Haptic feel is subjective, but most of what makes a click feel crisp or sloppy can be measured. The usual method mounts an actuator or a finished device on a known mass, records its motion with an accelerometer, and reads strength, timing, and decay off the trace. No single number replaces trying the device on real hands, and each measurement below has limits worth knowing before you compare parts or products.
The test setup
Actuator datasheets usually report performance on a standard test mass, commonly around 100 grams, suspended on soft foam or elastic so the fixture can move freely without the bench damping it. An accelerometer fixed to the mass records the motion while the actuator plays a defined input signal. Because the mass is standardized, figures from different suppliers can be compared roughly, but fixtures, mounting direction, and drive conditions vary, so check those details before putting two datasheets side by side.
For a finished product, the meaningful test is on the product itself. Hold or mount it the way a user would, place the accelerometer where the skin makes contact, such as the back of a watch case or the face of a touchscreen, and measure there. A heavier device needs more force for the same acceleration, and the housing can boost some frequencies and absorb others.
Acceleration
Peak or RMS acceleration, quoted in g, is the headline strength figure. More acceleration generally feels stronger, but perception isn't linear and it depends on frequency. The Pacinian receptors that dominate vibration sensing detect the smallest displacements at around 200 to 300 Hz, so the same acceleration can feel noticeably different at two different frequencies.
Note which axis is being measured. An ERM shakes in the plane of its rotation and an LRA moves along a single axis, so a single-axis reading taken in the wrong direction understates the output. Many test setups record all three axes and report the combined magnitude alongside the main axis.
Rise time, stop time, and ringing
Timing often shapes perceived quality more than strength does. Rise time describes how quickly the actuator reaches its target output after the drive starts, though definitions vary, and some suppliers quote the time to half of rated acceleration while others use 90 percent. Stop time, sometimes called fall or brake time, is how long the motion takes to decay below a set threshold after the drive ends, and ringing is the residual oscillation in that tail.
Short rise and stop times are what make a click feel like a click. An ERM may take tens of milliseconds to reach full speed, a well-driven LRA can deliver most of a click's energy within a few cycles, and a piezo actuator can be faster still. Look at the whole envelope on the trace, since a pulse with a fast rise and a long tail still feels soft, and a decaying wobble after the main pulse can register as a second, fainter tap.
Frequency response
Sweeping the drive frequency at a fixed input level and plotting output acceleration shows how much room an actuator gives the effect designer. A typical LRA produces a sharp peak at resonance and falls away on both sides, while wideband actuators and piezo parts produce a flatter curve. Running the same sweep on the finished product also exposes enclosure resonances, the frequencies where a loose panel or battery starts to rattle.
Acoustic noise
Haptics that buzz audibly undercut the point of silent feedback. Measure sound pressure with a microphone at a fixed distance in a quiet room, and test the device in realistic positions, including lying on a hard table, where a phone or watch can rattle against the surface. Square-wave drive, loose internal parts, and enclosure resonances are the usual suspects when a device is louder than expected.
Latency
End-to-end latency is the time from a user's action, such as a finger landing on a virtual button, to the moment the haptic output can be felt. It adds up touch sensing, operating system and application processing, driver startup, and the actuator's own rise time. Measuring it means capturing both events on one timeline, for example with a rig that logs the touch sensor and the accelerometer together or with a high-speed camera filming the finger and the device.
A noticeable gap between touch and response weakens the sense that a button was pressed. Acceptable latency depends on the task and on whatever visual or audio feedback arrives alongside the haptic cue, so measure with users in the real context and treat any single target figure with caution. The guide to haptics in car touchscreens covers a setting where delay is especially hard to hide.
What the numbers miss
Lab measurements don't capture how the output couples into a particular person. Strap tension, skin contact, body location, and age all change what reaches the receptors, and vibrotactile sensitivity generally declines with age. Real environments add masking too, since walking, driving, or a crowded bus can swamp a cue that's obvious on the bench.
Perceptual testing fills the gap. Paired comparisons, rating scales, and identification tests, where people try to name which of several patterns they felt, show whether measured differences are noticeable and whether patterns stay distinct under distraction. The article on designing a tactile language covers pattern identification in more depth, and the actuator comparison and waveforms and drivers guide explain where the measured differences come from.