What Most Companies Get Wrong, and How to Get Metrics Right

If you’ve been around corporate metrics programs, you’ve probably heard the horror stories. The fast food chain whose employees started making every chicken sandwich to order because management began measuring the amount of waste. Sounds good, but order time and wait time went up exponentially, and customer satisfaction scores dropped precipitously. Or the more common: the customer service rep who dumps callers abruptly after helping them in order to meet call volume targets.

It's Corporate Whac-a-Mole. Solve one problem, and another pops up. And that’s the dirty little secret: most metrics programs have become meaningless exercises in paper shuffling that either don't deliver the desired results or, even worse, deliver unintended bad results in another area.

Why are metrics so hard to get right? Your car has a metric, and it encourages exactly the behavior it should. When the gas gauge alerts you that you are low on fuel, 99% of people (save the teenage population) will find the next gas station and refuel. Or, they'll watch this important metric and refuel before it signals an immediate need. So what can we learn from the fuel tank metric? When a metric measures something simple, does it clearly, and is very closely tied with a negative outcome, it will have a clear effect on behavior.

A corporate metrics program should do the same: alert your managers to issues, so they can be fixed before they become problems.


There are several reasons metrics programs don't achieve their desired results, and chief among them is the guy in the mirror. It's basic human nature. When managers set their own departmental metrics, they are likely to set the hurdle at a height they know they can clear. Beyond this natural fear of accountability – which is not to be underestimated – there are several supporting reasons metrics often go awry: they’re too complex, they can't be attributed to one group, or they just don't measure the right thing. Let's pick these apart.


Most metrics are owned by the CFO or the COO, who is setting the agenda based on what he or she is rewarded on, so the entire company dashboard becomes financial in nature. Managers will say they measure Return on Invested Capital (ROIC) or Net Present Value (NPV), but they have no idea how their day to day activities impact those metrics. ROIC and NPV are ethereal numbers, culled from complicated calculations and multiple spreadsheets, in a fire-drill of a quarterly process with an end result that means nothing to anybody but the CFO.

Take, for example, the industrial conglomerate that measured its financial success with a complex metric called "See-FROG-See" - [CFRoGC] Cash Flow Return on Gross Capital. That metric, the result of intense efforts to account for risk adjustments and cross currency aggregations, was impossible for employees to understand and couldn't guide decisions or actions.

Better to break these complicated metrics down to their component parts - the exact balance sheet or income statement line item they correspond to (Revenue, COGS, SG&A, Working Capital, Fixed Capital Efficiency, or Effective Tax Rate). Then, hunt down the specific processes that drive those numbers, and measure these, because they are your "value levers." Making sure everything that is measured links to higher level operational and financials metrics owned by the executive is the key to having the right information to drive decisions.

The best financial metrics accurately track progress toward specific goals, so wise goal-setting is key. Review last year's actual performance, then set goals for income, cost of goods, gross margin, operating expenses, and net profit. Set goals grounded in the reality of past performance, making sure your monthly financial statements are accurately adjusted for work in process, and summarizing your statements in a budget tracking report.


Which brings us to another point. A lot of companies use metrics to measure financials. But if your metrics program doesn't have operational or business process metrics in addition to financial ones, it's hard to tell how the business is doing.

Suppose you have a complex operational metric, like Customer Satisfaction Index, which has multiple embedded nuances like churn or repeat purchases in it. You must be able to find out what actually causes your customers to be satisfied. Is it on-time delivery (OTD)? Or is it delivery against customer's expectations? The two can be very different.

If you tell a customer you'll be there between 12 and 4, and you arrive at 4, that may feel like 100% OTD to the company, but it certainly doesn't feel that way to the customer, who blocked off that four-hour slot, and has already tweeted four times how angry she is. From her point of view, she blocked off 12-4 for your visit; now she has to stay an extra hour or more while the actual repair work is done.


A frequent complaint we hear from managers when it comes to metrics is that they are so inconsistent: "One cycle inventory is too high, the next cycle we have significant stock-outs…I can't drive processes to an optimal outcome."

A metric is an absolute measure. But no one action in an enterprise is isolated from all other actions. A single isolated metric can lead to some very undesirable outcomes.


For example, a recent client was witnessing wild swings in inventory as a result of periodic overcorrections. One month, manufacturing was told they were being measured on productivity, so they expanded batch sizes and produced at record levels. Of course that created a very high level of inventory, which sales couldn't offload, so they were given short-term incentives to dump excesses into the discount channels in order to save working capital from being adversely affected that quarter. They were in a massive cycle of overcorrection between sales and manufacturing.

How did they resolve the problem? They agreed that Manufacturing and Sales would jointly own an Inventory metric. Manufacturing would still measure productivity and sales would still measure its team's performance, but the two would have to collaborate in the Sales & Operations Planning process to get inventory levels just right.

