Scoring as a Service

Why padel’s score problem is not an AI problem — and what any solution has to pass

In short: A padel scoring system has to pass five requirements at once — the score must be visible to all four players continuously, audible on every change, updated within a few hundred milliseconds, entered without anyone walking anywhere, and correctable by any player from where they stand. These are gates, not points: four out of five is not eighty per cent of a scoring system, it is zero. AI cameras are strong at highlights, analytics and streaming, but automatic scoring needs 100% accuracy rather than the ~95% computer vision currently achieves, and a camera does not remove the need for a scoreboard — it adds one, along with an outdoor high-brightness display and its cabling on every court. Watch and phone apps fail on visibility: the score sits on one wrist and the other three players have to ask.

Padel has solved most of its software problems. Booking is mature. Tournament management is mature. Ranking algorithms exist and are getting better. What is still unsolved — and what everyone in the industry quietly agrees is unsolved — is the score itself.

Not the score after the match. The score during the match: the number four players are actually playing to, right now, at 4-4 in the third with everyone tired and nobody quite sure whether that last game went to the reds or the blues.

Two approaches are competing to own that moment. Manual scoring, where a human tells the system what happened. And AI camera scoring, where a computer vision system watches the court and decides what happened. The debate usually gets framed as “accuracy” — can the AI see the ball or not. That framing is too narrow, and it lets both sides avoid the harder question.

The harder question is: what does a scoring system have to do, at the level of the court, to be usable at all?

What does “AI camera for padel” actually mean?

“AI camera for padel” currently describes four different products, and conflating them makes this discussion impossible.

Highlights and replay — the camera clips the tweener, the por tres, the winner off the back glass, and pushes it to the player’s phone. The club gets free social content out of it.
Performance analytics — shot counts, rally length, court coverage, heatmaps, time in the transit zone.
Live streaming — broadcasting the court to YouTube or a club screen.
Automatic scoring — the camera decides who won the point and maintains the state of the match.

Only the fourth competes with a scoreboard. The first three are good products with real demand, and nothing in this article argues against them. A club that buys highlight cameras and gets members sharing clips has made a sensible purchase. That is a marketing and engagement product, and it is orthogonal to scoring.

It is worth being precise about how much of the market is actually in category four, because it is much less than the marketing suggests. Search the padel space for automatic scoring today and what comes back is overwhelmingly manual: tap-to-score apps for Apple Watch and iPhone, one-tap zones, undo buttons, haptic confirmation, server and side tracking. The computer vision work that does exist on the scoring problem tends to appear as an OEM subsystem — one consultancy describes building event detection for serve, rally, out and end-of-point as modules delivered into a client’s platform, not as a product a club can buy. The clearest “AI smart referee for automatic scoring” claim in the racket-sports space right now sits on a tennis-oriented ball machine module, not on a padel court camera.

So the comparison in this article is narrow and deliberate: camera-inferred scoring versus player-entered scoring. Everything else the cameras do is a separate purchase with a separate justification.

What does a padel scoring system have to do?

Before comparing technologies, it is worth writing down what a scoring system must satisfy on court. These are not nice-to-haves. Fail any one of them and players stop using the system, no matter how clever the technology behind it is.

1. The score must be visible to all four players, continuously

Not on a phone in someone’s bag. Not on a tablet at the entrance. Not in an app that has to be opened. Visible from anywhere on court, at any moment, without anyone having to ask for it or move to get it.

This matters because the primary function of a scoreboard in amateur padel is not record-keeping — it is dispute prevention. Two players remembering the score differently is the single most common friction point in a social match. A score that has to be requested does not prevent that dispute; it only resolves it after it has already started.

There is exactly one alternative to a visible display: announce the score aloud after every change. It works, and it is genuinely awful. A synthetic voice reading out “cuarenta, treinta” ninety times in two hours is intrusive on the court it serves and worse on the three courts either side of it, where players are hearing a score that is not theirs. Any club with more than two courts rules this out immediately. Announcement is a supplement to a display, not a substitute for one.

2. Every score change must be announced with sound

A visual-only update is a passive update. Someone has to happen to be looking at the right moment. Sound makes the change an event that all four players register, whether they are looking or not.

The dispute-prevention benefit is obvious. The less obvious benefit is integrity. If the score can change silently, it can be changed silently — a remote pressed twice, a point quietly added during a changeover, an “adjustment” nobody witnessed. A buzzer does not make cheating impossible. It makes it impossible to do unobserved, which is nearly the same thing in a sport played by four people standing twenty metres apart.

