One Serve, Two Questions: Did It Land, and Did It Come Back?

First-serve percentage tells you how often a player lands the first serve. First-serve unreturned rate tells you how often those made first serves produce no successful return under the data provider’s counting rules. The first measures reliability; the second measures immediate damage. Read them together, but never assume they use the same denominator.

The short answer: consistency and damage are different

The distinction is easiest to see side by side:

• First-serve percentage: Of all first serves attempted, how many landed in?
• First-serve unreturned percentage: Of the first serves that landed in, how many were not successfully returned?

The standard first-serve percentage formula is:

First-serve percentage = first serves made ÷ first-serve attempts

Under a first-serve-only convention, the corresponding unreturned formula is:

First-serve unreturned percentage = unreturned first serves ÷ made first serves

That denominator matters. The second calculation does not penalize the server for missing a first serve; it evaluates what happened after a first serve went in. The Tennis Abstract serve-stat glossary uses first serves, rather than all service points, as the opportunity base for first-serve-only statistics.

So first-serve percentage is primarily a consistency or reliability metric. First-serve unreturned percentage is an immediate-effectiveness or damage metric. A player can land many first serves without making them especially difficult to return. Another can land fewer first serves but produce a high rate of aces, service winners or failed returns when the serve goes in.

Neither percentage substitutes for the other.

The denominator changes the meaning

Two percentages can both describe first serves while answering different questions:

Metric	Numerator	Denominator	Interpretation
First-serve percentage	Made first serves	First-serve attempts	How consistently the first serve lands in
First-serve unreturned percentage	Unreturned first serves	Made first serves	How often a made first serve produces no successful return
Unreturned first serves per service point	Unreturned first serves	All service points	How often any service point ends with an unreturned first serve

The first metric uses every first-serve attempt as its opportunity base. The second is conditional: it looks only at first serves that landed in. The third returns to an all-service-point denominator.

A display may shorten “unreturned first serves per service point” to something resembling “first-serve unreturned rate.” That does not make it equivalent to the conditional percentage. If one figure divides by made first serves and another divides by all service points, they cannot be compared as though they share the same base.

Before comparing figures, check:

1. Provider: Who recorded or derived the statistic?
2. Definition: What counts as unreturned?
3. Denominator: Made first serves or all service points?
4. Date range: One match, a tournament, a season or a rolling window?
5. Sample: Which matches and how many opportunities are included?

Percentages from different providers, periods or samples should not be multiplied or combined unless their definitions and coverage are compatible.

Worked example: three useful percentages from 100 service points

Consider a purely hypothetical match containing 100 service points:

• 60 first serves land in.
• 18 of those 60 made first serves are unreturned.

The first-serve percentage is:

60 ÷ 100 = 60%

The first-serve unreturned percentage is:

18 ÷ 60 = 30%

To find the share of all service points ending with an unreturned first serve, multiply the compatible rates:

60% × 30% = 18%

That 18% result must be named precisely. It is the share of all service points ending with an unreturned first serve. It is not the first-serve unreturned percentage, which remains 30% because that statistic uses made first serves as its denominator.

The same arithmetic helps distinguish serving profiles. Suppose two hypothetical players each serve 100 points:

Hypothetical profile	First serves in	Unreturned after landing	Unreturned first serves per service point
More consistent, less damaging	70%	20%	14%
Less consistent, more damaging	55%	35%	19.25%

The first player accesses the first serve more frequently. The second gets fewer first serves in but does more immediate damage when successful. The product shows how those dimensions combine across all service points, without declaring either profile universally superior.

A two-axis way to read serving profiles

A useful serving chart places first-serve percentage on the consistency axis and first-serve unreturned percentage on the damage axis. Without imposing universal benchmarks, it produces four relative profiles:

Profile	Consistency axis	Damage axis	Practical reading
Consistent and damaging	Relatively high	Relatively high	Frequently lands first serves and often prevents a successful return
Consistent, less damaging	Relatively high	Relatively low	Gets many first serves in, but more are returned
Damaging, less consistent	Relatively low	Relatively high	Misses more first serves, but made serves often end the point immediately
Lower on both	Relatively low	Relatively low	Lands fewer first serves and gains fewer unreturned outcomes from those made

A high first-serve percentage does not prove that the serve is difficult to return. Conversely, a high unreturned rate does not show how often the player gains access to that effective first serve.