Like quantum physics, all processes in an enterprise are interconnected, and so are the metrics. There is no individual metric that can really be controlled by a single manager. So causal linkages in a metrics hierarchy should be well understood. Otherwise, one individual's actions or operational decisions may have an undesirable impact on another area of the same process.


On the opposing page, we offer a simple diagram for selecting the right metrics, ensuring ownership, and making sure they encourage the right behaviors.


Any measurement strategy must address what it is measuring (process inputs and outputs), who it is measuring (process owners and participants that can impact the outcome) and how it is being measured (through appropriate use of information and technology). Today, in many organizations, metrics reporting is considered a chore. This framework recommends automating the delivery of metrics - elevating it to the status of a critical business capability.

Using technology to enable stable and reliable data acquisition is the last (but a fundamental) piece of the metrics puzzle. Ultimately, alignment with strategy implies a strict prioritization - measure what is important and what directly impacts strategic objectives - not more, not less. Let's go clockwise around the diagram.



The biggest mistake companies make is that they start with a metric they want to measure rather than a business objective. A classic example is the one where an airline wants to cut costs, so it measures fuel used, and pilots adjust their behavior by flying more slowly and leaving the air conditioning off. This drives on time arrivals down and causes customer satisfaction scores to plummet. If all the customers begin flying another airline, lower fuel costs won't matter.

A better approach is to start with the key business objectives. Then identify the processes that drive those objectives. You'll want to also note the "customers" of each process, and what their interests are before you move on to selecting metrics. So the airline above would have realized that a metric on fuel usage might impact customer experience or on-time arrivals, and would instead have found where fuel is wasted, and set a metric around that.

Select metrics that measure the behaviors you want to change or monitor. Always ensure that the trade-offs implicated and the associated metrics hierarchy are aligned.

A key approach to selecting the right metrics is to ensure that you are measuring the critical customer requirements/outcomes from the process in question. Keep the metrics program simple with few metrics to drive focus. A good rule of thumb is to monitor a minimum of three and a maximum of five metrics on a dashboard. You could have additional metrics that are relevant in special cases, but three to five metrics are adequate to help you track the steady state performance of a process or operation. In fact, every business has one "magic" metric that holds the key to its performance. It may be sales revenue per transaction, returns, page views, or percentage of growth, but chances are, the CEO and Board unconsciously know what it is.Select metrics that ultimately measure value to the shareholders and customers.

How does the metric you own roll up through the organization to produce shareholder value, and how can you influence it?

Once you understand that relationship, you can drive behavior correctly. Select metrics that not only provide a historical view of performance but also include leading indicators. A historical view and context of the metric is critical to understanding what the metric is telling us, and therefore what action to take. Intuitively, you would know that holding too much inventory is bad. But a historical view would allow a team to determine how much is too much, and whether the inventory level is in fact rising with time.

Be open to evolving and refining the metrics from one period to the next in line with changing priorities and capabilities of the organization. Metrics should change to accommodate your team’s changing focus or your company's shift in strategic priorities. Of course, you don’t want wild swings in what you measure, or you'll never have a baseline or something to compare progress against. But you need to make sure each metric is not encouraging a similar tradeoff in behavior somewhere else.



Select metrics that align the organization with strategic priorities by linking CXO metrics with those of CXO-1 level, and -1 level metrics with -2 level, and so on for the rest of the organization. Conversely, think about your own performance metrics. If they do not directly enable your supervisor's metrics, then you may not be completely aligned with the organization. Ensure ownership and accountability for each metric. If a metric has no clear owner, consider taking it out of the program. Why measure something if no one is responsible for changing it?



Don't try to measure anything for which you don't have accurate and timely data available. If you cannot support metrics program with the right technology, it will eventually fail. So start by measuring what you can, and evolve from there. Once you begin assigning metrics to individuals or departments, make sure the metric owners have continuous visibility to data so they can adjust levers and\ behaviors to drive performance. Having a scorecard of performance as a visible reminder of pending action is a significant motivator. Visible success is an even stronger motivator. Once an individual, team or the whole organization can visibly see improvement in metrics towards the desired target, they will be motivated to sustain the momentum.


Be careful what you measure, because you may get more of it than you want. While this is an often quoted cliché, it is true without exception. To make a metrics program successful, it must avoid the common pitfalls and: a) measure the right performance given the company strategy, b) allow for the right ownership, and c) be systematic and consistent. So while metrics are absolutely critical, relying solely on them to understand what's happening in a business is a surefire path to failure. A little qualitative analysis and anecdotes can be quite telling and should complement any metrics program. Most importantly, get engaged; you'll see if a process is achieving the desired outcome, and if not, why.

About the authors
Sharad Elhence

Sharad Elhence
Sharad Elhence is a Partner in the Strategy practice of Infosys Limited and works with F500 clients in the Manufacturing, Retail, and Distribution industries.

Saurabh Agrawal

Saurabh Agrawal
Saurabh is a Senior Principal in the Strategy practice of Infosys Limited.