This article was first published in 2018 on Jeudegolf.org. It was extensively updated three years later for MyGolfMedia and is being republished in 2026, one year after the website was hacked. In 2018, a scientific study compared the Zepp Golf 2 with a TrackMan 4. Looking back, the story of this small sensor raises a question that remains just as relevant today: when does data become reliable and relevant enough to actually help a golfer improve?
In 2015, I tested the Zepp Golf with considerable interest. It was a small sensor attached to the golf glove and connected to a mobile app. At a time when access to swing data was still largely restricted to professionals using expensive launch monitors, the promise was compelling: for around €150, an amateur golfer could start measuring his or her own swing.
My first review was positive. Three years later, after using both the Zepp and Zepp Golf 2, hitting thousands of balls with them and, most importantly, buying a TrackMan 4, my opinion had changed dramatically.
I subsequently published a second article explaining why I could no longer recommend the product. With even more hindsight, however, the story has become more interesting than Zepp Golf itself. It also illustrates the limitations of reviewing golf technology after using it for only a short period of time.
Why I believed in Zepp Golf
The concept behind Zepp was clever. The small inertial sensor attached to the lead-hand glove and recorded the golfer’s movements during the swing.
The app provided several metrics, including swing speed, club plane, hand plane, tempo, backswing angle and hand speed. Hip rotation could also be estimated by carrying the smartphone in a pocket.
In 2015, this was fairly remarkable.
Data was beginning to play a growing role in golf instruction and club fitting, but the average golfer obviously didn’t have a TrackMan sitting in the garage. I had already been using TrackMan regularly since 2012, without ever imagining that I would eventually buy one.
Zepp therefore appeared to offer a bridge between feel and measurement. Instead of simply thinking that my swing was too fast or too long, I suddenly had numbers and a three-dimensional representation of the movement. I could hit a ball, look at my phone and receive immediate feedback.
That was what convinced me.
But a device capable of producing a lot of data is not necessarily capable of producing the right data with enough reliability to guide practice.
Three years of use changed my opinion
When I returned to the subject in 2018, I had something that was inevitably missing from my first review: time.
I had used Zepp extensively and had also tested the second-generation product. More importantly, I had eventually bought a TrackMan 4. That experience gradually changed the way I looked at data.
With Zepp, for example, I had spent a considerable amount of time looking at my hand plane. The app provided reference values, and I tried to change my movement to get closer to them. In hindsight, I should have started with a much simpler question: what ball-flight problem was I actually trying to solve?
The same applied to club plane or hand speed. A metric can be interesting to observe and may change through practice without necessarily being something a golfer should try to alter directly.
TrackMan forced me to think differently. Club path, face angle, dynamic loft, quality of strike and ball-flight characteristics created a much more direct relationship between an action and its outcome. I could identify a problem, make a change to the swing and immediately see whether the resulting ball flight was moving in the desired direction.
The difference was therefore not simply about having more data. It was about having more relevant data.
When data can send your practice in the wrong direction
Backswing angle is probably the best example of something I misinterpreted at the time.
I was trying to increase my driver speed. Looking at the Zepp data, I started trying to make a longer backswing. That could indeed produce more clubhead speed, but I had turned a possible correlation into a technical objective.
With more experience, and after more discussions with golf coaches, I began focusing on other areas: coordination between the upper and lower body, sequencing, quality of strike and, ultimately, ball speed.
A longer backswing is not automatically a better backswing. For some golfers, artificially trying to increase its length can make coordination more difficult or compromise strike quality.
The problem wasn’t that Zepp displayed backswing angle. The problem was that I was turning that information into an objective without first establishing whether it addressed the problem I was actually trying to solve.
Were my doubts about speed justified?
Clubhead speed was where my reservations became strongest.
Because Zepp was attached to the glove, I questioned how a sensor located on the hand could determine, with sufficient precision, the speed of a point at the opposite end of the golf club.
In my 2018 article, I went too far by simply describing the measurements as “incorrect.” A study published that same year allows that criticism to be expressed much more precisely today.
Researchers from Chemnitz University of Technology and the University of Gothenburg compared the Zepp Golf 2 with a TrackMan 4. The study involved 20 golfers with handicaps ranging from 0 to 54 and analysed 300 swings with different types of clubs.
Two limitations need to be made clear from the outset. This was a short paper published in the proceedings of the International Sports Engineering Association conference, and the study examined clubhead speed only. It therefore tells us nothing about Zepp Golf 2’s hip rotation, hand plane, tempo, backswing angle or 3D swing representation.
For clubhead speed, however, the findings are particularly interesting.
Reasonably accurate on average, but not precise enough shot by shot
To understand the results, we need to distinguish between two concepts: accuracy, or how close the average measurement is to the reference value, and precision, or how consistently the device measures from one shot to another.
In terms of average accuracy, the Zepp Golf 2 performed reasonably well.
Compared with TrackMan 4, the average difference was +4.6 km/h with a wedge, +0.9 km/h with a 7-iron and −2.7 km/h with woods.