A community chart comparing the two dimensions used a preceding 52-week window and required at least 15 Tennis Abstract charted matches. That is a useful visual framework, not a complete tour ranking: charted matches are a selective subset, and an eligibility threshold does not guarantee complete or equally representative coverage for every player.

What counts as an unreturned serve depends on the provider

Under the Tennis Abstract convention, “unreturned serves” comprise three categories:

• Aces
• Service winners
• Return errors

That is not universal terminology. The LTA tennis-statistics glossary distinguishes an ace from an unreturned serve based on contact: an ace is untouched, while an unreturned serve is touched but not put back into court.

Neither convention is inherently unusable. The problem arises when a chart using “unreturned” to include aces is compared with one using “unreturned” only for touched serves. Identical-looking labels can represent different event sets.

When reading a serve chart, determine:

• Are aces included in “unreturned serves,” or listed separately?
• How are service winners classified?
• Are return errors included?
• Does touching the ball count as a return?
• Is the denominator made first serves or all service points?
• Were the events coded consistently across the entire sample?

If those details differ, the displayed percentages are not directly interchangeable.

Do not confuse unreturned rate with neighboring serve metrics

Several familiar serve statistics overlap with unreturned rate without measuring the same thing.

Ace rate is narrower. If a provider’s total-unreturned category includes aces, service winners and return errors, ace rate captures only one part of that category. It should not be used as a synonym or automatic proxy.

First-serve points won is broader. The server can win a first-serve point after the return lands in and a rally begins. Unreturned rate stops at the return outcome; first-serve points won includes everything that happens later in the point.

ATP Serve Effectiveness is broader still. It credits aces, service winners, unreturned serves and serves that create an attacking first ball. The ATP’s published tour-level illustration divided average first-serve Serve Effectiveness of 58% into 16% aces, 22% unreturned serves and 20% attacking first balls. Those are the ATP page’s separate component labels and should not be forced into another provider’s definition of total unreturned serves. See the ATP explanation of Serve Effectiveness for its framework.

Short-rally serve efficiency is another broader outcome. It can include points won on the server’s next shot after the return, so it measures early point control rather than unreturned serves alone. A serve can therefore have zero unreturned credit but still contribute to first-serve points won, Serve Effectiveness or short-rally efficiency.

Which metric is better—and what the evidence cannot tell us

The better metric depends on the question:

• Use first-serve percentage to evaluate reliability.
• Use first-serve unreturned percentage to evaluate immediate damage after the first serve lands.
• Use both to describe a serving profile more completely.
• Use their compatible product to calculate unreturned first serves across all service points.

The available evidence does not establish a direct universal correlation between the two percentages or an ideal balance that applies across players and levels. It would be wrong to claim that raising first-serve percentage necessarily raises or lowers unreturned rate.

One observational analysis of U.S. Open men’s singles found a positive but smaller association between first-serve percentage and short-rally serve efficiency than between serve speed and that outcome. But its outcome included first-serve points won by the server’s next shot, so it did not directly measure unreturned rate. Its observational design also cannot establish that changing first-serve percentage causes a change in effectiveness. The scope and model are detailed in the paper on drivers of serve effectiveness.

A rigorous direct comparison would require harmonized point-level data and one stable definition of an unreturned serve. It should also separate or control for factors such as opponent, surface, player level and match context. Otherwise, an apparent relationship may reflect who was serving, who was returning or where the match was played.

The most useful reporting rule is simple: when the underlying data permit, report both unreturned first serves per made first serve and unreturned first serves per total service point. The first isolates conditional damage; the second shows how often that damage occurs across the server’s full workload.

Does touching the return mean the serve was returned?

It depends on the provider’s counting rule. Under the LTA glossary, a touched ball that is not put into court is an unreturned serve, while an untouched serve is an ace. Under the Tennis Abstract convention, the broader unreturned total includes aces, service winners and return errors.

For analysis, “touched” and “successfully returned” are therefore not always equivalent. Check whether the provider defines a return by contact, by putting the ball back into play or through a broader coded outcome.

Is a 15-match charting minimum enough for a tour-wide ranking?

No. A minimum-match eligibility rule is not proof of tour-wide coverage or estimate stability. If matches are selectively charted, players may be represented by different tournaments, opponents and conditions.

Treat such a chart as a view of the available charted sample, not a definitive ranking of every professional server.

First-serve percentage tells you how often the first serve landed in. First-serve unreturned rate tells you how damaging those made first serves were under the provider’s rules. Read both, verify the definition and denominator, and multiply compatible figures only when you want the share of all service points ending with an unreturned first serve.