Sixty-one female football players from two Swiss U19 women's national league teams and two U19 grassroots teams volunteered to participate in this study. A total of 36 players (mean ± SD age: 16.7 ± 1.2 years; Team 1: n = 9, Team 2: n = 10, Team 3: n = 12, Team 4: n = 5) met the inclusion criteria: (1) participation in both pre- and post-test, (2) participation in at least five intervention training sessions, (3) outfield playing position (goalkeepers were excluded). On average, elite players had 10.1 ± 1.9 years and grassroots players had 5.4 ± 3.0 years of football experience. Written informed consent was obtained from all participants or their legal guardians in the case of the underage participants. The ethical aspects of the study were approved by the institutional review board of the Swiss Federal Institute of Sport Magglingen SFISM (196_LSP_04_2023) and the study was conducted in accordance with the latest version of the Declaration of Helsinki.

Training content was designed in collaboration with the participating teams’ coaches to foster scanning development within a representative learning design. Thus, four game formats with variations and four exercises were developed to expose players to a wide range of game situations and sudden changes, such as shifts in game direction or moving visual cues that need to be identified during play. These conditions required players to make quick decisions and continuously reassess the game environment. Table 1 provides an overview of the game formats and exercises from the playbook used during the scanning-focused training designed to enhance scanning behavior in the intervention group. Details are provided on the team setups, objectives of the exercises and game formats, and key training elements. The detailed instructions are provided in the supplementary information (Supplementary File 1).

Before implementing the intervention, exercises and game formats were pre-tested with an independent team to refine instructions and optimize the new playing formats. Based on these pre-tests, coaches of the participating teams received printed instructions detailing all exercises, game formats, and corresponding instructions. Before the start of the study, coaches were introduced to the game formats through pre-tested video footage and personal on-field instructions allowing the coaches to clarify any potential questions. During the intervention period, coaches lead the scanning-focused training sessions in order to maintain normal training conditions and natural player behavior. For each scanning-focused training session, the coaches could choose the exercises and game formats (including at least one game format) to match the overall focus of the training session. For intervention fidelity, coaches documented each training session immediately afterward. Reports included details on selected exercises, game formats, modifications, duration, remarks, and participating players. These reports were descriptively reviewed by the study leader to ensure consistent implementation of the intervention across teams and to confirm appropriate variation in selected exercises and formats from the playbook. No substantial differences between the teams or outliers were identified. Furthermore, each team was observed twice from the sideline by the study leader during intervention training sessions, without any interference. Post-training discussions with coaches were conducted without the players’ presence to ensure an open exchange of feedback and observations. Slight modifications (mainly refined instructions) to exercises and game formats were conducted based on coaches’ feedback throughout the intervention period to further refine the study approach. These adaptations were primarily designed to help coaches manage variations in player numbers or infrastructure and ensure consistent training conditions throughout the intervention.

As the teams were located up to 90 km apart from each other, players were assigned to either the control or intervention group based on their team affiliation, following a quasi-experimental rather than completely randomized design. Both the control as well as the intervention group consisted of one grassroots and one elite team. The intervention was conducted over five weeks, including one to two scanning-focused training sessions per week. Only one scanning-focused training session was conducted in weeks with two official matches. This was the case for the elite teams in two out of the five-week intervention period. Each scanning-focused training session included a 30-minute block of scanning-focused exercises and game formats, which was conducted immediately following the warm-up. During that 30-minute block, the control group followed their regular training routine, emphasizing ball control and first touch within their game formats and exercises. However, they did not engage in scanning-specific game formats and exercises (Supplementary File 2).

Test sessions were conducted one week before (pre-test), one week after (post-test), and three weeks after (retention test) the five-week intervention period. Coaches assigned outfield players from their team to four balanced sub-teams, ensuring equal distribution of physical, technical, and tactical abilities. Players remained in the same sub-teams throughout all test sessions. After a standardized 15-minute warm-up, the test session included two sets of 4 × 4-minute SSG, with a 2-minute break between games and a 5-minute break between the sets. The games were played on two 40 × 30 m pitches, following modified 11v11 rules: no offside rule and goalkeepers were allowed to restart play after goals and out of play situations (6). Throughout the session, multiple balls were available for the out-of-play situations and were made available immediately by coaches positioned around the pitch to maintain continuous gameplay. A 1:2:1 playing system was implemented to ensure that players rotated through all four positions within each set and to minimize position-specific effects on performance. All 4v4 games were manually recorded using two 4 K digital cameras (HDR-CX700VE, Sony, Minato, Tokyo, Japan). The cameras were mounted on tripods and positioned 2 m from the long side and 7 m from the short side of each pitch. No zoom was applied during filming to maintain a consistent field of view.