This applies with equal force to automated systems. An AI that silently corrects a point from three rallies ago has done something players cannot audit.

3. The update must be effectively instantaneous

Padel is fast. The gap between the end of one point and the serve of the next is often under ten seconds, sometimes under five. A scoring system that takes two or three seconds to reflect a point has already lost the moment. Worse, it creates a specific failure pattern: the player presses, nothing visible happens, so they press again. Now the score is wrong, and the system that was supposed to prevent disputes has caused one.

The tolerable budget here is a few hundred milliseconds — fast enough that the update feels like a consequence of the button press rather than a separate event.

4. The input must come to the player, not the player to the input

This is the requirement most often violated, and the one that quietly kills adoption.

Any system that requires a player to walk to the glass, the net post, or a wall-mounted panel to register a point imposes a movement cost on every single point of the match. Across a two-hour session that is a hundred or more unnecessary walks. Players will comply for the first two games and abandon it by the fourth. The behaviour is entirely predictable and it is not laziness — it is a rational response to a system that asks them to interrupt their own game to feed it data.

Score input has to live in a pocket, on a wrist, or on the fence within arm’s reach of where the point actually ended.

5. The players must be able to verify — and correct — every point

Whatever generates the score, the four people on court have to be the final authority on it. That means two capabilities, not one: they see and hear each change as it happens, and they can fix a wrong one immediately, from wherever they are standing.

This is not a concession to distrust of AI. It is the principle that governs every officiated sport — someone with standing has to be able to say no, that one was ours, and be heard. Remove it and you have a system that is right most of the time and unchallengeable all of the time, which is a worse experience than a system that is wrong occasionally and correctable in one press.

A verified score also has a property an inferred score does not: it is defensible. When a ranking point is contested three weeks later, “all four players saw it, heard it, and did not object” is evidence. “The model was confident” is not.

How well does manual scoring work in padel?

Traditional manual scoring — no system at all — trivially satisfies requirements 3 and 4 and fails 1 and 2 completely. It is instant and frictionless because it happens in human memory, and it is unreliable for exactly the same reason. No record, no stream overlay, no ranking data, and a dispute roughly every third match.

Watch and phone scoring apps are the fastest-growing category and the clearest illustration of why partial compliance fails. They are genuinely good at requirements 3, 4 and 5: the tap is instant, the device is already on the player’s wrist so nobody walks anywhere, and undo is one press. Some add haptic confirmation, which is a private version of requirement 2.

And they fail requirement 1 completely, which is enough to sink them.

The score lives on one wrist. The other three players cannot see it. To find out the score they have to ask the person wearing the watch — which is precisely the interaction the system was bought to eliminate. Worse, it quietly appoints that player as the umpire. One person now owns the record, decides when it was a mistap, and is the only one who can audit it. In a friendly that is mildly awkward; in a league match with ranking points attached it is untenable.

The obvious fix is to add a display, and this is where the approach collapses under its own logistics rather than its technology:

Where does it go? A phone or tablet has to be readable by four players spread across twenty metres of court. Mount it on the net post and it is unreadable from the back glass. Mount it on one end fence and it faces away from half the court. The screen size that solves this is not a tablet.
It cannot be an LCD tablet outdoors. A consumer tablet runs a few hundred nits against roughly 10,000 nits of ambient sunlight, and in direct sun on a Spanish afternoon it does not merely wash out — it thermally shuts down. This is not a theoretical failure.
Who charges it, and when? Eight courts means eight devices that staff have to collect, charge and redeploy every day, and one forgotten charger takes a court out of service.
Whose device is it, anyway? If it is a player’s phone, the score walks off the court when that player leaves, and the next four arrive with nothing. If it is the club’s, it needs mounting, weatherproofing, theft protection and a support process.

Solve all of that properly and you have not built a phone app. You have built a scoreboard — just an expensive, fragile, badly-shaped one.

Tap-a-tablet or tap-the-glass systems typically satisfy requirement 1, sometimes 2, and fail requirement 4 badly. They are the most common failure mode in club deployments because they look excellent in a product demo — someone taps a nice interface, the score updates — and the demo never simulates the ninetieth walk to the glass. These systems are usually abandoned in place: installed, powered on, ignored.

