7shifts' Sales Forecast
7shifts' sales forecast has the ability to predict a restaurant's sales with only a few weeks of sales information. Integrate your POS to get ~95% accurate sales forecasts so you can build labor-perfect schedules that your team can trust.
What's in a forecast?
Our sales forecasting system is comprised of three separate algorithms, which forecast short-term (recent trends), mid-term (seasonal trends), and long-term trends (year-over-year).
7shifts' machine learning model considers multiple sets of data from different historical time periods to forecast the sales of each specific day.
Advantages of Machine Learning:
- Easily recognize sales patterns
- Reveal hidden correlations to decrease human-based bias
- Captures non-linear relationships
When will my Sales Forecast reach peak accuracy?
The accuracy of your forecast is a journey, not a one-time calculation. Because 7shifts uses machine learning, the model needs to "experience" your business cycles to predict them accurately.
The more historical data the system has, the more models it can activate to refine your numbers. Here is the typical timeline for a new location:
| Timeframe | Forecast Status | What the Model is Learning |
|---|---|---|
| 0–2 Weeks | Learning | No projections are shown while the system collects its first data points. |
| 2–4 Weeks | Foundational | The Short-term (Recent Trends) model activates to identify basic day-of-the-week patterns. |
| 1–6 Months | Improving | The Mid-term (Seasonal) model joins in, recognizing weekly patterns and gradual seasonal shifts. |
| 6-12 Months | Near Peak | The Mid-term model reaches full strength, stabilizing seasonal trends for your specific location. |
| 12+ Months | Maximum Accuracy | The Long-term model (Year-Over-Year), if activated, allows the system to account for full annual cycles. |
How accurate is 7shifts' Sales Forecast?
7shifts' data team performs an annual review of the sales forecast model to ensure our accuracy claim can be backed up by data. Our most recent data audit shows an average daily error rate of 5.51%.
It is important to note that "~95% accuracy" is a statistical average. While many restaurants can see even higher precision on a typical day, your individual accuracy may vary based on your business's consistency, the amount of historical data available, and external factors like weather or local events.
What can cause inaccuracies in 7shifts' Sales Forecasts?
In most cases, our forecast model accurately accounts for and anticipates restaurant sales variances. However, the model is most effective when sales follow clear, repeatable patterns. If a restaurant experiences high day-to-day variability—such as large, unpredictable swings in weekly revenue—the forecast will act as a helpful directional guide rather than a precise prediction. Additionally, sudden and significant changes in sales caused by unexpected weather events, sharp increases in business volume, or major adjustments to historical sales data may result in temporary inaccuracies. These discrepancies are typically short-lived and will self-correct as the model adapts and integrates the outlier data.
Will my POS sales sync time affect my forecast?
No. 7shifts' sales forecasting model is trained with daily sales numbers pulled from your POS. All of 7shifts' integrated POS sync their sales with 7shifts at a minimum of once a day.