The Employee Influence Index

Disclosure: I work for Pluxee UK, but this article and idea have no association with my employer.

Perspective and proposal | Employee recognition, social networks, gamification and organizational network effects

Executive summary

My experience across employee benefit and recognition platforms, alongside building and observing social products, keeps bringing me back to the same contrast. Social platforms are remarkably effective at propagating interaction and influence. Employee recognition platforms are built around an inherently positive interaction, yet participation can fade, activity can stay within familiar groups, and conventional gamification can reward volume rather than meaningful cultural reach.

This paper proposes the Employee Influence Index, or EII. It combines three bodies of thinking: social networks show how influence travels through relationships, gamification shows how visible incentives shape behavior, and recognition research shows that acknowledgement can change subsequent performance and helping. The proposal is to change the objective from rewarding more recognition to rewarding the expansion of meaningful positive influence across the organization.

Recent experiments on peer-recognition systems suggest this is not an abstract concern: what a platform ranks and how widely it is used measurably change helping behavior, sometimes for the worse. EII responds by valuing reach, cross-boundary relationships, mentoring and useful feedback, with diminishing returns for repeated interactions inside the same small circle.

Its most valuable output is an organizational view of how positive influence is distributed, not a personal score. Any individual view should be private, reflective and never used for performance or promotion decisions. EII is presented as a testable proposal, not a validated model, and this paper sets out the evidence, the design, the risks and the hypotheses needed to find out whether it works.

1. The observation behind the proposal

The starting point for this proposal is practical rather than purely theoretical. Working with employee benefit and recognition platforms has exposed me to a recurring challenge: turning a valuable capability into a behavior that employees adopt consistently. Working with social platforms has shown me the opposite problem: systems that are exceptionally effective at stimulating interaction, creating feedback loops and concentrating influence, even when the quality of that influence is highly variable.

That contrast led to a simple question. What could employee recognition learn from the mechanics through which social platforms propagate human influence, without inheriting their less desirable characteristics?

Recognition platforms and social networks look different, but their underlying mechanics overlap. Both depend on person-to-person interaction. Both create networks. Both produce visible signals of approval. Both can affect subsequent behavior. The main difference is the objective being optimized. Social platforms typically optimize for participation, distribution and attention. Recognition platforms seek to reinforce desirable behavior, strengthen belonging and help people feel valued. The opportunity lies not in copying social media, but in redirecting some of its network mechanics towards a deliberately positive organizational outcome.

2. Recognition has effects beyond the transaction

The first foundation of the proposal is that recognition changes subsequent behavior, and not only for the person recognized. In a controlled field experiment, Bradler, Dur, Neckermann and Non hired workers for a one-off data-entry job and gave some of them unexpected public recognition. Recognition increased subsequent performance, and much of the measured increase came from workers who had not themselves been recognized (Bradler et al., 2016).

Two caveats matter. The setting was a short, temporary job rather than an established workplace with ongoing relationships, and the finding does not prove a network effect of the kind proposed here. It does show that recognition carries social meaning beyond the individual reward, and that observers respond to it.

Peer-recognition research adds a second layer. In a laboratory experiment, Black (2023) found that peer-recognition systems can increase helping behavior, but that the effect was stronger for helping members of one's own group than for helping people outside it. This identifies precisely the limitation EII targets: without deliberate design, recognition systems may reinforce existing social boundaries rather than extend beyond them.

The important unit may therefore not be the recognition transaction itself, but the relationship that the interaction creates, strengthens or extends.

3. The design of a recognition system changes behavior

A growing body of experimental work shows that design choices inside peer-recognition platforms are not neutral. They change how people help one another, and the direction of the effect depends on what the system measures and displays.

Evans, Presslee and Vandenberg (2025) compared leaderboards within a peer-to-peer recognition program. Ranking employees by the number of recognitions they received reduced proactive helping compared with having no leaderboard at all. Ranking employees by the number of recognitions they gave increased it. The same underlying activity, ranked on a different basis, pushed behavior in opposite directions. This is the closest existing evidence for the central claim of this paper: the metric a recognition system chooses shapes the culture it produces.