RF remote or wearable input into a dedicated display is the only configuration that satisfies all five. The player who won the point presses a button they are already holding, the display updates in the time it takes the radio packet to arrive, a buzzer sounds, all four players see it. There is no round trip to anywhere. The failure modes are hardware failure modes — battery, radio range, interference — which are engineering problems with known solutions, not conceptual limitations.

The genuine weakness of manual scoring is that it is only as honest as the person pressing the button, and it captures nothing beyond the score. No rally length, no shot data, no video.

Can an AI camera score a padel match automatically?

AI camera systems are doing real work in padel and it would be dishonest to dismiss them. Ball tracking, in-or-out calls, movement analysis — these are things no button can deliver, and the category is growing fast, with systems deployed across hundreds of clubs in Europe. But most of that deployment is categories 1 to 3. What follows applies only to the claim that a camera can own the score.

Against the five requirements, that claim runs into trouble in several places at once.

On requirement 3 (latency), the objection is weaker than it used to be. The early generation of AI scoring was cloud-based, and a cloud round trip is genuinely fatal on court. The current generation increasingly runs inference on the camera itself. One vendor advertises a 12 TOPS on-device processor running 4K at 90 FPS and explicitly markets “no cloud round-trip.” A computer vision consultancy publishing padel case-study figures reports game-event recognition under 200 ms with all analytics running client-side rather than in the cloud. If those numbers hold in the field, edge-inference AI can meet the latency requirement.

So the latency argument should be aimed precisely: it kills cloud-dependent scoring, not on-device scoring. Anyone making the broad version of the claim will be corrected by a vendor who has done the engineering.

On requirement 1 (visibility), AI systems have no answer of their own. A camera knows the score; it does not show it to the players. This is why it is telling that AI camera vendors are now shipping LED scoreboards alongside their cameras — one of the leading padel AI camera products lists a bright LED scoreboard controlled by wireless buttons or an Apple Watch as part of the offer. The camera company has independently arrived at the conclusion that you still need a display and you still need a button.

On requirement 2 (audible confirmation), the same gap applies, with an added problem: automated corrections. When a vision system revises its own earlier judgement, the players have no way to know it happened or to contest it.

On accuracy, 95% is not a passing grade. For a scoreboard it is a failing one. Published padel computer vision figures land around 95% for ball and player tracking and above 92% for bounce and out detection. Those are impressive numbers for a genuinely hard problem — tracking a ball through glass reflections, variable outdoor light and four bodies in four metres of space is hard-won engineering, and nobody should be sneering at it.

But the required accuracy for a scoring system is not “high.” It is 100%, and there are two structural reasons why.

Every point counts, and errors do not cancel out. In analytics, a 5% error rate is noise — the heatmap is still useful, the average rally length is still roughly right, and mistakes in opposite directions partially offset. A score has none of that forgiveness. A padel match runs 60 to 90 points. At 95% per-event accuracy that is three to five wrong events per match, and it only takes one. A missed point at 3-3 does not make the system 95% right about the match; it makes the system wrong about every subsequent state of the match until a human intervenes. The error propagates instead of averaging.

There is no natural correction point. An analytics error surfaces later, in review, where it can be fixed calmly by someone looking at footage. A scoring error surfaces instantly, in front of four people, mid-match, and has to be resolved in the ten seconds before the next serve. Which means the system needs a correction path that is immediate and available on court — a button in a player’s hand — and once you have accepted that, you have accepted that the AI is a proposal engine and the players are the authority.

Nobody in the industry is claiming 100%, and the honest published figures are the ones quoted above. That is not an indictment of the technology. It is a statement about what job those numbers qualify it for: analytics, video, highlights, line-call assistance. It does not qualify a camera to be the unsupervised author of a result that feeds a ranking.

And there is the coverage problem. The overwhelming majority of padel courts in Europe will never have an AI camera installed. Not because clubs are unconvinced, but because the capital cost and installation requirements only make sense on a subset of courts at a subset of venues. A scoring layer that only works on camera-equipped courts is not a scoring layer for the sport.

Do AI cameras remove the need for a scoreboard?

This is the part that rarely makes it into the sales conversation, and it is the part that decides whether a club can actually afford the system.

If requirement 1 holds — and it does — then an AI camera does not remove the need for a scoreboard. It creates one. The camera knows the score. The four players do not. Something on that court has to tell them, continuously, and that something has to survive a padel court.

