‘ You had to be the Detective ’ : Implementing Workfare in British Employment Services

British employment service delivery has shifted towards a model primed on core ‘ workfare ’ objectives – that is, enforcing behavioural compliance to work-related duties and expanding participation in work. Nevertheless, significant gaps remain in current knowledge about how workfare is implemented daily by frontline staff. The existing international street-level research on employment service delivery reveals how workers use a range of discretionary practices to achieve workfare objectives. Yet this research largely ignores how, in practice, a key aspect of enforcing behavioural compliance and encouraging work participation is through contending with its opposite – behavioural non-compliance. Analysing 13 interviews with frontline staff, this article contributes to street-level knowledge by revealing the ways managers and workers in British employment services are encouraged to detect and correct variations of claimant non-compliance.


Introduction
The last four decades have seen the unfolding of a multi-national 'workfare project' (Brodkin, 2013a). Social policy responses to poverty and unemployment across the world have moved away from employment services prioritising need, legal entitlements and voluntary (re-)training programmes. These services have become increasingly primed on a common set of compliance-focused policies for both benefit claimants and frontline staff. For those claiming out-of-work benefits, policies mainly converge around individualised case management, routinised performance of mandatory work activities (job-searching, work/training placements), and (threats of) benefit reductions, suspensions or disqualifications for a (wider) range of non-compliant behaviours. Meanwhile, employment service staff have seen their daily work activities become increasingly governed by digitalised performance monitoring and compliance auditing management systems geared heavily towards achieving job outcomes and/or reducing claimant numbers (Considine, 2001;Brodkin and Marston, 2013). The workfare project has been frequently justified by political officials across multiple nations with reference to a perceived need to correct the behavioural dysfunctions of poor, unemployed groups (Soss et al., 2011: 28-38;O'Sullivan et al., 2021: 32;Fletcher and Redman, 2022).
A growing body of international scholarshipcommonly referred to as 'streetlevel' research (Lipsky, 2010) has investigated how frontline workers delivering employment services have constituted the workfare project in daily, situated practice. One strand of street-level scholarship, among many other strands, has explored how frontline staff translate and implement the policies at their disposal to enforce behavioural compliance in their everyday capacity.
The present article reveals new evidence suggesting that existing street-level research largely ignores important features of enforcing compliance in frontline practice. It is known that benefit claimants can struggle against and subvert compliance-focused policies and practices in a variety of ways to prioritise their own needs and interests (Redman, 2021). Far less is known, however, about how frontline staff detect occurrences of claimant non-compliance. Moreover, while there is a sizeable body of international research exploring how compliance is enforced, it is contended here that there is more to learn about how, upon detecting noncompliance, frontline staff may then go about using compliance-focused policies and practices to correct non-compliance. Analysing 13 interviews with frontline workers occupying a range of positions in British employment services, this article contributes to street-level knowledge by demonstrating how staff go about detecting and correcting variations of claimant non-compliance.
The interviews also revealed how practices conceived here as 'detection' and 'correction' are not inevitable or homogenously implemented but subject to the same inconsistencies typically present on the frontline of service delivery (Lipsky, 2010). Consequently, this article additionally explores the driving factors underlying practices of detecting and correcting variations of claimant non-compliance. It shows how the inconsistent implementation of such practices tend to be determined by one or a combination of the formal (compliance-focused) policy framework, managerial and administrative pressures, the discretionary power wielded by frontline staff and their perceptions of claimant behaviours and appearances.
To adequately explain how existing research largely neglects practices of detecting/ correcting non-compliance and how these practices can be inconsistently implemented, the article begins with a tailored review of the street-level research tradition by outlining the strand of adjacent international literature that explores how the workfare project is constituted in the delivery of employment services. This is followed with an explanation of how the workfare project has unfolded in Britain. The methods used are then discussed. The article concludes with an analytical presentation of the research findings and some brief reflections on how these findings might be interpreted by actors situated at opposite poles of the workfare project.
2. The workfare project in street-level research Over the last four decades, researchers across the world have set out to investigate how the unfolding workfare project has impacted the day-to-day delivery of employment services. This research tends to draw heavily on a distinctive research tradition, advanced by Lipsky (2010), whose 'street-level' approach to policy analysis notably examined why disparities often emerge between policies in their paper form (as conceived by policy makers) and policies in their practiced form (as executed by frontline workers). Lipsky (2010) suggests that disparities between policy conception and execution stem from one or any combination of: discretionary power possessed by frontline staff to interpret and implement policy; organisational cultures/goals, performance management systems and (lack of) available resources in public service offices; and relationships/interactions between managers, frontline workers and service-users. More recently, Zacka (2017: 32) has emphasised the role that 'moral dispositions' play in shaping how frontline workers 'make use of their discretionary power', with workers inherently leaning toward softer, indifferent or tougher approaches to service delivery. These variables can all shape and distort policy when implemented in day-to-day practice.