Burke, Sommerfeldt and Wang (2025) examined whether a peer-recognition system is used broadly across all subgroups of an organization or narrowly by only some of them. Broad use strengthened a norm that asking for help is acceptable. Narrow use increased help-seeking among members of the groups using the system, but reduced it among non-members compared with having no system at all. Put simply, how widely recognition is distributed across the organization matters, not only how much of it there is.

There is also a cautionary strand. Black, Cecchini and Newman (2024) found that after one company rolled out a peer-recognition system, employees on average reported feeling less appreciated, and their follow-up experiment linked this to social comparison driven by public recognition and leaderboards. Wang (2023) similarly found that public peer-recognition information can make employees who feel overlooked less willing to help colleagues. Any system that makes influence more visible has to take these effects seriously.

Most of these studies are experiments with modest samples, and several took place in laboratory settings rather than live workplaces. Taken together, though, they establish that recognition platforms are behavioral systems, and that the choice of what to count, rank and display is a design decision with consequences.

4. From recognition activity to influence

Most recognition platforms naturally report activity: how many recognitions were sent, who received them, which values were selected, which teams participate and which rewards were redeemed. These are useful measures of adoption, but they say little about how recognition moves through the organization.

Consider two employees. One sends twenty recognitions in a month, almost all to the same three colleagues in their immediate team. Another sends eight specific recognitions that reach several functions, include people they have never recognized before, and create relationships across organizational boundaries.

An activity metric favors the first employee. A network view finds the second pattern more interesting. The Employee Influence Index is an attempt to formalize that difference by asking not only how often an employee participates, but how their positive interactions are distributed through the organization.

5. The organizational network matters

Organizational network research gives this distinction a scientific foundation. Granovetter's work on weak ties showed that relationships outside a person's close circle are disproportionately valuable because they connect people to information their close contacts do not have (Granovetter, 1973). Burt extended this with the idea of structural holes: people who bridge otherwise disconnected groups are exposed to more diverse ideas and are more likely to have good ones (Burt, 2004).

Cross and Cummings studied knowledge-intensive workers in a petrochemical company and a strategy consulting firm. Both the structure of an individual's network and the characteristics of particular relationships were associated with individual performance, and ties that crossed organizational, physical and hierarchical boundaries gave access to distinct information and perspectives (Cross & Cummings, 2004).

None of this implies that every additional relationship improves performance, or that network breadth should become an individual target. It does establish that the informal network through which work happens contains information that an organization chart cannot describe. Organizational network analysis is already an established people-analytics practice. EII applies a similar lens to a specific layer of that network: positive interpersonal influence.

An organization chart represents formal authority. A collaboration graph represents interaction. EII would describe a third layer: where acknowledgement, help and constructive influence actually flow.

6. Three ideas brought together

6.1 Social platforms: influence travels through networks

Social platforms demonstrate at enormous scale that influence is relational. It depends not only on what a person says, but on who they are connected to and whether the network amplifies their actions. EII borrows this network perspective but reverses the usual objective. Consumer networks tend to concentrate influence. A workplace may benefit from distributing it.

6.2 Gamification: incentives shape the behavior that is measured

Gamification supplies the behavioral mechanism, and a warning. Mekler and colleagues experimentally tested points, levels and leaderboards. These elements increased the quantity of activity, but did not significantly increase intrinsic motivation or perceived competence (Mekler et al., 2017). If a recognition system rewards the number of recognitions, employees will optimize for the number of recognitions. Volume is not the same as cultural impact. The leaderboard findings in Section 3 show the same principle applied directly to recognition.

6.3 Recognition: the positive signal

Recognition supplies the prosocial behavior the network is meant to propagate. Rather than manufacturing an artificial interaction for the sake of gamification, the system starts from a behavior organizations already want: noticing valuable contribution and acknowledging it.

7. The central hypothesis

Positive workplace influence can be modeled as a network, and carefully designed incentives can encourage employees to expand that network without reducing recognition to a volume competition.

The change in language is small but important. Instead of asking employees to recognize more people, the system encourages them to grow their positive sphere of influence. The aim is not a workplace popularity contest. It is to make visible how constructive relationships form, where they reach beyond familiar groups, and where they do not.