A normal screen will not do it. Direct sunlight is on the order of 10,000 nits of ambient light. A consumer TV or monitor runs 250–400 nits and is simply invisible outdoors. Industry guidance puts 2,500 nits at readable in shade but struggling in sun, 3,500 at good in most conditions, and 5,000+ at reliably readable in anything. An outdoor court on the Costa del Sol needs the top of that range. High-brightness panels of that class are a different product category from a television, with a price to match.

It has to survive the weather. IP65 as a floor, sealed enclosure, anti-reflective toughened front glass, active thermal management for a panel whose surface hits well over 100 °C in direct sun, and an operating range that covers both a Spanish August and a German January. Rain, condensation, dust off a synthetic court and salt air near the coast all get a vote. Every one of those requirements adds cost before a single pixel is lit.

It has to be cabled. Mains power at a position where all four players can see it, plus a data path from the camera or a local controller. At an existing club that means trenching, conduit or surface trunking across a facility that was never designed for it. This is frequently the largest single line item and the one that converts a per-court upgrade into a construction project requiring an electrician, a permit conversation and courts out of service.

And all of it is per court, not per club. Multiply the display, the enclosure, the mounting and the cabling by every court the club wants covered.

The honest summary is that “AI camera scoring” is never just a camera. It is a camera, plus an outdoor-rated high-brightness display on every court, plus the civil and electrical work to get power and data to each one — and the display side of that invoice is often larger than the camera side.

Contrast the alternative. A self-contained scoreboard stands under the net — the one position on a padel court that is equidistant from all four players and blocked from none of them — and displays the score on both faces at once.

Four things follow from that, and each one deletes a cost line from the section above:

LED dot matrix instead of LCD. This is the correct technology for the job. LEDs emit rather than fight a backlight against the sun, so sunlight readability comes almost for free, and a score needs a handful of characters rather than four million pixels. The brightness problem that makes an outdoor LCD panel expensive simply does not arise.
A lux sensor on each face, set independently. This matters more than it sounds. On an outdoor court one side is usually in sun while the other is in shade, and the sun moves through the afternoon. A single global brightness setting is wrong for one side of the net by definition — too dim to read on the bright side, or glaring and uncomfortable on the dark side. Sensing and driving each face separately is the only way to be readable from both ends of the court at the same time, and it is a thing a wall-mounted single-faced display physically cannot do.
No mains at the display position. Battery or passive PoE over a single Cat5 run. No trenching, no electrician, no permit conversation, no courts out of service.
No cabling to the point of use at all. Input arrives over 433 MHz from a remote in a player’s pocket.

And because none of it is bolted to the structure, it can be moved to another court when the club reconfigures, or put in the boot of a car and driven to a tournament.

The camera is a fixed installation. The scoreboard does not have to be, and the moment you accept that a display is mandatory either way, the capex comparison stops being close.

Why does padel scoring have to cover every court?

The scope correction above has a direct commercial consequence, and it is the one that decides whether a club can deploy scoring at all.

Highlights, analytics and streaming are amenities, and amenities work on one court. That is not a criticism — it is the business model. The club equips its show court, gets a marketing asset, sells a subscription to the members who care, and the scarcity is part of the appeal. Deploying on one or two courts is the correct decision for that product.

Scoring is not an amenity. It is infrastructure. And infrastructure that exists on one court out of eight does not exist.

Consider what scoring as a service actually has to do. A club runs an internal league across all its courts. A tournament schedules fifty-five matches across six courts over a weekend. A ranking system ingests results and expects them to be comparable. In every one of those cases, an uncovered court is a hole in the data. You cannot tell the four players on court 6 that their result does not count towards the league because they were not on the camera court. You cannot run a tournament where half the matches produce a live feed and half produce a paper sheet handed in at the desk. The moment coverage is partial, the club reverts to the paper sheet for everything — a process that only works sometimes is operationally worse than a process that always works badly.

This is why the two deployment shapes are incompatible. The economics of AI cameras push towards one or two premium courts. The requirements of scoring push towards all of them. And a club that is delighted with its highlight cameras on courts 1 and 2 has not moved a millimetre closer to solving scoring — not on the other six courts, and not on courts 1 and 2 either, because the camera it bought was never sold as a scoring system in the first place.