Some researchers have drawn inspiration from Lipsky's seminal ideas to reveal how compliance-focused policies can be scarcely or inconsistently implemented in the routine delivery of employment services. This is an outcome often attributed to the heightened administrative burdens associated with enforcing compliance, such as insufficient staffing or training, or additional paperwork (Bryson and Jacobs, 1992: 47;Wright, 2003: 128-130). It can also be indicative of variations in frontline workers' 'personal dispositions' towards compliance-focused policies and/or their service-users, with some opting to use discretionary powers in favour of more lenient approaches to service delivery (Wright, 2003: 226-229;Dubois, 2010: 92-97;Fletcher, 2011).
Nevertheless, contemporary research has also shown how the gradual integration of stringent performance management systems can operate in conjunction with the compliance-focused policy framework to curtail workers' discretionary powers and produce more standardised forms of service delivery that centre on enforcing compliance and expanding work participation. In America, Soss et al. (2011: 207-8) note how the introduction of performance management systems had curtailed workers' discretion, encouraging them to adopt 'a disciplinary stance towards clients' (cf. Brodkin, 2013b: 156-159). As staff had 'few tools at their disposal : : : to motivate client compliance', intense managerial pressure to achieve job outcomes frequently encouraged them to resort to (threat of) sanctions (Soss et al., 2011: 227-229). In Australia, McGann et al.'s (2020: 484) longitudinal surveys (1998-2016 found that workers have become more attentive to implementing compliance-focused policies while simultaneously feeling increasingly unable to decide 'what to do' with their caseloads, which correlates with the intensification of compliance auditing and outcome measurement tools in employment service delivery (O'Sullivan et al., 2021: 11-24). Relatedly, in Britain, Jordan (2018), Fuertes and Lindsay (2016) found that service delivery mainly revolved around a standardised set of practices, such as encouraging independent job-search activities, speculative 'scattergun' job applications, discussing interview techniques and CV polishing. Another strand of street-level research suggests that the integration of performance management systems may not so much curtail discretionary power but redirect it in accordance with the compliance-focused policy framework. In Germany and the Netherlands respectively, Senghaas et al. (2019: 621) and Grandia et al. (2020) demonstrate how staff use discretion to enforce compliance 'in a persuasive rather than a coercive way', utilising a range of rapport building and motivational techniques to win the trust of service-users. Conversely, research in Britain found that staff can use discretion in coercive and socially harmful ways to meet performance targets . New managerial pressures and stigma-laden perceptions of claimants can sedate workers from the harmful outcomes of their actions, encouraging some to pursue targets by developing innovatively pernicious frontline practices. Kaufman (2020: 216) argues that new managerial pressures have 'opened up a new disciplinary axis of discretion', altering both the whereabouts and ways in which workers exercise discretion. It is now located in workers' decisions to 'intensify or moderate' the degree to which they enforce compliance and expand work participation. This can be done by articulating their power and authority via direct or indirect threat of sanctions, by increasing or decreasing the number of work-focused appointments, by mandating or abrogating tasks perceived by claimants as difficult or unattractive, and by adjusting claimant employment aspirations upwards or downwards to align with their perceived capabilities. Kaufman (2020: 216) argues that decisions to intensify or moderate compliance are rooted in a range of 'moral, normative and calculative' rationalessuch as workers' perceptions of service-users' deservingness (of support or sanctions) or their proximity to the labour market and the perceived likelihood of achieving job outcomes.
The present article reveals evidence suggesting that the existing street-level literature largely ignores a key dimension to the new 'disciplinary axis of discretion'. Researchers have shown how frontline staff develop 'moral categorisations' of claimants based on their perceived level of (non-)compliance; often harbouring suspicions about the veracity of information provided around their willingness to work and/or the availability of alternative income streams (Wright, 2003;Dubois, 2010;Kaufman, 2021;O'Sullivan et al., 2021). Nevertheless, this existing research largely ignores how frontline staff may deploy another set of practices to verify or falsify and act on their suspicions of claimant non-compliance. The findings section(s) of this article will address this gap, revealing how frontline staff can deploy a range of discretionary practices to detect variations of claimant non-compliance and inform their perceptions of claimants who 'deserve' more intensive compliance-focused policies/practices. The findings also reveal how, upon detecting 'deserving' claimants, staff have a 'few [more] tools at their disposal' (cf. Soss et al., 2011: 227-229) to correct non-compliance than existing research presently suggests.
