Francesco Gizzi
Data
How to choose the right business KPIs and cut useless reports
In many companies reporting grows by accumulation. Someone asks for a number for a meeting, the number goes into the report, the meeting passes and the number stays. After a few years the monthly report has become a twenty-page document that takes days of work to prepare and a few minutes of attention to read. Everyone knows this is the case, but nobody takes responsibility for cutting it.
The problem is not the amount of data. It is the lack of a criterion for deciding which numbers deserve the attention of the people running the company. Understanding how to choose business KPIs means building that criterion: a few indicators, each tied to a concrete decision and to a person who is accountable for it.
In this article I focus on the choice itself. If you are more interested in how to structure tracking and dashboards for management, marketing and sales, you will find everything in the article on KPIs and dashboards for companies that want to make better decisions. Here we talk about what to keep, what to drop and why.
Vanity metrics and decision metrics
The first distinction to make is between numbers that feel good and numbers that help you decide. Vanity metrics are the ones that go up easily, get shown off in meetings and say almost nothing about the health of the business. Decision metrics are the ones that, if they change, force you to do something.
Some typical examples of vanity metrics:
- number of followers or engagement on social media;
- total website visits, with no distinction between target and non-target audiences;
- total number of leads, without knowing how many are qualified;
- number of emails sent or calls made, without knowing what they produced;
- gross revenue read on its own, without margin or cash collected.
And some examples of decision metrics:
- qualified leads by channel;
- proposal-to-contract conversion rate;
- customer acquisition cost;
- margin by job or by product line;
- average days to collect payment;
- customers lost in the period and the main reason.
The test I use with clients is simple. For each number I ask: if this value doubled tomorrow, or halved, what would you do differently? If the answer is "nothing" or "it depends", that number is not a KPI. It can stay in an operational report for whoever runs the individual channel or department, but it should not take up management's attention.
There is one caveat, though. A metric is not a vanity metric in absolute terms. Website visits are a vanity metric for a CEO, but they can be a useful indicator for whoever manages content, if read together with traffic quality. The criterion is not the number itself, but who reads it and to decide what.
Leading and lagging indicators
The second distinction is about time. Some indicators tell you what has already happened. Others tell you what is likely to happen. A good set of KPIs needs both.
Lagging indicators
Revenue, margin, customers lost, quarterly profit. These are the numbers that really count, but they arrive when it is too late to change them. When you see a drop in revenue in the end-of-month report, the causes took shape weeks or months earlier. Lagging indicators are for checking, not for steering.
Leading indicators
Number of open deals and their value, quotes sent, first meetings booked, response rate to proposals, hours already scheduled for next month, customer complaints. These numbers are less "final", but they move earlier and give you time to act.
The practical rule I propose is this: for every lagging indicator you care about, identify at least one leading indicator that comes before it. If your goal is quarterly revenue, look at the value of deals in progress every week. If you are worried about losing customers, look at unresolved support requests or customers who have cut back their orders. Many companies look only at lagging numbers, and that is why they notice problems when they have already become expensive.
The method: link each KPI to a decision and an owner
A KPI with no decision attached is just an interesting number. A KPI with no owner is a number everyone looks at and nobody manages. That is why, when I work on choosing indicators, I build a short sheet for each one with these fields:
- Name and precise definition. How it is calculated, with what data, what is included and what is not. Two people calculating the same KPI must get the same result.
- Data source. Which system it comes from and who updates it.
- Target value and alert threshold. Above what level all is well, below what level you need to act.
- Linked decision. What you do if the value crosses the alert threshold. For example: shift budget, review prices, give a salesperson extra support.
- Owner. A single person, with a first and last name, not a department.
- Review frequency. Weekly, monthly or quarterly, depending on how quickly the number moves and how quickly you can react.
Filling in this sheet is already a filter. Many indicators do not get past point four: nobody can say which decision they are supposed to guide. Others do not get past point five, because nobody wants to own them. Either way, it is a useful signal.