So the honest question to put to any vendor proposing camera-based scoring is not “what does a camera cost.” It is:

What does it cost to run automatic scoring on every one of our courts, for five years — including a sunlight-readable display and its cabling on each one — and what accuracy will you contractually stand behind?

That question stacks all three problems at once: recurring per-court cost, the display bill from the previous section, and the 100% requirement. It is uncomfortable not because AI cameras are bad products, but because scoring is a different job from the one they were built for, priced on a model that assumes selective deployment, and measured against a tolerance that analytics never has to meet.

A scoreboard has the opposite shape. One-off per-court purchase, modest recurring software cost, goes on the court nobody would ever film, unhooks and moves when the club reconfigures, and cheap enough at eight units that “every court” is the default rather than the stretch goal.

Manual scoring vs AI camera scoring: side by side

Requirement	No system	Watch / phone app	Tap-the-glass	RF remote + display	AI camera (cloud)	AI camera (on-device)
1. Visible to all four players, always	✗	✗	✓	✓	✗	✗ (unless paired with a display)
2. Audible on every change	✗	✗ (haptic, private)	Partial	✓	✗	✗
3. Sub-second update	✓	✓	✓	✓	✗	✓
4. No walking to input	✓	✓	✗	✓	✓	✓
5. Players correct in one press, in place	✓	Partial (one player only)	✗	✓	✗	✗
All five met	✗	✗	✗	✓	✗	✗
100% score accuracy by construction	✗	✓	✓	✓	✗	✗
Reliable digital record	✗	✓	✓	✓	Partial	Partial
No per-court cabling or civil works	✓	✓	✗	✓	✗	✗
No daily charging / staff handling	✓	✗	✓	✓	✓	✓
Affordable on every court, not just one	✓	✓	Partial	✓	✗	✗
Works on any court, movable	✓	✓	✗	✓	✗	✗
Analytics, video, highlights	✗	✗	✗	✗	✓	✓

Why do padel scoring systems fail on court? The requirements are a chain, not a scorecard

Read the table again and notice the shape of the failures. Almost every approach scores well on most rows. The watch app gets three of the five. The tablet on the glass gets two or three. The AI camera gets one or two on its own and more once you bolt a display to it. Nothing on the market is bad at everything.

That is exactly the trap. These are not points to be accumulated. They are gates, and the system is the product of them, not the sum.

Four out of five is not eighty per cent of a scoring system. It is zero. A system that is fast, effortless, accurate and correctable but that only one player can see does not produce a shared score — it produces one player’s private opinion of the score, and the other three are back to arguing. A system that is visible, audible, instant and correctable but wrong once a match produces a match record nobody trusts. A system that satisfies every requirement on court 1 and none on courts 2 through 8 produces a club that goes back to paper.

This is why so many of these products demo brilliantly and die in the field. A demo is a single-requirement test: someone taps, the number changes, everyone nods. The court is a five-requirement test run ninety times in two hours, in the sun, with four tired people who have no patience for a device that needs them to accommodate it.

It also explains why the surviving configurations look so unglamorous. A dual-sided LED display standing under the net, a buzzer, and a button in each player’s pocket is not an exciting product. It is simply the smallest arrangement of parts that passes all five gates simultaneously — visible to everyone at once, audible to everyone at once, instant because the radio link is direct, zero-displacement because the button is already in your hand, and correctable by any player without walking anywhere.

Every alternative is an attempt to remove one of those parts. Remove the display and only one player knows the score. Remove the button and the camera has to be right every single time. Remove the sound and changes go unwitnessed. Remove the per-court deployment and the league falls apart. Each removal looks like a simplification and is actually a gate failure.

What “Scoring as a Service” actually means

If scoring is infrastructure rather than an amenity, then the thing a club is buying is not a box. It is coverage — every court, every match, every day, with the result landing in the league table and the tournament bracket without anyone retyping it from a paper sheet.

That reframes what is being sold. A club does not want scoreboards; it wants to stop losing results, stop arbitrating disputes at the desk, and stop running its league on a spreadsheet. The hardware is how that gets delivered, not what is being bought — the same way a club buying a booking system is buying full courts, not a database.

Three things follow if you take the framing seriously:

Coverage has to be all-or-nothing. Partial deployment reverts the club to paper for everything, so the unit of sale is the club, not the court. Pricing that makes eight courts affordable matters more than pricing that makes one court cheap.

