When sales fall: how AI data analysis from Torgsoft shows where you are losing money

When sales stagnate or start to decline, intuitive decisions usually don’t work. “Guesswork” discounts, random promotions, or cutting the assortment often only make things worse. Data-driven analytics from Torgsoft lets you look at the business without emotions: understand what stopped selling, when it started, and where the business is really losing money — in prices, discounts, stock levels, or inactive customers.

Using artificial intelligence to analyze this data gives the business owner specific answers rather than generic advice. You can identify products that drag turnover down, customers who have almost stopped buying, and inventory that “freezes” cash in the warehouse. This allows you to act precisely: launch the right promotions, adjust prices, clear the warehouse, and win back customers without chaotic experiments. That’s why well-prepared Torgsoft reports become the basis for decisions that help you get out of a downturn and restart growth.

To get high-quality analytics with artificial intelligence (for example, ChatGPT, Claude, or specialized BI tools), you should export data from the Torgsoft accounting software that contains historical information about sales, customer behavior, and inventory movement.

Below is a list of key reports and data that work best for AI analysis, along with instructions on how to obtain them.

1. Sales and profitability for forecasting and trend analysis

This is the most important dataset for AI. It enables analysis of seasonality, margins, and pricing efficiency.

• Report: “Profitability of sales for the period” (located in the Analysis menu).

• What data it contains:

• Sale date (important for trend analysis by days/months).

• Product name, SKU, barcode.

• Quantity sold.

• Selling price (actual sale price).

• Cost of goods sold (important for calculating net profit).

• Discount amount (allows AI to evaluate promotion effectiveness).

• Profit and profitability.

• What AI can analyze: seasonal demand spikes, best-selling drivers, products with declining margins, discount effectiveness.

2. Customer base and behavior for segmentation and personalization

Data for building a customer profile and predicting churn (Churn Rate).

• Report: “Customer activity analysis” or “Customer purchase analysis” (the Analysis menu).

• What data it contains:

• Customer full name (or card ID).

• Number of receipts and total sales amount.

• Date of the last purchase (critical for RFM analysis).

• Average receipt.

• Phone/Email (useful for marketing strategies, but it’s better to anonymize for AI).

• What AI can analyze: run RFM analysis (Recency, Frequency, Monetary), identify VIP customers at risk of leaving, and suggest personalized recommendations.

3. Inventory and turnover for warehouse optimization

Helps avoid overstocking (“frozen money”) or shortages.

• Reports:

• “Inventory turnover statement by quantity” (the Warehouse menu): shows opening balance, receipts, issues, and closing balance for the period.

• “Slow-moving inventory analysis” (the Analysis menu): shows products with no movement for a specified period.

• “ABC and XYZ analysis”: classification of products by importance (ABC) and demand stability (XYZ).

• What data it contains: product name, stock balance, date of last receipt, period without sales.

• What AI can analyze: detect “dead stock,” calculate an optimal purchase volume, predict the date when an item will run out (stockout prediction).

4. Receipt analytics for cross-selling

Analysis of product compatibility in the basket.

• Report: “Product report for the period” (grouped by receipts) or “Analysis of number of receipts by amount”.

• What data it contains: contents of each receipt (which products were bought together), purchase time, receipt total.

• What AI can analyze: find patterns like “Product A is often purchased with Product B” (Market Basket Analysis) to create promo bundles or recommendations for sales staff.

How to export data to Excel for AI correctly

For AI to process the file correctly, it must be “clean” (structured).

1. Open the required form (for example, Analysis - Profitability of sales for the period).

2. Configure the columns. Click the column visibility settings button (usually at the top left of the table) and leave only the data needed for analysis. Extra columns can confuse AI or exceed token limits.

3. Export. Click “Report” (printer or Excel icon) -> “Data export settings”.

4. Important setting. In the export window, choose “Vertical layout of subordinate tables” or a simple table export (without complex headers and merged cells) to get a flat table. AI works best with the format: one row = one record (transaction/product).

5. Save. Select “Create an Excel document”.

Tip: if you use AI to analyze dynamics, export at least 12–24 months of data (for example, via the “Period” report or “Comparative report by months”) so the model can detect seasonality.