On the day following his birth, the deer fawn, Bambi, is on a walk in the forest with his mother. Still being unsure of his footing, he trips and falls. A young rabbit, Thumper, and his mother then have a short dialog.
Thumper: “He doesn’t walk very good does he?”
Thumper’s mother: “Thumper!”
Thumper: “Yes, mama.”
Thumper’s mother: “What did your father tell you this morning?”
Thumper: “If you can’t say something nice, don’t say nothing at all.”
From Walt Disney’s version of Bambi Directed by David D. Hand Based on the story by Salten (1923)
1. Introduction
Saying nothing at all may be good advice for animals in the forest, but it is bad advice for managers. The primary purpose of a performance review is to provide feedback to employees on their work-related performance to assist them in improving their performance (Cianci et al., 2010; Zhou, 2003). However, there frequently is a performance feedback gap, a gap between the performance and any feedback about that performance. The Bambi Effect is failing to provide any feedback, particularly negative feedback. Research has shown that performance ratings are higher for employees who report receiving more frequent and more specific performance communication from their supervisors (DeNisi & Murphy, 2017; DeNisi & Pritchard, 2006). Feedback, particularly negative feedback, is important as it can enable employees to correct their performance (Glassman et al., 2010; Heneman, 2005; Locke & Latham, 2002; Kluger & DeNisi, 1996). Nevertheless, many managers seem to struggle most with the final and perhaps most important step of the performance review process—providing feedback—especially when the feedback is negative (Simon et al., 2022; Shao & Lee, 2020; Brown, Kulik, & Lim, 2016; Jackman & Strober, 2003). The purpose of this article is to suggest how to close the performance gap and reduce or possibly eliminate the Bambi Effect.
2. Background
Prior research notes that employees not only want feedback, but they also want ongoing feedback because they are pursuing continuous personal and professional development (Henderson et al., 2019; Hattie & Timperlay, 2007; Anseel & Lievens, 2007). This appears to be particularly true for Millennials and Generation Z employees. Feedback usually has been given through the performance review, but its usefulness for enhancing employees’ attitudes and performance has been questioned. One reason for that lack of usefulness is that the feedback from managers is seen as biased (Roberson, Galvin, & Charles, 2008; Kingstrom & Mainstone, 1985; Holzbach, 1978).
The feedforward interview (FFI) or future-focused feedback approach has been suggested as a way to eliminate bias and improve performance (Budworth, Latham, & Manroop, 2015; Kluger & Nir, 2010). By focusing solely on the positive aspects of employee experiences instead of focusing on negative ones, FFI is intended to improve both performance and worker-manager relationships. A field test found that the training required to teach managers how to use the FFI tends to be relatively short. In addition, the test determined that since FFI eliminates the manager’s role as a judge, bias, whether real or perceived, would no longer be an issue for those being evaluated. FFI, thus, partially eliminates the Bambi Effect in that some feedback is provided. However, if the manager is not providing both positive and negative feedback or not providing it well, organizational members do not know whether their performance meets standards or exceeds standards. Nor do they know how to change work that is below standards into work that meets or exceeds standards if they don’t know that it is below standard. Absent that information, the organization also suffers because of the substandard performance of its members.
FFI provides feedback on positive aspects, but it fails to provide constructive feedback on how an individual can improve the negative aspects of their performance. Why is it so difficult to provide negative performance feedback to organizational members? Managers may fail to provide feedback for many reasons. They may fear confrontation, that is, they feel that performance discussions will lead to arguments and threats. Managers may lack confidence in providing feedback because they feel that they really do not have enough knowledge about the employees’ performance to be able to provide accurate feedback. They may want to avoid hurting an employee’s feelings. Managers may avoid providing feedback if they feel that it impacts the employee’s self-perception or personality. And, of course, performance evaluation can be time-consuming, so managers may avoid it altogether.
In addition, such assessments involve the perceptions of both assessor and assessee, which are not likely to match as they have different personal goals. Further, the managers may realize that they do not have the necessary skills to do the task well, especially to defend against possible disagreements with the employee. They may doubt their own self-efficacy. That is, they may doubt their own ability to maintain a positive interpersonal relationship with organizational members after providing negative feedback. It could be that the manager has just been promoted and received no formal training by the organization on providing performance feedback. In this case, both ability and self-efficacy could be impaired. In addition, providing negative feedback may result in additional work and stress for the supervisor. For example, it might mean creating a performance improvement plan or collecting information to justify a negative rating. It could cause fear in the supervisor’s mind of having a critical eye turned onto their own performance if their subordinates’ performance is somehow lacking. Finally, they may fear that providing negative feedback will lead to retribution by the organizational member. It might just seem easier and less risky to take the path of least resistance. All of these deserve the organization’s attention.