3. The workfare project in British social policy Detecting and correcting variations of claimant non-compliance has been a central pre-occupation of those (political and intellectual actors) who have played an instrumental role in unfolding the workfare project in Britain. Over the last four decades, the workfare project has been heavily influenced by (neo-)Conservative theories on the behaviour of people enduring poverty and persistent unemployment (Deacon, 2000). One influential theory, notably advanced by American scholar Mead (1986), held that the expansion of permissive, post-war anti-poverty programmes (e.g., cash benefit administration, employment services) had fuelled increasing rates of 'behavioural dysfunction' among poor communities-such as dependence on benefits, unwillingness to take low-wage menial jobs, and engagement in illicit/informal working activities. Dysfunction increased, according to Mead (1986: 54-61), because post-war programmes were designed on deterministic sociological assumptions that found it almost inconceivable to lay the blame for poverty and unemployment on the behaviour of poor people 1 . Consequently, benefit programmes failed to levy behavioural standards on users in return for support, exempting poor people from pressures to comply with mainstream standards that non-poor people typically endured through schools, neighbourhoods and workplaces. Mead (1986: 22) described this group as a growing 'underclass' who were adrift from the non-poor classes and required a range of compliance-focused policies to re-integrate them into mainstream society. Specifically, Mead (1997: 59) advocated for workfare in the narrow sense-i.e. shifting from voluntarist to mandatory work/ training programmes that 'reduce or deny benefits to clients who do not cooperate'. He also advocated for workfare in the broader sense-i.e. public services focused on 'helping and hassling' users through intensive personal 'case management' and 'monitoring' of their behaviour, with benefits conditional on routinely satisfying work-related requirements (Mead, 1997: 61-63). Mead (1997) thought that workfare could correct the behavioural dysfunctions of the 'underclass' by pushing people into available jobs at the earliest opportunity, by ensuring continuous pressure was placed on those who were out-of-work to comply with mainstream behavioural standards, and by 'smoking out' those who were unwilling to work or working illegally.
Mead's ideas have 'had a profound impact upon all points of the party-political spectrum' in Britain (Deacon, 2000: 8). Their lasting influence can be seen in the multiple re-framing(s) and re-design(s) of employment services pursued by successive governments since the 1980s (Price, 2000).
These ideas were again discernible in a series of post-2010 changes to the employment service, as the newly elected Coalition government summoned underclass language and imagery to legitimise a battery of 'narrow' and 'broad' workfare reforms   Grayling (2012) declared that the Coalition was 'fighting a battle to stop claimants slipping back into the benefits system by the back door'. The Coalition 'fought' this 'battle' by building on the workfare regimes left by previous administrations to design a more compliance-focused employment service for both (1) claimants and (2) frontline staff.
(1) Multiple benefit claiming subgroups were targeted with several major policy changes. Policymakers sought to expand work participation by reducing the real take-home value of cash-benefits through benefit freezes and an initial benefit cap. Mandatory Work Activity (MWA) placements lasting 30hrs per week over four-week durations were introduced in 2011, primarily targeting claimants perceived by frontline staff as lacking commitment to job search activity or those unmotivated and lacking work experience (ICFGHK and TNS-BRMB, 2012). MWA was followed by the introduction of an 'enhanced sanctioning regime', active from 2012, which increased the maximum period of disqualification from 26-weeks to three years and featured a tiered system to impose harsher penalties for recidivists (Adler, 2018: 37). The enhanced regime retained some core features of its predecessor, such as a separate referral and decision-making mechanism. In practice, employment service workers initially raise a non-compliance doubt by collecting and referring relevant evidence to an Independent Decision Maker who then assesses the case according to specific criteria in a separate office (DWP, 2017).
