Start with a question.
Use Kalopsia’s reports to understand your restaurant, test possible explanations and choose the next decision. This guide connects the evidence with the questions management needs to ask.
A worked example
Sales are down. Where should you look first?
For the same period and operating scope, compared with a suitable baseline:
Sales value = orders × average order value
- Sales value
- −8%
- Order count
- −11%
- Average order value
- +3%
Fewer orders are the main change. A higher average order value softens the fall in sales, but it does not tell you why orders fell.
Is the order decline widespread, or concentrated in a location, channel or time?
Illustrative, rounded figures. Fewer orders do not necessarily mean fewer distinct customers. The cause still needs investigation.
Choose your starting question
Demand and customers
Kalopsia’s existing reports and operational views support these questions. Add local context and specialist evidence to complete the picture.
Follow demand from locations and products to first purchases, returning customers and the activity behind sales.
Where is demand changing?
Break the result down by location, channel and time. Establish whether the difference is widespread or concentrated, then check affected orders. Compare several explanations: demand, availability, scheduling, delivery problems or restaurant execution.
Locating the change narrows the investigation; it does not establish its cause. Orders are transactions, not people.
Start in Kalopsia: Sales · Orders
What are customers choosing differently?
Compare absolute sales value, quantity sold and product or category share for the same period and scope. A growing share may simply reflect a decline elsewhere. Test price, promotion, availability and preference separately.
Value, quantity, share and profitability are different measures. A top seller is not necessarily the most profitable product.
Start in Kalopsia: Products · Categories
Do registrations become purchases?
Separate account validation from the first qualifying purchase. Compare registration groups after the same number of days, distinguishing customers who completed that window from those still pending. Check first, second and third purchases separately.
Pending customers have not failed to convert. A first purchase establishes purchase activity, not loyalty.
Start in Kalopsia: Clients: Activation · Cohorts
Are customers coming back?
Follow second and later qualifying purchases, comparing groups that have had the same time to return. First-time and returning counts for one period can overlap if someone buys more than once. Consider purchasing outside the registered-client population.
Returning purchasers in one period are not a long-term retention rate. Even a second purchase does not prove loyalty.
Start in Kalopsia: Clients: Overview · Activation · Cohorts
Who is behind the sales activity?
Compare registered and anonymous orders, sales shares and average ticket. A higher registered sales share with unchanged order share may reflect larger tickets. Check population filters carefully: customer-profile scope differs from sales-location scope.
Recorded orders and sales do not measure distinct people, visits or retention, or prove a campaign’s effect.
Start in Kalopsia: Engagement · Clients
Choose your starting question
Operations and business decisions
Test whether the operation delivers reliably and whether activity makes commercial and economic sense.
Can the restaurant deliver its promise?
Review availability, order progression, cancellations, reversals and feedback. Narrow the issue to a location, channel or service pattern. Use each exception to test a specific explanation, checking the relevant orders and customer comments before choosing an action.
Good sales can conceal a deteriorating experience. Different exceptions should not automatically become one diagnosis.
Start in Kalopsia: Orders · OrderHub · Feedback
Are higher sales making us better off?
Contribution is revenue less represented direct product costs, before other business costs. Check approved cost definitions, evidence completeness and product mix. Add accounting, payroll and banking information where needed for the full picture.
Revenue is not net profit or cash. A product margin view cannot supply the complete financial answer.
Start in Kalopsia: Sales · Product Margin
Did the commercial action achieve its objective?
Define the intended result and affected group. Compare outcomes with a suitable baseline; ask what might have happened without the action. Consider discounts, displaced sales and side effects. Decide whether to continue, change, stop or investigate.
Associated order value does not establish causation, incremental demand or profit. Define how the decision will be verified.
Start in Kalopsia: Promotions · Sales · Clients
What is client credit telling us?
Where client credit is used, review movements and the outstanding balance separately. Check which accounts and period are included, then reconcile unexplained movements with orders. Historical balances may be estimates and depend on complete history. Use aggregate evidence first.
Credit held is not cash, profit, additional sales or proof of loyalty. A balance discrepancy needs reconciliation.
Start in Kalopsia: Client Credit · Orders
Choose your starting question
Signals and persistence
Use signals and trends to focus investigation, then bring evidence and local context to human review.
What might I be missing?
Treat an unusual signal as a starting point. Check evidence quality, materiality and where it appears, then test explanations and missing context. An evidence-quality score is not a probability or a root cause.
Unusual does not automatically mean bad. Intelligence directs attention; people investigate, judge and decide.
Start in Kalopsia: Intelligence inbox · Sales · Orders
Is the decline persistent or local?
Compare transactions, sales and average ticket across completed, comparable weeks. Check which weeks are included and whether the pattern also appears at comparable locations in the same company. Excluded weeks are not zero activity.
A persistent trend does not establish its cause. Reviewing a warning is a separate management step.
Start in Kalopsia: Sales trends
A repeatable method
A method for every review.
Move from observation to decision, then return to the evidence.
Observe
What happened, in which period and scope?
Compare
What is a fair comparison?
Break it down
Where does the difference appear?
Explain
Which explanations fit, and what contradicts them?
Decide
Does this require action, investigation or continued observation?
Verify
How will we know whether the decision worked?
A breakdown locates a difference; it does not prove its cause. Test more than one plausible explanation.
Before you decide
Check these four things before you decide.
Period
Is the time window correct, complete and comparable?
Scope
Are the locations, channels and customer groups consistent?
Evidence
Is the information sufficiently complete for this conclusion?
Context
What operating changes or local events are missing?
Choose an action that fits the evidence, name the responsible person and agree when and how to review the result. Further investigation can be the right next step.