As noted earlier, the failure to provide performance feedback was attributable to many different reasons and one of those is the lack of managerial skill. The effort to help managers deal with that lack of skill has produced a flood of research theories, articles, and books on providing feedback, and training (e.g., Aguinis, Gottfredson, & Joo, 2012; Maurer, 2011; Atwater, Waldman, & Brett, 2002). While this has resulted in a huge set of recommendations and rules on how to conduct performance feedback interviews with organizational members, it has not necessarily reduced or eliminated the Bambi Effect as the performance gap persists (Bear et al., 2017; Moss, Sanchez, & Heisler, 2004). Many of the difficulties involved in providing feedback have been addressed through traditional training. However, such traditional training has not been entirely successful as it may be “situation-based,” which is seen as only relevant for a specific performance feedback situation. Or it seems distant or irrelevant if the context is not clear to the parties involved. And, of course, there are individual differences in the willingness or receptivity to learning.
Although the knowledge about how to conduct performance feedback may exist, not everyone who becomes a manager possesses it. Indeed, it seems safe to say that new, inexperienced managers rarely possess this knowledge, except for what they have picked up as employees receiving performance feedback. Thus, even with its limitations, the first suggestion is to use training, especially training dealing specifically with providing feedback. Training has been the recommended method of providing managers with a performance feedback knowledge base. We suggest that a complex training system focused more on corporate performance than individual performance combining feedback with goal setting and incentives be used.
Part of the problem in developing such a training system may be that given time and budget constraints, trainers may find it impossible to furnish managers with a sufficient amount of the performance feedback knowledge base for different situations and individuals with different learning abilities. And managers may feel overwhelmed by the size of the knowledge base or the number of factors, the number of rules, the complexity of the interpersonal relationship, and the many different possible outcomes which must be considered when providing performance feedback. As a result, the performance feedback gap still exists.
Because this traditional training has not been as effective as anticipated, other forms of training have been developed around so-called procedural knowledge—knowing how things are actually done. In this type of training session, a piece of the overall performance feedback knowledge base is extracted and modeled for the manager. For example, the manager of a store may be shown videos of how to provide corrective feedback to an employee about stocking the shelves. Sometimes the manager practices the desired behavior during a role-play exercise. Generally, these training programs appear successful at imparting to managers a portion of the performance feedback knowledge base and demonstrating that managers can, with training, provide performance feedback successfully to organizational members even if it is limited to a small subset of the whole performance feedback knowledge base.
Often, however, the managers can apply the “how-to” knowledge only to situations similar to those experienced in the training skills session. They make a direct link, it seems, only between the knowledge, the behavior, and the situation. For example, the store manager may learn how to counsel an employee about handling complaints regarding product quality but not transfer that knowledge when feeding back information about the employee’s personal habits. In other words, for some people, the training is useful only when the pattern of events matches those presented in the training session very closely. If the situation that presents itself does not closely mirror the training, the manager might lack the self-efficacy to attempt to adjust and apply the new knowledge to a new situation.
This type of training, while far from perfect, has seemed to be the best available for providing inexperienced managers with performance feedback skills. As noted earlier, because FFI eliminates the manager’s role as a judge, bias, whether real or perceived, would not be an issue for those being evaluated. However, there are other ways in which to reduce the judgmental aspect of performance feedback and to make it easier for management to deliver negative information and for employees to accept that feedback.