Two outputs follow from this. The primary output is organizational: a view of how positive influence is distributed across teams, functions and locations, which leaders can use to spot silos and isolated groups. The secondary output is personal: a private view that helps an individual see the shape of their own network. The design choices in the rest of this paper are made with that order of priority in mind.

8. A proposed model for the Employee Influence Index

Imagine each employee at the center of an organizational map. Every meaningful recognition, mentoring interaction or useful piece of feedback creates or strengthens an edge between two people. The sphere of influence grows as those edges reach different parts of the organization.

The resulting shape would not be a perfect circle. One person may have deep connections within Engineering and few elsewhere. Another may act as a bridge between Technology, Finance and Operations. A third may have a smaller network but repeatedly help junior colleagues develop. That picture is more informative than a leaderboard because it makes the objective connection rather than rank.

Possible dimensions of EII

Dimension	What it captures	Possible signal	Design implication
Reach	How many distinct colleagues are positively affected	Count of distinct recipients over a rolling period	Repeated interactions do not count as new reach
Breadth	Whether influence crosses teams, functions, locations or levels	Number of distinct units reached, weighted by organizational distance	Reward expansion beyond familiar groups
Novelty	Whether an interaction creates a new connection	First recognition between a pair with no prior recorded interaction	Diminishing returns for repeated pairs
Depth	Whether recognition is specific and meaningful	Minimum specificity, such as naming a concrete contribution or outcome	Gate other dimensions on depth to discourage shallow praise
Propagation	Whether an interaction is followed by constructive participation from others	Recipients or observers who go on to recognize or help someone new	Value the spread of behavior, not only sender activity
Bridging	Whether an employee connects otherwise weakly connected groups	Brokerage or betweenness measures on the recognition graph	Recognize boundary-spanning behavior
Mentoring	Whether influence is expressed through help and development	Mentoring or help interactions confirmed by the recipient	Extend beyond praise as the only positive behavior
Recipient impact	Whether feedback or support was useful	Optional one-tap recipient signal such as “this was useful”	Credit constructive value, not feedback volume

8.1 An illustrative scoring approach

The following sketch shows how the dimensions could be operationalized. It is illustrative rather than calibrated, and every parameter would need testing.

Each recognition first passes a depth check. A message that names a specific contribution or outcome qualifies; a generic “thanks, great job” is still delivered and still appreciated, but carries little index weight. Qualifying interactions then contribute to an edge between two people, with each repeat between the same pair weighted less than the last. For example, the nth recognition to the same colleague within a period might count for 1/√n of the first. Breadth and bridging are then calculated on the resulting graph rather than on raw counts: how many distinct units a person's edges reach, and whether those edges connect groups that are otherwise weakly linked. Mentoring and feedback contribute only when the recipient confirms they were useful.

Two practical rules follow. Scores should be computed over rolling periods so that the index describes current behavior rather than accumulated history. And the organizational view should be computed first, with individual views derived from the same graph rather than being the primary product.

8.2 Tensions between dimensions

The dimensions do not all pull in the same direction, and the model has to manage that. Novelty rewards recognizing people you have not interacted with before, but recognition of someone you barely work with is likely to be less informed and more superficial, which conflicts with depth. Gating novelty on depth, and weighting new connections more highly when they are followed by continued interaction, is one way to reconcile the two. Propagation rewards interactions that stimulate further participation, but mutual back-and-forth between two people is also the most obvious gaming pattern, so propagation should count only when it reaches someone new.

9. Diminishing returns are central to the design

A basic point system would be easy to game. Two colleagues could repeatedly recognize one another. A team could organize recognition exchanges. Employees could send superficial praise to raise their activity.

EII therefore requires diminishing returns. The first meaningful interaction with a previously disconnected colleague may have substantial network value. Repeated recognition between the same pair still matters relationally, but contributes progressively less evidence that the sender's sphere of influence is expanding.