Filming and tagging procedures followed the methodology outlined by Hintermann et al. (6). All video footage was analyzed using the same tagging panel as in the previous study and assessed by the same two expert raters. Interrater reliability was assessed using intraclass correlation coefficients (ICCs), calculated via mixed-effects models under the assumption of absolute agreement with a single rater. Estimates and 95% confidence intervals (95% CI) were computed using the psych-package in R (28). A randomly selected subset of 108 game situations (4% of total) from four games involving four players was independently analyzed by both raters. The resulting ICC was 0.89 [95% CI (.86,.92)], indicating good to excellent interrater reliability (29). To assess intrarater reliability, each rater independently re-analyzed four games of two randomly selected players (4% of total). The ICCs demonstrated good to excellent intrarater reliability for both raters [rater 1: ICC = 0.86, 95%CI [.79,.90]; rater 2: ICC = .93, 95% CI [.81,.99], respectively] (29).

To address the main research question and account for dropouts, scanning behavior was compared between the intervention and control groups, which pooled elite and grassroots players. Skill level and the percentage of positive subsequent actions were then analyzed in the subgroups. Thirty-one out of 36 players (14 in the intervention and 17 in the control group) participated in the retention test two weeks after the post-test and were analyzed as another subgroup.

All data are presented as mean ± SD and the alpha level set at p < 0.05. The assumption of normality was checked through visual inspection, evaluating the predicted and standardized residuals. Normal distribution was confirmed by a random distribution around zero in the scatter plot, a diagonal straight line in the QQ-plot, and a Gaussian distribution in the histogram (30). A two-way repeated measures analysis of variance (ANOVA) with Bonferroni-adjusted post-hoc comparisons was calculated to determine potential differences and interaction effects in the mean number of scans per game situation between groups (control vs. intervention) across two or three time points (pre- vs. post- vs. retention test) and two skill levels (elite vs. grassroots). Effect sizes were interpreted using partial eta squared (η²), with thresholds of 0.01, 0.06, and 0.14 indicated a small, medium, and large effect (31).

To assess the performance of subsequent actions, only game situations with scanning behavior before ball reception were analyzed. Mean percentages of positive outcomes were compared using a two-way ANOVA, with skill level (elite vs. grassroots) and time (pre-test vs. post-test) as independent variables. Additionally, a Mann–Whitney U test was used to compare the mean percentage of positive outcomes between game situations with and without scanning before first ball contact.

The results of the training intervention are presented in Table 2. In line with our first hypothesis, the intervention group significantly increased their mean number of scans per game situation from the pre- to the post-test (p = 0.002). Additionally, during the post-test mean number of scans were significantly higher in the intervention than the control group (p = 0.002). During the intervention period the control group also exhibited a slight increase in mean number of scans per game situation, which however, did not reaching statistical significance (p = 0.088). The mean changes from pre- to post- (delta) in the number of scans per game situation were 0.55 ± 0.37 (95% CI: 0.37–0.74) in the intervention group and 0.31 ± 0.35 (95% CI: 0.16–0.46) in the control group.

Figure 1 provides an overview of the relative change in the mean number of scans per game situation from pre- to post-test for intervention and control groups.

The two-way ANOVA showed that elite players had a significantly higher percentage of positive subsequent actions compared to grassroots players (elite: 75.4%; grassroots: 65.1%; F_1,68 = 7, p = 0.011, η² = 0.09). There was a trend for time that indicated a potential decrease in percentage of positive outcomes from pre- to post-test, though this effect was not statistically significant (pre-test: 74.3%; post-test: 67.1%; F_1,68 = 3, p = 0.074, η² = 0.05). This result does not support our second hypothesis, which predicted that an increase in the number of scans would be associated with a higher success rate. However, a comparison of game situations with and without scanning behavior showed a significantly higher percentage of positive outcomes when players scanned before their first ball contact (with scanning: 70.6%; without scanning: 62.5%; z = −3.45, p = 0.001).

