Short answer. Analysing your sales means answering six questions: what keeps me alive, who do I depend on, is the month good, who has left, what did I miss, where does the margin go. Each has its reading: rank by margin and not by revenue; compare like periods, not like dates, because Ramadan moves about eleven days earlier each year; count the customers who do not come back. And each requires a piece of data: the purchase cost on every sales line, the customer identified, a single record per product. Six weeks in the example to go from a monthly total to six readings.
This page is for the manager of a trading, distribution, manufacturing or service company that has software full of figures and few answers, and for the founder based in France who wants to read, from afar, what their company in Algeria really sells. The definitions come from public sources: Pareto principle, ABC analysis, gross margin, seasonality, the Hijri calendar, same-store sales, moving average, cohort analysis, customer attrition, stockout, star schema, fact table, data quality. The eight products, their indices, the twelve weeks and the calendar are a constructed example. This page does not cover accounting or tax: the margin discussed here is a management margin, and the accounts belong to your accountant. The real-time dashboard, its freshness and its alerts are the subject of another guide.
Six questions, and the reading that answers each
The figure pairs each question with its reading and the data it requires.
A total answers nothing. “Twelve million this month” says neither where it comes from nor what is left of it. The six questions in the figure cut that total along one axis at a time: by product, by customer, by period. The Pareto principle. For many outcomes, roughly 80% of consequences come from 20% of causes; it is an adage of business management that 80% of sales come from 20% of clients. It is not a law: it is a first reading, to be checked on your figures. It says where to look first.
Three classes. ABC analysis divides items into three categories, from the most important to the least: applied to sales, it separates the handful of products that weigh, the middle, and the long list that fills the shelves. One question, one decision. A reading that changes no decision does not deserve a screen. “Who do I depend on?” serves to decide whom to visit; “who has left?”, whom to call back, with the tools of the guide to CRM.
The same products, two rankings
The figure ranks eight products by revenue, then by margin.
Revenue is not gain. Gross margin is the difference between revenue and the cost of what was sold, divided by revenue. In the example, the best seller has a margin of 6%: it takes up the warehouse, the drivers and the cash, and earns less than five others. Three products out of eight make almost 60% of the margin. What to do with it. Push what earns, renegotiate the purchase of what keeps you busy, and stop celebrating a sales record made on a 6% product.
The condition. To rank by margin, every sales line must carry its purchase cost on the day of the sale. It is the data most often missing: the software knows the selling price, not what the item cost. Discounts. The second leak is the discount: it is read by seller and by customer, by comparing the list price with the price sold. A promotion that swells sales can destroy the margin, as the guide to promotions shows.
«Ranking by revenue looks at the work, not at the gain.»
Analysing your sales, in six steps
The second step is accented: complete every sales line.
Clean, complete. A product entered under two records has its sales cut in two, and will be first nowhere; a customer in three copies will never appear in the top twenty. Data quality, accuracy, completeness, consistency, is settled before any analysis: that is the work described in the guide to the product record and in the guide to taking over your data. Rank, compare. Then the two basic moves: rank along an axis, compare two periods that resemble each other.
Explain, decide. A gap is not an explanation. A fall may come from a stockout, an event that causes inventory to be exhausted, and which the sales figure does not show: you do not sell what you do not have. Noting the requests not served is the only way to see it, along with a correct stock, the subject of the guide to stock. One action per reading. Each reading ends with a decision, a name and a date; otherwise analysis becomes one more report.
A sales analysis, in five layers
The accented layer is the sales line.
The sales line. Everything starts from it. In data warehousing, a fact table consists of the measurements of a business process, here one line per item sold; the star schema, the simplest and most widely used, links that table to dimension tables: products, customers, sellers, time. That arrangement is what lets the same sale be grouped by product in the morning and by customer in the afternoon. Records. A dimension is worth what its records are worth: a family per product, a city per customer.
One definition per figure. “Revenue”: returns deducted or not, at order or at delivery? The definition is written once and applied everywhere, otherwise two screens give two totals. Where the lines come from. From management software that really records sales: when there are four that do not talk to each other, analysis starts with a single base. Every morning. The readings on this page are computed overnight; what must be seen within the hour belongs to the dashboard mentioned above.
Comparing the right week
The figure compares twelve weeks of sales over two years.
