A sales dashboard is useful when it lets you choose an action: call ten customers, resolve a late delivery or revisit an offer. Eight indicators defined without ambiguity are worth more than forty charts whose scope nobody knows.
Write a measurement contract before the first chart
For each indicator, set a formula, a source, the date it attaches to and an owner. “September revenue” can mean the orders entered, the invoices issued or the payments received. Those amounts answer different questions and must not be displayed under the same label.
To follow sales made, take invoices excl. VAT, net of credit notes, at their issue date, for example. To manage the order book still to be delivered, take confirmed, uncancelled orders, then subtract the quantities already shipped. Show the time of the last sync: a dip at midday may be nothing more than an import that is still incomplete.
Eight indicators and the decision each one informs
| Indicator | Definition to display | Use |
|---|---|---|
| Net invoiced revenue | Invoices excl. VAT minus credit notes over the period | Compare with the budget and the expected pace |
| Contribution | Net revenue minus the defined variable costs | Spot sales that are expensive to serve |
| Buying customers | Distinct outlets with a confirmed order in the period | Separate the breadth of the network from key accounts |
| Average order value | Net revenue excl. VAT of confirmed orders / number of those orders | Read the mix and the effect of minimum orders |
| First orders | Customers whose very first order falls in the period | Measure openings without counting duplicates |
| 90-day reorder rate | New customers who reordered within 90 days / cohort observable for 90 days | Assess the quality of the openings |
| Visit coverage | Targeted customers with at least one visit made / targeted customers | Reallocate field time |
| Late orders | Orders still to ship after the promised date | Prioritise sales admin exceptions |
These definitions are proposed reporting choices, not a universal standard. A company that counts invoiced customers or complete deliveries may adopt other rules. What matters is displaying them and keeping the same convention in exports, filters and meetings.
Read the mix before concluding that things have improved
An average order value can rise without any rep selling better. Fictional example: in the first month, 80 direct orders at €100 and 20 central buying orders at €1,000 give €28,000 for 100 orders, that is an average order value of €280. In the second, 50 direct orders and 50 central buying orders at the same values give €55,000, that is an average of €550.
For reorders, wait until the window is complete
An opening on 20 August cannot be assessed on 6 September against a 90-day reorder rate. Keep only the cohorts whose window has finished, or show recent cohorts separately, marked “still being observed”. Comparing customers observed for three months with others observed for ten days manufactures an artificial fall.
Compare territories that are comparable
The number of visits depends on travel time, the appointments accepted and the portfolio assigned. A dense area with 70 pharmacies does not necessarily represent the same workload as a rural territory with 70 pharmacies. Set activity, coverage and results side by side before ranking people.
A change of territory also raises a question of attribution: does the historical revenue follow the customer, or stay attributed to the rep of the time? Keep both readings if they serve different decisions, with explicit labels. The territory planning method gives an example of a workload calculation.
Show the limits of the data
- →Duplicate customers: the same outlet under two codes inflates openings and distorts reorders.
- →Incomplete history: the first order imported is not necessarily the first commercial relationship.
- →Late credit notes: a closed month can still move; say whether the figures are provisional or final.
- →Missing data: an unknown cost is not zero. Show the coverage rate of the contribution calculation.
The Sirene database distinguishes the legal unit, identified by its SIREN number, from its establishments, identified by their SIRET numbers. This distinction, documented by Insee, the French national statistics office (French), helps to choose the right level of counting. It does not remove the need to handle changes of operator or foreign customers.
Are the data comparable and complete enough?
- No / to be checkedGo back over the scope
Check dates, sources, credit notes, synchronisation and the composition of the groups.
- YesLook for the cause
Gather the facts, then name an action, an owner and a due date.
A change in an average can come from the mix of customers or products.
End the meeting with a list of actions
A weekly review can fit into four questions: which change is real, what is its likely cause, which check is missing and who acts by when? A spike in late deliveries calls for a sales admin action. A fall in buying customers calls for an analysis of the ordering pattern, then a follow-up adapted to inactive customers.
Keep the decision beside the indicator and check its effect at the next meeting. If a chart leads to neither a decision nor a check for several weeks, take it out of the main view. It can stay in the detail without occupying the whole team’s attention.
By Anatole Briant
By Anatole Briant
By Anatole Briant
COO of Salesia
Pierre Lerebours is COO of Salesia. He is your contact for implementation, training and ongoing support.