The obligation is continuous. A service means the scores keep arriving — replacements when a unit fails, firmware that keeps pace with rule changes such as the FIP formats, integrations that keep working when a tournament platform changes its API. A club that buys a box owns a maintenance problem. A club that buys a service does not.

The score has to leave the court. Visible and audible on court satisfies the players; the club needs the same score in its league table, its bracket, its ranking feed and its stream overlay, from one source of truth. That integration layer is the service, and it is the part a camera vendor and a scoreboard vendor are equally obliged to build.

One caution on the term, though. “As a service” sets an expectation that hardware is included in a subscription rather than bought up front, and that is a real commercial decision, not just a naming one. Bundling hardware into a monthly fee is the version that best matches the argument in this article — it removes the capex barrier that stops clubs from covering all eight courts, which is precisely the failure mode described above. It also means the vendor finances the hardware, carries the working capital, and absorbs the loss when a club churns after four months with eight units in the field. A split model — hardware purchased, platform and integrations on subscription — is honest, common, and still legitimately a service. What is not honest is using the language of a service while shipping a box and walking away.

What does a working padel scoring setup look like?

The framing of “manual versus AI” is a vendor framing, not a player framing. Players do not want a scoring philosophy. They want the score to be right, visible, and instant, and they want to not think about it.

The configuration that actually delivers that is one where the two layers do what each is good at:

AI proposes. The camera detects the end of a rally and infers who won it. It handles video, highlights, analytics, and line calls — everything a button cannot do.
Hardware confirms and displays. The score appears on a display both teams can see, with a sound on every change. A remote in the player’s hand corrects it in one press when the inference is wrong, without anyone walking anywhere.
The display is the source of truth. Not the camera, not the cloud. What the players can see and hear is what gets written to the match record, pushed to the tournament platform, and fed into the ranking algorithm.

That last point is the one worth insisting on. A score that four players have visually confirmed and audibly witnessed is a socially verified score. A score that only a server knows is an assertion. For rankings, for tournaments, for anything with stakes attached, the difference matters — and it will matter more as ranking systems start carrying real consequences.

How to evaluate a padel scoring system: eight questions

Ignore the demo. Ask eight questions:

Standing at the back of the court, mid-rally, can I see the score without turning around or asking anyone?
When the score changes, do all four of us know — including the two who were not looking?
From the moment the point ends to the moment the display changes, is it fast enough that I would never press twice?
How many steps does it cost me to register a point?
When it gets it wrong, how long does it take to fix, from where I am standing, and am I allowed to fix it?
What is the claimed per-point accuracy — and what happens to the rest of the match after the first error?
What is the total cost to put a sunlight-readable display on every court, including power, data and installation?
What is the five-year cost of running this on all our courts — and what happens to our league and tournaments on the courts it does not cover?

Any system that answers all eight well is a good scoring system, whatever technology is inside it. Most systems on the market today fail at least two. It is almost always visibility and input distance that break first on court — and questions 7 and 8 that break the deal in the club owner’s office.

Frequently asked questions

Can AI cameras score padel matches automatically today? Mostly not as a purchasable club product. The AI camera market in padel is dominated by highlights, performance analytics and live streaming. Automatic scoring exists largely as computer vision subsystems sold into other companies’ platforms, and the products clubs actually buy for scoring are still manual — remotes, watch apps and tap interfaces.

How accurate does automatic padel scoring need to be? 100%, or it needs an instant human correction path. Published padel computer vision figures sit around 95% for ball and player tracking. In analytics a 5% error rate averages out; in scoring it propagates, because one missed point makes every subsequent state of the match wrong until someone intervenes.

Why is a scoreboard needed if the camera already knows the score? Because the camera knows it and the players do not. All four players need the current score visible continuously, which means a display on the court — outdoor-rated, and bright enough to read against roughly 10,000 nits of direct sunlight, where a consumer screen manages 250–400 nits.

Why do Apple Watch and phone scoring apps not solve this? They solve input but not visibility. The score lives on one player’s wrist, so the other three have to ask — the exact dispute the system was meant to prevent — and that player becomes the de facto umpire. Adding a shared display brings mounting, charging, sunlight readability and ownership problems that end in building a scoreboard anyway.

What does “scoring as a service” mean for a padel club? Coverage rather than hardware: every court, every match, with results landing in the league table, the tournament bracket and the ranking feed without anyone retyping them from paper. Partial coverage does not degrade gracefully — as soon as some courts are uncovered, the club reverts to paper for all of them.