Seasonality. It refers to the trends that occur at regular intervals of less than a year: weekly, monthly, quarterly. In Algeria, part of the season does not follow the solar calendar: the year of the Hijri calendar has twelve months of 29 or 30 days, 354 or 355 days in all, about eleven days fewer than the solar year, so its months cycle in reverse through the seasons. Ramadan, the ninth month, starts earlier every year. What that breaks. Any comparison of “this month against the same month last year” that touches Ramadan or a holiday.
Like for like. Chains with several shops compare same-store sales: the revenue of the same outlets over a given period. The same care applies to time: the second week of Ramadan against the second week of Ramadan, the back-to-school week against the back-to-school week, and the same number of working days. Smoothing. A moving average, which replaces each point with the average of its neighbours, removes the noise of a public holiday. Who has left. Cohort analysis breaks customers into groups, for example those won in the same year, and follows each one: it shows attrition, the loss of customers, which the total hides when new ones arrive.
«You do not compare two dates: you compare two periods that resemble each other.»
Six weeks to the first reading, in 2027
Six weeks in the example, the first to open the data.
- What the manager receives: the inventory of what the sales record and what is missing; product and customer records with no duplicates; a sales table fed every night; six readings, including margin per product and the like-for-like comparison; the calendar of the business; a weekly appointment.
- What the company no longer does: celebrate a revenue record made on a product with no margin, compare March with March in a Ramadan year, find out at year end that a big customer has left, redo the same table by hand every month.
- What is measured: share of sales lines carrying a cost and a customer, products and customers that make 80% of the margin, like-for-like change, last year's customers with no purchase, requests not served, average discount per seller.
- What Symloop does: the data inventory, cleaning the records, carrying the cost and the customer on the lines, the sales table and its axes, the six readings, the business calendar, the connection to your management software or its overhaul, then the first reading with you.
Start by looking. The first week produces no screen: the software's sales table is opened and what a sale really records is noted. Is there a customer, or “counter customer” everywhere? A cost, or an empty field? A seller? That list decides the rest. The longest part. Carrying the cost and the customer on the lines, for the past when possible, for the future in every case: one entry rule changes, and analysis becomes possible three months later.
The first reading. The six readings are read once with the manager, figures in hand: what surprises is checked line by line before it is believed. Then one decision per reading. Afterwards. A fixed appointment, Sunday morning, twenty minutes, always the same six readings: it is regularity, not the tool, that makes a company know its sales.
- How do you analyse a company's sales?
- By asking six questions and reading, for each, a ranking or a comparison: products by margin, customers by revenue and by margin, the month like for like, customers who do not come back, requests not served, discounts. Before that, check that every sales line carries a cost and a customer.
- Which indicators should be followed to analyse sales?
- Few, but always the same: revenue and margin, by product and by customer; like-for-like change; number of active and lost customers; average discount; requests not served. Six readings read every week are worth more than thirty looked at once.
- How do we know which products earn the most?
- By ranking them by margin and not by revenue. That requires the purchase cost on every sales line. In this page's example, the best seller is only sixth by margin, and three products out of eight make almost 60% of it.
- How do you compare a month with the previous year when Ramadan changes date?
- By comparing periods that resemble each other, not dates: the second week of Ramadan with the second week of Ramadan, back-to-school with back-to-school. The lunar year has 354 or 355 days: Ramadan starts about eleven days earlier each year.
- How do we know which customers have left?
- By listing the customers who bought last year and nothing since a set delay, for example three times their usual rhythm. That assumes the customer is identified on every sale. The sales total hides these departures as long as new customers arrive.
- Can sales be analysed with Excel?
- Yes, to start with: an export of sales and a pivot table give the first rankings. The limits come quickly: duplicate records, missing cost, work to redo every month. A tool takes over when the reading becomes weekly.
- What must the data hold to analyse sales?
- On every sales line: the product, the customer, the date, the quantity, the price sold, the purchase cost and the seller. Around it: a single record per product and per customer, with their family. Without the cost, no margin; without the customer, no loyalty.
- What is the 80/20 rule in sales?
- It is the Pareto principle: roughly 80% of consequences come from 20% of causes, and it is often said that 80% of sales come from 20% of clients. It is an order of magnitude, not a law: work out the real proportion on your figures.
- Is a business intelligence tool needed?
- Not first. The tool does not create the missing data: start with the cost and the customer on the lines, then six simple readings. A charting tool becomes useful when those readings are stable and several people consult them.
- How long does it take to set up a sales analysis?
- Six weeks in this page's example: one to inventory the data, two for records and costs, then the sales table, the six readings and the business calendar. The margin on the past depends on the costs that can be recovered.