Black's experiments support this principle indirectly: peer recognition more readily stimulates helping within existing groups than outside them (Black, 2023). Burke and colleagues show the organizational consequence: when recognition is concentrated in some groups, people outside those groups may become less willing to ask for help (Burke et al., 2025). If the goal is to reduce silos, simply encouraging more recognition may reproduce the existing network. The incentive must distinguish between reinforcing an existing tie and extending positive behavior into a new group.

10. Hierarchy and the risk of a flattery index

Hierarchy creates another source of distortion. Managers are expected to recognize their teams, and senior leaders naturally interact with more employees because of their role. Neither should automatically dominate the index.

The reverse is equally important. If recognizing senior colleagues produces a large influence reward, the platform may inadvertently incentivize performative praise. A system intended to democratize influence could become a flattery index.

The model therefore needs role normalization and relationship context. Peer recognition, manager-to-employee recognition, upward recognition and cross-functional interaction may all be valuable, but they should not be treated as identical evidence of network influence. The governing principle is that influence is earned through behavior and relationships, not inherited from a position on the organization chart.

11. Feedback belongs in the model, but carefully

Recognition is not the only way employees influence one another. Mentoring, knowledge sharing and constructive feedback can have substantial developmental value, and a mature influence model should eventually represent them.

The research on feedback requires caution. Kluger and DeNisi's meta-analysis synthesized 607 effect sizes covering 23,663 observations. Feedback interventions improved performance on average, but more than one third of the effects were negative. Their Feedback Intervention Theory argues that outcomes depend partly on where feedback directs the recipient's attention, and that feedback becomes less effective when attention shifts from the task towards the self (Kluger & DeNisi, 1996).

This is a strong reason not to award influence merely because someone gives feedback. The relevant question is whether the interaction was constructive, which is why a lightweight recipient signal matters more than the existence of the feedback itself. Positive recognition can remain public where appropriate. Constructive criticism should be private. Both can contribute to an influence network without being treated as the same social act, and the content of private feedback should never appear in any shared view.

12. The organizational graph is the real product

At sufficient scale, the most valuable outcome is not the individual EII at all. It is the organizational graph produced by the interactions underneath it.

HR and leadership could see whether recognition is concentrated in a small number of teams, whether some departments are structurally isolated, whether particular employees quietly bridge functions, whether new joiners are building connections, and whether influence becomes more or less concentrated after a reorganization or merger. These are questions most organizations currently answer by anecdote.

This changes the ambition. Rather than maximizing the highest individual score, the organization would aim for a healthier distribution of positive influence across the network. The long-term question is not “who is the biggest influencer?” but “how widely is constructive influence distributed?”

It also shapes how the individual view should work. Once a personal score is visible to managers, it will be treated as a performance measure regardless of what the policy says. The safer design is an individual view that is private to the employee by default, framed as a map of their network rather than a number, and explicitly excluded from performance, pay and promotion decisions.

13. Inverting the social media model

Social platforms theoretically give everyone the ability to publish and participate, yet reach tends to concentrate among a small minority of accounts. Network science has long described this pattern, in which well-connected nodes attract a disproportionate share of new connections (Barabási & Albert, 1999). That concentration may be entirely rational for a consumer attention network.

Inside an organization, the desired structure is different. A healthy workplace may benefit from many employees exerting modest positive influence across overlapping groups: one person's network intersects with another's, recognition crosses departmental boundaries, knowledge moves between communities and previously weak ties become usable relationships.

EII borrows the idea of an influencer while deliberately rejecting the idea that influence should be scarce. Its purpose is to create more positive influencers, not a smaller elite of them.

14. Privacy, consent and trust

EII is a proposal to map relationships inside a workforce. Handled badly, it would feel like surveillance, and it would fail for that reason alone. Trust is therefore a design requirement, not a compliance afterthought.

In the UK and EU, relationship data about identifiable employees is personal data. Any implementation would need a clear lawful basis under UK GDPR or GDPR, a data protection impact assessment, and attention to regulator guidance on monitoring workers, such as the Information Commissioner's Office guidance published in 2023. In many European countries, works councils or employee representatives would need to be consulted before introduction.