3. Artificial Intelligence (AI)
One such approach is the use of artificial intelligence (AI) (Tambe, Cappelli, & Yakubovich, 2019; Van Fleet, Peterson, & Van Fleet, 2005). Just as a form of AI, Alexa or Siri, can notify you of appointments or times to take medication, some other form of AI could remind and even assist a manager in providing performance feedback. John McCarthy coined the term AI in 1955 (https://www.britannica.com/biography/John-McCarthy). AI incorporates approaches from multiple disciplines to simulate human intelligence. AI may incorporate image processing, cognitive science, neural systems, and machine learning. The promise of AI in performance management is still in the early stages, but the gap between the promise and reality is closing (Hroncich, 2019; Maulana, Effendi, & Hidayat, 2014; Shaout & Al-Shammari, 1998). One issue in developing a usable AI system is the complexity of Human Resource (HR) outcomes (Cappelli, Tambe, & Yakubovich, 2019; Jia et al., 2018). What really is a “good employee”? Especially in today’s environment, simple measures are insufficient – jobs are interdependent; there are subjective elements to virtually all jobs, and there may be intangibles that are difficult to measure. AI may well be a way to incorporate intelligent decision support systems, methods such as 360-degree performance evaluation, and “scores” by supervisors, the employee, and peers into one system. Such a system yields decisions that are easier for an individual to accept since they are made by an algorithm rather than a human, especially when they are negative. There is a large body of research that shows that algorithms are less biased and often more accurate than human decision-makers (Cowgill, 2018; Miller, 2018).
At present, AI has been used in management for making recommendations to employees about actions they may take—training, career moves, benefits to choose from, and how to handle an interview for example. One vendor, Work Compass, has tried to move away from check-list-based performance reviews and toward continuous discussions, facilitated by phone-based apps, facilitated by natural language processing software. It uses text messages to drive merit pay decisions. While promising, the current state of these systems cannot replace existing practices as they are still incomplete. The difficulties faced by AI systems are the very complexity of performance; measurement issues; fairness, equity, and legal issues; and varying individual reactions to “management by computer”. Nevertheless, organizations should monitor what vendors have available since vendors can combine data from many organizations as they develop their programs. One benefit of AI systems is through a series of questions, the AI system can home in on the specific performance feedback situation that the manager is facing and then provide concrete suggestions. For example, in a negative performance feedback situation, the expert system might ask: “Have you provided performance feedback to this employee in the past?” If the answer were yes, the AI system would branch to appropriate questions about what to say during this feedback session. However, if the answer was no, one of the suggestions made by the AI system might be: “At this stage, you cannot fire the employee per company policy. Do not threaten to fire the employee.”
3.1. Expert Systems
So, while a fully functioning AI system for performance evaluation is still in the future, a particular form of AI known as an expert system (ES) has considerable promise and is currently used by organizations (Mikulić, Lisjak, & Štefanić, 2021; Zhou, Fang, & He, 2011; Hosseininezhad, Nadali, & Balalpour, 2011). ES is an AI software that uses knowledge stored in a knowledge base to solve problems that would usually require a human expert (Van Fleet, Peterson, & Van Fleet, 2005; Peterson & Van Fleet, 1990). ES is familiar in several industries for technical problems but less familiar for managerial ones. So, ES represents another approach that has the capability to reduce the performance gap and the Bambi Effect. Numerous knowledge-based ESs have been developed that claim to provide managers with managerial knowledge and skills in functional areas such as accounting, marketing, and finance and for remedial human resource tasks such as smoothing out workflow, answering routine HR questions and inquiries, and providing employee engagement analytics (Chakraborty et al., 2020; Wagner, Otto, & Chung, 2002: Mentzer & Gandhi, 1992). They are also becoming available in general management for recruiting and hiring plus ESs to provide managers with the knowledge and skill to give performance feedback (Fitriana et al., 2019; Šooš, 2019; Peterson, Van Fleet, Smith, & Beard, 2003).
A conceptual model of an expert system is shown in Figure 1. The expert system also known as a knowledge-based system consists of facts about a specific situation such as giving performance feedback and a set of heuristic knowledge (rules expressed as if/then statements). This knowledge-based is collected from both human experts and the literature on the topic. Given that there is an extensive knowledge base on how to give performance feedback and many human experts, this knowledge and rule base could be acquired and then entered into an expert system shell such as visirule and GeneSys DX. Once the knowledge base is created, the expert system shell has an already built-in inference engine that
![]()
Figure 1. A conceptual model of an expert system.
applies if/then rules to the knowledge base to provide advice on a given topic in this case the ability to provide performance feedback. Through a user interface and the inference engine, which translates inquiries asked of the user, the user is provided knowledge to solve a complex problem such as giving performance feedback (Khan, Rehman, & Amin, 2011).
The following insert provides a simple example of a manager interacting with the expert system and the recommendation from the system to the manager.
A reminder: a critical question the system would ask is: Have you provided feedback to this employee in the past. If the answer is no, the system would tell the manager that they could not fire the employee at this point. Many inexperienced managers get frustrated with poor performance and their first reaction is to fire the employee. This often leads to the employee filing a grievance and the manager having to back down and by so doing losing credibility and lowering their self-efficacy.