From 2013, policymakers began to place a wider range of claimant sub-groups under compliance-focused policies with the (ongoing) rollout of Universal Credit (UC). UC extends work-related activity pressures and threat of sanctions to those in low-pay, insecure or part-time work, encouraging in-work claimants to find more or better paid work. Individualised case management, operational since the 1990s, became increasingly digitalised under UC. Most claimants now manage their claim, provide job-search evidence and liaise with frontline 'work coaches' through an interview or a combination of face-to-face work-focused interviews (typically fortnightly), signing on appointments at a local Jobcentre Plus (JCP) office and interactions through an online journal. This includes applying for jobs through 'Find a Job' (formerly 'Universal Jobmatch'), a vacancy website which can allow work coaches to monitor job applications. Case management increasingly revolves around continual supervision and maintenance of the 'claimant commitment' (replacing 'jobseekers agreements'). Commitments consist of detailed back-to-work plans which require most claimants to provide written evidence of full-time (35hrs per week) work-related activity. The claimant commitment progresses a major shift in eligibility criteria, ongoing since the mid-1980s (Bryson and Jacobs, 1992: 18-9), away from full-time work availability towards full-time work-related activity. Since claimants can be sanctioned for failure to comply with any item written in their commitment, it also broadens the scope of non-compliant behaviours liable for sanction (Fletcher and Wright, 2018). Face-to-face interactions in JCP offices are supported by (G4S) security officers who, since the 1990s, have been contracted into employment services to manage claimant inflows and pacify any hostile or anti-social behaviours.
Changes were also made to the operational framework governing (2) frontline employment service staff. The introduction of Work Programme (hereafter WP) in 2011 saw large sections of the claimant count outsourced to (quasi-)private contractors for more intensive employment support. WP was targeted at long-term unemployed people and those receiving illness/disability benefits subject to work-related requirements. It operated on a radical payment-by-results (PbR) model. Private service providers were required to absorb risk and invest capital upfront, receiving payments in exchange for job starts and sustained job outcomes at 4, 13 and 26-week intervals. For frontline WP 'welfare-to-work advisors', PbR most often translated into stringent managerial pressure to achieve initial job starts and sustained job outcomes. While WP providers were afforded some flexibility to design services, welfare-to-work advisors were mandated to implement a standard case management approach, using digital software to schedule high-volume caseloads (typically around '120') for regular work-focused interviews and filing sanction referrals for any breach of requirements. This placed a wider range of claimants under more intense pressure to demonstrate work-related activity and find work, or else face (threat of) the enhanced sanctioning regime.
Meanwhile, in 2011, new 'off-benefit flow' targets were introduced in JCP offices. For 'work coaches', this meant that successful outcomes were achieved when claimants ended their claim irrespective of whether they entered employment. This stimulated new street-level adaptations, encouraging some staff to develop innovative methods of securing claim closures . There were also greater informal pressures placed on work coaches to make sanction referrals (Webster, 2016). A PCS union (2014) survey of JCP staff found that 23% of respondents had an explicit target to make sanction referrals. Formal off-benefit flow targets, informal sanctioning pressures, and placing a wider range of claimants under more intense work-related pressures via WP, were instrumental factors in a huge surge in sanctioning rates between 2010-2013. Sanctions reached unprecedented heights while the claimant count stagnated and fell: Loopstra et al. (2015) found that those who were sanctioned and subsequently ceased to claim benefits were far more likely to enter unknown destinations than to find work. This evidence cumulatively suggests there was a significant change to the operational logic driving employment service delivery in the first years of the 2010s. The objective was not only to enforce compliance and expand work participation, but also to secure claim closures irrespective of claimant destination.
Beyond 2015, there has been continued effort to 'make work pay' by reducing the take-home value of cash-benefits via another benefit freeze, another benefit cap, and the two-child limit. Although there has also been a softening of some compliancefocused policies. Sanctioning rates have fallen significantly since 2013. Between 2013 and 2016, Webster (2016: 2) identifies a '5.2 percentage point' fall in the proportionate rate of monthly sanctions to JSA claimants; attributing this primarily to a 'decision by ministers to ease off' on sanction referrals and declining WP referrals. The maximum sanctioning period was reverted back to 26-weeks in 2019. WP has ended and investment in private providers has been downsized (now 'Work and Health Programme'), though provider payments are still indexed to job outcomes. More recently, ministers have again looked towards expanding compliance-focused policies as one way of dealing with national labour shortages.

