Business analytics for small business: the numbers worth tracking

Most small businesses are not short on data. They are short on data they believe. When the sales figure in the spreadsheet does not match the POS, and neither matches what the owner remembers, nobody uses either number and decisions go back to instinct.

Fix trust before you fix reporting

Analytics built on data that three systems disagree about produces confident nonsense. Before any dashboard, settle where each number comes from. One system owns sales. One owns stock. One owns customers. Every report reads from those, and nobody maintains a parallel spreadsheet on the side.

This is unglamorous and it is the whole job. Once it is done, reporting is almost free. Skip it, and you will spend a year arguing about which figure is right.

The six numbers that earn their place

1. Revenue by product or service

Not total revenue — the breakdown. Almost every business finds that a small part of the catalogue produces most of the income, and that some products they promote heavily contribute almost nothing. The decision this changes: what to feature, what to reorder, what to quietly discontinue.

2. Gross margin by product

Revenue without margin is a trap. Your best-selling item can be your worst earner once you count landed cost, packaging and the shipping subsidy you offer to compete on a marketplace. Track margin per line, and check it again whenever supplier prices move.

3. Cost to acquire a customer

Total marketing spend for the period divided by new customers gained. Simple, and often uncomfortable. Compare it against what a customer is worth to you over a year. If acquisition costs more than the first purchase yields, you are buying revenue rather than earning profit — which is fine deliberately and fatal accidentally.

4. Repeat purchase rate

What share of customers buy a second time, and how long the gap is. This single number tells you whether you have a business or a series of one-off transactions. It also tells you where to spend: if repeat rate is low, more ad spend just pours customers into a leaking bucket.

5. Stock turnover

How fast inventory converts to sales. Slow-moving stock is cash sitting on a shelf in Cainta while you are short on the item everyone actually wants. For any business holding goods, this is usually the fastest route to freeing working capital without borrowing.

6. Response time to inquiries

Minutes from a message arriving on Messenger, Viber or your website form to a real reply. In a market where customers message three suppliers at once, this is often the deciding factor in who gets the order — and it is the number most SMEs have never measured.

Where these numbers come from

• Revenue and margin — your POS or order system, provided cost price is recorded per item.
• Acquisition cost — ad platform spend plus any agency fee, matched against new customer records in the CRM.
• Repeat rate — the CRM or order history, which requires customers to be identified consistently rather than entered as walk-in every time.
• Stock turnover — the inventory system, using opening and closing stock over the period.
• Response time — your assistant or support platform, timestamped at message received and first human or automated reply.

A number that does not change a decision is decoration. Before you add anything to a dashboard, name the decision it will change and who makes it.

Why most SME analytics projects fail

1. 1.Too many metrics. Forty charts means nobody looks at any of them. Six is enough to run a small business well.
2. 2.No owner. If a number is not assigned to a specific person, nobody acts when it moves.
3. 3.Manual assembly. A report that takes four hours to build gets produced monthly at best, and skipped entirely in a busy month.
4. 4.No baseline. Without last quarter's figure, this quarter's number is trivia rather than information.
5. 5.Reporting on vanity. Page views and follower counts feel productive and rarely change what anyone does on Monday.

Build the habit before the dashboard

Set one thirty-minute review each week. Same time, same six numbers, one person presenting. Ask two questions of each: what moved, and what are we doing about it. Write the answer down so next week you can check whether the action worked.

A printed sheet reviewed every week will do more for your decisions than an expensive dashboard nobody opens. The tooling matters far less than the meeting.

When to add prediction

Forecasting demand, flagging customers likely to lapse, suggesting reorder quantities — these are genuinely useful, and they need history to work from. Twelve months of clean data in one system is the practical entry point. Before that, you are predicting from noise. Get the six numbers reliable first; the predictive layer becomes straightforward afterwards.

Start with what you can measure this month

Pick two of the six. Get them accurate and automatic. Add the rest one at a time as the underlying systems come online. Our builds include the admin panel and reporting as standard, along with training and documentation so your own team can pull answers without waiting on anyone, and a free lifetime 24/7 AI assistant that captures the response-time data from day one.