales patterns vary by month/season

Impact: Off-season predictions inaccurate

Solution:

  • Enable seasonal learning in Settings
  • Manually adjust thresholds seasonally
  • Use 90-day lookback for seasonal context

3. New Product Launch

Issue: Launch week has abnormal velocity

Impact: First month predictions unreliable

Solution:

  • Wait 30 days for data to stabilize
  • Use conservative thresholds initially
  • Monitor manually during ramp-up

4. Irregular Sales Patterns

Issue: Product sells in bursts, not steadily

Impact: Velocity varies wildly day-to-day

Solution:

  • Use longer lookback period (30 days minimum)
  • Consider Shopify's low-stock alerts instead
  • Check "confidence level" indicator

Adjusting Lookback Period

The lookback period dramatically affects accuracy:

7 days: Best for trending products, very responsive

14 days (default): Good balance for most products

30 days: Best for stable, consistent sellers

90 days: Best for seasonal planning

To change:

  1. Go to Settings
  2. Find "Velocity Calculation"
  3. Select lookback period
  4. Save changes

Understanding Confidence Levels

Stock Sentry shows confidence for each prediction:

High (80%+): Trust the prediction

Medium (50-80%): Use with caution

Low (<50%): Treat as rough estimate

Low confidence indicates:

  • Insufficient sales data
  • Irregular pattern
  • Recent changes in velocity

Best Practices

For Accurate Predictions

  • [DO] Use appropriate lookback period for product type
  • [DO] Adjust thresholds before promotions
  • [DO] Monitor confidence levels
  • [DO] Review velocity trends weekly

What to Avoid

  • [AVOID] Short lookback for seasonal products
  • [AVOID] Ignoring confidence indicators
  • [AVOID] Same settings for all products
  • [AVOID] Dismissing all alerts as inaccurate

Still Having Issues?

If predictions remain inaccurate after adjustments, contact support with:

  • Product name and typical sales pattern
  • Lookback period being used
  • Screenshot of velocity chart
  • Recent order history

Was this article helpful?

Yes, this helpedNo, still need help