Several design principles follow. The organizational view should be aggregated, with minimum group sizes so that individuals cannot be identified from team-level patterns. Individual views should be visible only to the employee unless they choose to share them. The content of private feedback should never enter any shared view; at most, the fact that a useful interaction occurred contributes to the graph. Employees should be told plainly what is collected, what is shown to whom and what is never used, and the rule that EII does not inform performance, pay or promotion decisions should be written into policy rather than implied.

15. Fairness and equity

Recognition data is not a neutral record of contribution. It reflects who is visible, who works in public, who shares a location with whom, and which roles naturally cross boundaries. An index built on that data could amplify existing inequities if it is not designed to counter them.

Remote, frontline and part-time employees may have fewer opportunities for visible interaction. Specialists whose work is deep rather than broad may appear less connected while being indispensable. Introverted or neurodivergent employees may express influence through fewer, more substantive relationships. Recognition patterns may also differ across demographic groups in ways that reflect bias rather than contribution.

A responsible implementation would monitor the distribution of EII across these groups from the start, normalize for role and working pattern where appropriate, and treat systematic gaps as a signal about the organization rather than about the individuals concerned.

16. The strongest objections

The concept has serious failure modes. The most obvious is Goodhart's law: when a measure becomes a target, people optimize the measurement rather than the behavior it was designed to represent. Employees could manufacture new interactions. Extroversion could be mistaken for influence. Popularity could be confused with cultural contribution. Employees might feel pressured to build relationships with little relevance to their work.

There is also a motivational risk. Research on the overjustification effect shows that tangible, expected rewards can undermine intrinsic motivation for activities people would otherwise do willingly (Deci, Koestner & Ryan, 1999). If recognition starts to feel like a way to earn index points, authentic praise may become transactional, which is exactly the outcome the recognition-backfire studies in Section 3 warn about.

A further objection comes from the network literature itself. Cross, Rebele and Grant (2016) found that collaborative work tends to fall disproportionately on a small number of highly connected employees, who are at greater risk of overload and burnout. An index that rewards bridging could, if designed carelessly, increase the load on precisely the people who already carry the most. This is another reason to measure the distribution of influence across the organization rather than celebrate its peaks.

These risks argue firmly against treating EII as a performance rating, promotion criterion or universal measure of employee value. A highly effective specialist with a small network is not a worse employee than a broad organizational connector. They contribute differently. EII should represent one dimension only: the breadth and constructive character of interpersonal influence. If it begins to claim more than that, the model has exceeded what its evidence can support.

17. Testable hypotheses

The proposal becomes more useful when expressed as hypotheses that can be falsified rather than as a product feature assumed to work.

H1: Employees exposed to incentives that reward network breadth will establish recognition relationships with a wider range of colleagues than employees rewarded primarily for recognition volume.

H2: Recognition that crosses existing team boundaries will be followed by more cross-team interaction than repeated recognition within established relationships, after controlling for role and prior collaboration.

H3: Diminishing rewards for repeated recognition between the same pair of employees will increase recognition network diversity without materially reducing total participation.

H4: When feedback and mentoring are credited only after the recipient confirms they were useful, recipients will show more subsequent helping behavior than when such interactions are credited on submission.

H5: In a randomized pilot, units assigned to breadth-oriented recognition design will show larger gains in belonging and cross-functional collaboration than control units, and these gains will be associated with a reduction in how concentrated the recognition network is.

H6: Colleagues who observe cross-team recognition, without receiving it themselves, will subsequently increase their own cross-team recognition and helping compared with matched colleagues who were not exposed. This can be tested through a staggered rollout that varies exposure over time.

18. A practical path to validation

A useful first study would not require a complete production implementation of EII. An organization could run a pilot across comparable teams or business units, randomly assigning some units to conventional recognition mechanics and others to prompts and incentives designed around network breadth, new connections and specific cross-team recognition. Randomizing at the unit level rather than the individual level reduces contamination between conditions.

The primary outcomes should not be the EII score itself. They should include network diversity, new cross-boundary ties, continued interaction after an initial recognition, helping behavior, participation persistence, perceived authenticity of recognition, belonging and self-reported collaboration, measured before the pilot, during it and several months after. Negative outcomes should be measured with equal care, including pressure to participate, perceived gaming, feelings of being overlooked, unwanted feedback and workload on highly connected employees.