Taken on their own, those numbers sound reassuring. But the 95% limits of agreement tell a very different story. For an individual 7-iron shot, the difference from TrackMan could range from approximately −12.9 to +14.7 km/h. With a wedge, the range was approximately −9.7 to +18.9 km/h, and with woods it was −18.7 to +13.3 km/h.
The researchers estimated the random error at around 12% and referred to potential measurement errors of up to 27 km/h.
For a golfer trying to improve, this is not merely a statistical detail. It is the heart of the problem.
Imagine a player spending several weeks or months trying to gain a few kilometres per hour with a 7-iron. If the difference between the Zepp reading and the reference measurement can exceed 10 km/h in either direction on an individual swing, the measurement noise is greater than the improvement the golfer is trying to detect.
A swing reported as faster by Zepp therefore doesn’t necessarily tell you whether the golfer has genuinely gained speed or whether some of that difference simply comes from measurement variability.
That is a much more accurate criticism than the one I made in 2018. Zepp did not necessarily produce systematically incorrect speed readings. Across a series of swings, its average could be reasonably close to the reference, while individual measurements lacked the precision required to interpret some swings with confidence.
What the study does not tell us about Zepp
It would be tempting to use this study as retrospective confirmation of every criticism I made of the product. But the researchers only investigated clubhead speed.
My assessment of tempo, hand plane, club plane, hip rotation and the 3D swing representation therefore remains based on my own experience as a golfer and equipment tester, rather than on scientific validation.
The 3D representation is a good example. In 2018, I described it as “perfectly accurate.” In reality, I had no independent measurement system capable of verifying that claim.
What I can say today is that the representation appeared consistent with my swing and was one of the product’s most convincing features.
That distinction says quite a lot about how my approach to equipment testing has evolved: separating what I observe, what I feel, what I can actually measure and what I can demonstrate.
Not everything about Zepp was useless
Changing your mind about a product doesn’t mean rewriting history and pretending everything about it was bad.
Zepp made me aware, for example, of how fast my tempo was. I cannot retrospectively guarantee the precision of that measurement because the 2018 study did not evaluate tempo, but it encouraged me to work on an aspect of my swing to which I had previously paid too little attention.
Its other benefit was even more indirect: Zepp made me want to practise.
I hit hundreds and eventually thousands of balls with the sensor. I looked at the data, experimented and tried to understand my swing. I cannot claim that I improved because of Zepp’s algorithms. But I certainly practised more because the device made training more engaging and measurable.
What Zepp Golf can still teach us in 2026
Zepp Golf now belongs to an earlier generation of connected golf technology. But its story remains an excellent case study.
In 2015, I had a technically appealing product in front of me. It displayed speeds, angles and graphs. The apparent precision of those numbers naturally increased my confidence in them.
Yet producing a number and producing a measurement reliable enough to guide practice are two different things. There is another hurdle after that: even if a variable is measured accurately, it still has to be relevant to the problem the golfer is trying to solve.
That was probably my biggest mistake during my first Zepp review. At times, I confused measuring more with understanding more.
Five questions to ask about any golf measurement device
Before using a launch monitor, sensor, watch or app to make changes to your swing, I now believe five questions are essential:
- What does the device actually measure? Is the variable measured directly, or calculated from other information using an algorithm?
- Is the measurement accurate? Across a sufficiently large sample, how close is it to an established reference?
- Is it precise? Is measurement variability low enough to distinguish genuine improvement from measurement noise?
- Is the data relevant to the golfer’s problem? Just because something can be measured doesn’t mean it needs to be changed.
- Does changing the metric actually improve the outcome? In golf, the final judge remains the ball: strike quality, speed, direction, trajectory and dispersion.
A system can therefore measure a variable correctly without that variable necessarily being a useful training metric.
Testing also means knowing when to change your mind
When I wrote my second article in 2018, I opened with a sentence that I find even more important today than I did at the time:
“I owe my readers an apology.”
I had recommended a product three years earlier, and I felt that the experience I had accumulated since then required me to reconsider that recommendation.
I don’t disown the original review. It reflected what I had genuinely observed and believed in 2015. What it lacked was precisely what a test lasting a few hours or a few days can rarely provide: enough time to discover whether the data generated by a device genuinely changes the golfer’s understanding of the swing and the way he or she practises.
Another eight years of hindsight now lead me to qualify some of the overly categorical statements I made in that second article as well.
That, too, is part of testing golf equipment.
In the case of Zepp, it took thousands of golf balls, two generations of the product, experience with other measurement systems and comparison with an independent study for me to better understand what I had in front of me back in 2015.
Zepp Golf had a very good idea: making swing measurement accessible to ordinary golfers. What its story ultimately taught me is that there can be a considerable gap between producing data and producing data that is reliable, relevant and actionable enough to help a golfer improve.
That lesson remains every bit as relevant in 2026.
Scientific source: Paul Lückemann et al., Validation of an Inertial Sensor System for Swing Analysis in Golf, Proceedings of the 12th Conference of the International Sports Engineering Association, 2018. Read the study on MDPI