In an extensive expert system such as one designed to provide advice on giving performance feedback, there could be as many as 500 if/then rules and thousands of bits of knowledge in the knowledge base. No manager would be expected to answer all inquiries since the expert system makes decisions based on the answers of the user. For example, if the answer to the first inquiry above was “No negative performance feedback will not be given,” the expert system would ignore all the inquiries about negative feedback and would not provide advice on that situation to the manager.
3.2. Examples
But do they work? Proton Expert Systems & Solutions Pvt. Ltd suggests that it can be used in various human resource activities including recruitment, employee development, talent management, and workforce management (https://www.protontech.in/erp_iconnect.php). BambooHR does performance appraisal along with other HR functions (https://www.bamboohr.com/). Expert system applications in performance evaluation have been reported for a Mexican firm (Aguilar Lasserre et al., 2014), an Iranian IT company (Hosseininezhad, Nadali, & Balalpour, 2011), and CV. Artha Mandiri, a travel company in Indonesia (Fitriana et al., 2019). Applying fuzzy set theory in an ES has been shown to improve a performance evaluation system in a university (Walek & Farana, 2017). One widely used platform, Culture Amp, uses predictive analytics and so while not precisely an expert system it is similar in its operation (https://www.cultureamp.com/). In addition, research has shown that ESs can work (Mikulić, Lisjak, & Štefanić, 2021; Cascante et al., 2002). To verify those results, the authors conducted their own study of an ES (Peterson & Van Fleet, 1990). In an experiment, they found that the use of that ES would enable subjects to perform more effectively the task of giving negative feedback—an ES can really help managers do their jobs better. This is critically important since Millennials and Generation Z organizational members do not want the old-style command and control boss but want managers who can coach them in personal and professional development. We all know that one of the critical skills of a good coach is the ability to deliver timely performance feedback.
4. Results
Two major conclusions are indicated here regarding training time and applicability. First, if the system is well designed and tested before implementation, training time and costs may be substantially reduced. Obviously, time depends upon the “user-friendliness” of the ES. The training involved in the two studies mentioned here was brief [about one hour] and yet it yielded good results. This would mean that new managers could more rapidly become as effective as experienced ones.
Second, organizations may need to use several different implementation procedures when introducing ESs. A major moderator might be experience—and not just the managerial experience of the personnel involved but also experience with computers and software. Experience with the task may speed up or eliminate the need for the ES; experience with computers and software may make the training faster or easier.
Training directors and organizations should seriously consider the benefits of using an ES in the area of performance feedback, as it clearly has the potential to help managers provide more useful feedback in a less threatening style. An ES should be developed with the participation of those whose jobs are to be involved, and possibly the assistance of a consultant. Involving those impacted increases, the acceptance of the Es, and, since the manager’s role is greatly diminished, bias is also reduced or eliminated. An ES then would seem to be an efficient and effective way of eliminating the Bambi Effect—using technology that is already familiar in industry to overcome management’s reluctance to say nothing but good. Using an ES, managers can actually help employees understand what they need to do to improve performance. In a recent webinar by Nick Gallimore, Director of Talent Transformation and Insight at Clear Review, another software tool for conducting performance management, thirty-nine percent of participants in the webinar reported that their current performance management process did not improve performance, engagement, or wellbeing. The Clear Review software product leads the manager and the employee through the process of setting clear goals, frequent feedback sessions, and meaningful coaching conversations driven by rich performance data. Performance reviews augmented with AI have a positive and significant impact on the effectiveness of the human resource management process.
The ES used in the experiment has since been merged into another information technology company. ES is helping facilitate the transition of the boss to a coach. Expert systems are starting to play a role in performance management and coaching. For example, tools including Bravely, BetterUp, and Marlow are mobile apps that provide confidential personalized assistance to workers by connecting them with coaches. Fast food restaurants such as Dig Inn have adopted this new technology and so has M&M’s the manufacturer of Mars candy bars. An expert system that can provide 360-degree feedback to an employee has been developed (Aguilar Lasserre et al., 2014).
So, FFI is a partial solution, AI has future potential, and ESs in conjunction with training should close the performance gap and reduce or possibly eliminate the Bambi Effect. Try one and see for yourself (Leonard-Barton & Sviokla, 1988).