The purpose of the pilot is to determine whether the proposed mechanism actually changes the network and, more importantly, whether those changes correspond to better outcomes for employees. If they do not, the construct should be revised or rejected.

Conclusion

Employee recognition began with a simple and valuable idea: notice good work and acknowledge it. Digital platforms made that behavior scalable. Recent research shows that the way those platforms are designed changes what the behavior becomes. The next opportunity is to design deliberately for what happens after recognition occurs.

One interaction can strengthen a relationship. Relationships form networks. Networks shape how information, support and social influence move through an organization. The Employee Influence Index proposes treating recognition as that kind of network phenomenon, and testing whether positive influence can be deliberately broadened without making it artificial.

What if, instead of creating a few influential people, we designed the workplace to create many?

Research boundary

The studies cited in this paper support individual components of the proposed model. They do not establish the Employee Influence Index itself as a validated construct. EII, its proposed signals and weighting, and the claim that incentivizing network expansion will improve organizational outcomes remain hypotheses requiring empirical testing. This distinction is intentional and central to the proposal.

An invitation to discuss

I am sharing this as a proposal to be tested and challenged rather than a finished answer. I would particularly value perspectives on three questions. If you work in HR or people analytics, would an organizational view of recognition flow change decisions you make today? If you build or run recognition platforms, which of these dimensions could you realistically measure, and which would employees accept? And for everyone: where do you think this idea is most likely to go wrong?

References

• Barabási, A.-L., & Albert, R. (1999). Emergence of scaling in random networks. Science, 286(5439), 509–512. https://doi.org/10.1126/science.286.5439.509
• Black, P. W. (2023). The effect of peer-to-peer recognition systems on helping behavior: The influence of rewards and group affiliation. Accounting, Organizations and Society, 109, 101454. https://doi.org/10.1016/j.aos.2023.101454
• Black, P. W., Cecchini, M., & Newman, A. H. (2024). When being recognized makes employees feel less appreciated: Evidence regarding when and why peer-to-peer recognition could backfire. Accounting, Organizations and Society, 113, 101565. https://doi.org/10.1016/j.aos.2024.101565
• Bradler, C., Dur, R., Neckermann, S., & Non, A. (2016). Employee recognition and performance: A field experiment. Management Science, 62(11), 3085–3099. https://doi.org/10.1287/mnsc.2015.2291
• Burke, J., Sommerfeldt, R. D., & Wang, L. W. (2025). To ask or not to ask: The effects of broadly and narrowly adopted peer-recognition systems on help seeking. Management Science, 71(10), 8267–8288. https://doi.org/10.1287/mnsc.2023.00318
• Burt, R. S. (2004). Structural holes and good ideas. American Journal of Sociology, 110(2), 349–399. https://doi.org/10.1086/421787
• Cross, R., & Cummings, J. N. (2004). Tie and network correlates of individual performance in knowledge-intensive work. Academy of Management Journal, 47(6), 928–937. https://doi.org/10.5465/20159632
• Cross, R., Rebele, R., & Grant, A. (2016). Collaborative overload. Harvard Business Review, 94(1), 74–79.
• Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation. Psychological Bulletin, 125(6), 627–668. https://doi.org/10.1037/0033-2909.125.6.627
• Evans, J. H., III, Presslee, A., & Vandenberg, A. J. (2025). Peer-to-peer recognition leaderboards and employee proactive helping behavior. The Accounting Review, 100(5), 157–181. https://doi.org/10.2308/TAR-2023-0581
• Granovetter, M. S. (1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360–1380. https://doi.org/10.1086/225469
• Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119(2), 254–284. https://doi.org/10.1037/0033-2909.119.2.254
• Mekler, E. D., Brühlmann, F., Tuch, A. N., & Opwis, K. (2017). Towards understanding the effects of individual gamification elements on intrinsic motivation and performance. Computers in Human Behavior, 71, 525–534. https://doi.org/10.1016/j.chb.2015.08.048
• Wang, P. (2023). When peer recognition backfires: The impact of peer information on subsequent helping behavior. Accounting Perspectives, 22(3), 341–374.