What is a baseline metric, and why your pilot needs one

On-Off Group

October 2026

A baseline metric is the number your customer journey produces today, measured before you change anything. Without it you cannot say whether a pilot worked, because you have nothing to compare the after number against.

If you are running a pilot on a customer journey this quarter, this is the bit that decides whether anyone believes you at the end of it.

A baseline metric is the number the journey produces today, measured before you change anything. Turnaround time on onboarding. Completion rate on the application form. Calls to the branch per hundred applications. One number, measured once, written down with the date.

Most pilots skip it. Then the boss asks whether it worked, and the honest answer is a shrug.

What it looks like when there is no before number

The project ends. The new flow is live. Everyone agrees it feels faster. Someone asks how much faster, and the room goes quiet.

What comes out instead is a report on activity. Number of workshops run. Number of staff trained. A satisfaction score from the session. All true, none of it an answer to the question that was asked.

This is how a six month project ends in a slide pack. Not because the work was bad, but because nobody wrote down the turnaround time in month one. By the time you need the comparison, the old flow has been switched off and the old data is gone or was never kept in a form you can use.

The cost is not only the report. Without a before number you cannot defend the budget for the next phase, and you cannot tell which of the four things you changed actually moved anything.

Where the before number comes from

There are three usual sources and you should try them in this order.

First, systems. Someone in operations already runs a monthly report with turnaround time in it. It may be defined oddly, but it exists and it goes back years.

Second, a manual count. Pull fifty applications from last month, open them, and record how many days each took from submission to decision. Slow, boring, completely defensible.

Third, watching real people. Five customers doing the task on video, timed, with a note of who finished and who gave up. That gives you a completion rate and a duration you can repeat later. There is a separate piece on onoffgroup.com on how to come out of a customer journey session with a drop-off number instead of a wall of sticky notes.

Five users is enough to start. It is not a survey. It is a measurement you can run again in the same way.

Pick the number your boss already uses

The temptation is to invent a better metric. Resist it. If the bank runs on turnaround time, measure turnaround time, even if you think completion rate matters more.

A number nobody recognises has to be explained before anyone believes it, and that is where the meeting goes off track.

Write the definition down in one sentence before you measure anything. "Working days from customer submitting the form to decision sent, for retail accounts only, excluding cases returned for missing documents." That last clause is the one people will argue about in six months, so decide it now while nobody has a result to defend.

One number is fine. Two is manageable. A whole dashboard usually means nobody wanted to commit to one number.

Measure the after exactly the same way

The after number is only worth something if you collected it the same way as the before. Same definition, same exclusions, same source.

This sounds obvious and it is where most comparisons quietly fall apart. The before came from a manual count of fifty cases. The after came from a system report that counts calendar days instead of working days. Now some of the improvement is just the change in counting, and finance will spot it.

The pairs that hold up are simple and dull. With a large insurer, five-day design sprints took journeys that used to need six months or more and got them designed, tested with customers and live in four weeks. Completion rates on the new journeys rose 80 per cent and the old drop-off points were gone. The client measured both ends. The design sprint with Meralco was measured the same way, before and after: onoffgroup.com has the write-up.

"26 weeks down to 4" is easy for your boss to repeat in the next meeting. It only exists because someone knew what 26 was.

Small honest numbers beat big vague ones

Sometimes the baseline shows the problem is smaller than everyone assumed. Turnaround time is four days, not the two weeks people complain about, and the real pain is the three phone calls in between.

Say so. A pilot that moves a small number by a real amount is easier to fund again than a big claim with nothing under it. And if the before number shows there is no real problem, you have saved months of work on the wrong thing.

What to do this week

Pick one journey and write down, in one sentence, the number you will be judged on.
Ask operations whether that number already sits in a monthly report, and get the last six months of it.
If it does not exist, count fifty recent cases by hand, or time five real users doing the task.
Write the definition, the source and today's date on the same page as the number, and send it to your boss before the pilot starts.
Book the same measurement in the diary for the week after the pilot ends, using the same source.

Questions people ask

How many baseline metrics should a pilot have?
One or two. Pick the number your boss already asks about, usually turnaround time or completion rate, and measure that. A dashboard of fifteen numbers gets argued over instead of used.

What if the before number is embarrassing?
That is the useful case. A bad baseline gives you the biggest visible improvement later, and it is far safer to show it now than to have it found after the pilot.

Can you get a baseline if the system does not log anything?
Yes. Sit with five real customers or branch staff, time the task with a stopwatch, and count how many finish. Five timed runs is a defensible starting number.

How long does it take to get a baseline?
Usually about two weeks, and often less if the data already sits in a report someone runs monthly. Most of the time goes on agreeing the definition, not on the measuring.

On-Off Group trains teams, tests products with real customers, finds where a transformation has stalled and builds what gets it moving, for banks, insurers and enterprises in the Philippines, since 2015. How we help with baselines and measurement.