Review frequency deserves particular attention. One of the most common forms of waste I come across is looking every week at numbers that only move meaningfully over a quarter, such as margin by product line. The result is that random fluctuations get discussed as if they were trends. Conversely, some leading indicators, such as open deals, lose much of their value if read only once a month. Matching the frequency to the speed of the number makes meetings shorter and discussions more sensible.
How many KPIs should you keep? There is no magic number, but management at an SME rarely needs more than eight or ten indicators in total. Each area head can have a few more for their own work, as long as they are linked to management's.
How to eliminate useless reports
Choosing the right KPIs is half the job. The other half is having the courage to remove the rest. This is how I usually proceed:
- Inventory. A list of all the reports produced in the company, with who prepares them, how long they take and who they are sent to.
- Interviews with recipients. For each report I ask the people who receive it what decision they have made thanks to that document in the last three months.
- Trial suspension. Reports nobody uses are suspended for a month, without any grand announcement. If nobody asks for them, they are eliminated.
- Consolidation. The remaining reports are reduced to one document per level: one for management, one for each operational area.
Trial suspension works very well because it avoids theoretical debates. Everyone tends to defend a report when you talk about eliminating it, but few notice when it stops arriving. The time freed up is often surprising, and it can be reinvested in analyzing the few numbers that matter.
A practical example
Take a hypothetical plant maintenance services company with about forty people. Every month the administration team prepares a report of more than twenty pages with some sixty indicators. It takes three days of work to put together. The monthly management meeting lasts two hours and almost always ends without clear decisions.
Applying the method, management arrives at eight indicators, grouped under six headings:
- margin by contract type, owned by the operations director;
- hours scheduled for next month compared with team capacity, also under the operations director;
- value of open quotes and acceptance rate, assigned to the sales manager;
- contracts expiring in the next 90 days, with the sales manager;
- jobs running late against the promised timelines, with the technical lead;
- average days to collect and overdue receivables, with the finance manager.
Each one has an alert threshold and a linked decision. If hours scheduled for next month drop below 80% of capacity, for example, sales focuses on expiring contracts and on the most mature open quotes. If days to collect rise above the threshold, the administration team starts chasing the main customers and management considers whether to revise payment terms.
The report becomes one page. Preparation goes from three days to half a day. The meeting gets shorter and, above all, every item closes with an action and a name. The other indicators do not disappear: they remain available in operational reports, for whoever needs them in day-to-day work.
Mistakes to avoid
- Choosing KPIs based on the data available. Start from the decisions to be made, then look for the data. If the data is missing, decide whether it is worth collecting.
- Copying other companies' KPIs. The right indicators depend on your business model and the stage you are at.
- Assigning a KPI to a department. When the owner is "marketing" or "sales", in practice there is no owner.
- Looking only at the past. Without leading indicators you notice problems when you can no longer avoid them.
- Never reviewing the list. The right KPIs today may not be right a year from now. A yearly review, or one whenever the strategy changes, is enough.
- Confusing control with distrust. KPIs are there to help you decide better, not to find culprits. If they are used to punish, the numbers soon stop being reliable.
Where to start
A good first step can be taken in an afternoon. Take the report management receives every month and, next to each number, write which decision it has guided in the last three months and who is accountable for it. The rows left blank will tell you a lot about how much of your reporting is actually useful.
If you want to go further and build a solid set of indicators, with shared definitions, thresholds and owners, that is the work I do in my data, KPI and dashboard consulting projects. When the issue is mainly the link between marketing, sales and margins, I also tackle it as part of management control for marketing and sales. Write to me and tell me how your company reads its numbers today: we will start from there.
FAQ
How many KPIs should an SME's management track?
There is no fixed number, but management rarely needs more than eight or ten indicators. Each area can have a few more, as long as they are linked to the main ones.
What is a vanity metric?
It is a number that grows easily and looks good, but does not lead to any concrete decision. Followers, total visits and the number of unqualified leads are typical examples.
What is the difference between leading and lagging indicators?
Lagging indicators, such as revenue and margin, measure results that have already happened. Leading indicators, such as the value of open deals, move earlier and give you time to act.
Doesn't eliminating reports risk losing information?
Usually not, if you do it through a trial suspension. If nobody asks for a report that has been suspended for a month, that information was not guiding decisions.