Differences between elite and grassroots players are presented in Table 3. The table shows mean number of scans per game situation before and after the training intervention for both skill levels, across the intervention and control groups. The two-way ANOVA revealed significant main effects of time (pre- vs. post-; p < 0.001) and condition (control vs. intervention; p < 0.001) in the elite group. In the grassroots group, a significant main effect of time (p = 0.007) but not condition (p = 0.367) was found. In elite players, the mean changes (delta) in the number of scans per game situation from pre- to post- were 0.66 ± 0.41 (95% CI: 0.41–0.91) in the intervention group, compared to 0.39 ± 0.43 (95% CI: 0.11–0.67) in the control group. In contrast, in grassroots players, the corresponding mean changes were 0.35 ± 0.12 (95% CI: 0.24–0.45) in the intervention group and 0.25 ± 0.29 (95% CI: 0.09–0.41) in the control group.

Among elite players, those in the intervention group performed significantly more scans in the post-test compared to the control group (p = 0.005). Additionally, elite players in the intervention group exhibited a significant increase in scanning behavior from pre- to post-test (p = 0.001). In contrast, no significant post-hoc effects were observed among grassroots players.

An overview of the relative changes in the mean number of scans per game situation from pre- to post-test in both the intervention and control groups, across both performance levels, is presented in Figure 2.

The 2-way repeated measures ANOVA (n = 31) reported statistically significant main effects of time (pre- vs. post- vs. retention test; F_2,87 = 12, p < 0.001, η² = 0.21) and condition (control vs. intervention; F_2,87 = 23, p < 0.001, η² = 0.21). The post-hoc test indicated an increase in the mean number of scans in the intervention group from pre- to post-test [pre: 0.86 ± 0.36, 95% CI [0.67; 1.05]; post: 1.47 ± 0.53, 95% CI [1.19; 1.74]; p = 0.004] and from pre- to retention test [retention: 1.47 ± 0.55, 95% CI (1.18; 1.75); p = 0.004]. As previously shown in Table 2, no significant differences were observed in the control group between pre- and post-measurements. Furthermore, no significant changes in number of scans were detected between the post- and retention test in the intervention or control group. However, there were significant differences (post: p = 0.018; retention: p = 0.015) in post- and retention test results between the two groups.

The progression in mean number of scans per game situation from pre- to post- and retention test is illustrated in Figure 3, with comparisons across intervention and control groups (Figure 3A) and performance levels (Figure 3B).

While the findings of the present study provide valuable insights into the development of scanning behavior in U19 female football players, several limitations should be acknowledged. Firstly, teams were not completely randomly assigned to intervention and control groups. Instead, group allocation was based on team affiliation and regional proximity, which may have introduced selection bias. Secondly, from the original sample size of 61 players, only 36 completed enough training and test sessions to be included in the final data set, which reduces statistical power and generalizability despite significant differences. Thirdly, the study was conducted with U19 female players and future studies have to verify the findings in other age groups and skill levels. Lastly, the retention test was conducted three weeks after the intervention. To investigate long-term sustainability of the observed improvements longer follow-up periods are required.

The results of this study provide helpful guidance for coaches aiming to promote scanning behavior in young female football players. Since significant improvements were only demonstrated by elite players, the study emphasizes the necessity to actively foster and develop scanning behavior in high-performance settings where perceptual-cognitive skills already play a central role (18). Rather than treating scanning as an isolated skill, coaches should consider it an integral component of overall player development. To support this, training environments should follow the principles of representative learning design. For example, SSGs ensure that the practice task simulates the perceptual and decision-making demands of actual match play (25). Due to their adaptability, SSGs are particularly well-suited to manipulate environmental, individual, or task-related constraints, hence achieve specific tactical learning objectives (40) and provide a variety of options for adapting exercises and game formats. Scanning cues can be added to existing drills to enhance perceptual demands. The flexibility of the proposed game formats (see Table 1) enables coaches to adapt the content to suit their players’ developmental stage and training environment. For elite players who often possess a solid foundation of perception and motor skills, incorporating scanning task into SSGs appears to be an effective method of refinement. In contrast, grassroots players may first benefit from more foundational training in motor control and perception-action coupling before similar interventions produce comparable improvements.

This study aimed to initiate the establishment of evidence-based long-term training methods for scanning behavior in sports. Future research should investigate further age categories to provide a more comprehensive understanding of the development of scanning behavior across the career of female football players. While the game formats and exercises were well received by both players and coaches, the results obtained from the SSGs in the test sessions still need to be confirmed during 11v11 match play.