Methodology
This article uses data collected from 13 frontline employment service staff. Ten interviews were conducted in 2019 as part of a research project examining contextualised agency in public service interactions between frontline staff and young, unemployed men. A further three interviews were carried out in 2021 as part of a different project (still ongoing) and supplement the initial interviews. Accessing employment service staff has presented significant challenges across both projects, which partly reflects the DWP's increasing sensitivity to external scrutiny . To overcome this, contact was established with retired civil servants and adjacent service providers where former staff had migrated into new roles. Some participants were then snowballed through initial contacts. The original sample (n = 10) comprised one JCP manager; three JCP work coaches; one JCP staff and later welfare-to-work advisor; four welfare-to-work advisors (WP, one also had Work and Health Programme experience); and one decisionmaker. The additional sample (n = 3) comprises one JCP work coach and two G4S security officers (working in JCP offices). All participants reflected on experiences of working during the post-2010 period in present focus. A majority (n = 9) were not active at the time of interview and were reflecting entirely on former experience. One work coach in the original sample and the additional sample (n = 3) were active at the time of interview. This was significant for the work coaches, who were either prompted or independently reflected on frontline experiences in the 2010-15 period and reported noteworthy changes to employment service delivery in more recent years. It was less significant for security officers, who indicated that the core aspects of their role (maintaining order on the shopfloor) had remained constant. Nevertheless, as all participants were recalling experiences from previous years, memory fallibility could be a limitation. Although, shared experiences of 'detection' and 'correction' practices, as well as various other street-level adaptations to policy and managerial pressures, were in evidence across the sample. This was in spite of respondents occupying a range of positions in the employment service and no two working in the same office.
Interview schedules revolved around the general structure of everyday work, performance targets, inter-personal work with claimants, and general attitudes towards the role. Transcripts were re-read and re-coded for the production of this article. A 'theoretical thematic analysis' was selected to provide a detailed analysis of an aspect of the data set (Braun and Clarke, 2006). A set of codes were developed to identify specific patterns around practices of 'detection' and 'correction' in response to perceived exhibitions of claimant 'non-compliance'. A further set of codes was developed from existing street-level literature to identify key drivers of frontline practice (e.g., 'managerial pressures', 'administrative pressures'; 'moral', 'normative' and 'calculative' 'discretion').
All specific details pertaining to name, place, gender, age, ethnicity, job tenure and so forth are withheld or pseudonymised. Ethical approval for both research projects was received by Sheffield Hallam University Research Ethics Committee.
5. The workfare project in street-level practice Claimant non-compliance can take a variety of forms. It can occur directly, in public service interactions-in discreet, non-confrontational forms (Finn, 2021;Whelan, 2022: 87-104) or in overt, hostile forms (Fletcher and Flint, 2018). Non-compliance can occur indirectly, outside of public service interactions-in refusals of undesirable work or engagement in illicit/informal activity (Casey, 2021;Redman, 2021). Though uncommon, non-compliance can also signify the presence of more deeply held anti-work values (Frayne, 2015). Respondents indicated that frontline staff play an important role in detecting and correcting these variations of claimant noncompliance.

Detection
Detecting non-compliance took several forms. One form was to communicate with a small network of actors who, inadvertently or otherwise, were appropriately positioned to observe claimant behaviour. When claimants are unable to demonstrate compliance to work-related duties and instructions, this can be overlooked if they present a reason sufficiently explaining why they were unable to comply. Some welfare-to-work advisors conferred with JCP staff to verify (or falsify) alibis, to determine whether claimants were paying lip service: 'I always stress the importance of keeping, maintaining dialogue and relationships with the work coaches because we knew that they were playing us off against each other : : : have they been attending? Have they been engaging and embracing everything that we're trying to do up there? Not always' (Worker four; reference to Work and Health Programme) Other JCP and WP staff indicated that when they had scheduled a job interview for a claimant they would be in contact with employers. This was sometimes done to assess interview performance, but it was also done to probe for anti-work behaviour/ values (cf. O'Sullivan et al., 2021: 161): 'I knew he didn't want a job. So I sent them to see this [employer]' (Manager one; JCP higher executive officer). Although, one worker suggested that this had not been practiced in their office for many years due to complaints from employers who bemoaned 'fielding lots of phone calls from work coaches checking up on people who should have been for interviews' (Worker ten; active JCP work coach).
Worker two (welfare-to-work advisor; WP) repeatedly likened aspects of their role to being like a 'detective'. This seemed a fitting descriptor of a range of other practices deployed by frontline staff. Workers in two WP provider offices were encouraged to practice informal digitalised surveillance to monitor claimant behaviours and document any change in work-related circumstances. Worker two suggested that digital surveillance was done to find out whether claimants were availing of alternative income streams: '[management would] say "go on Facebook and have a look". "People are posting on there that they've been doing a bit of cash in hand" or you know "they're selling this on Ebay or whatever and advertising it on Facebook." They'd make you do anythingsome things you maybe shouldn'tyou know when you're looking on their Facebook and trying to find evidence that they're getting money from other means.' Gathering intelligence on non-compliance was important for some WP managers. Not only would this enable them to trigger mandatory sanctioning and fraud referral apparatus, but information on non-compliance could also facilitate more efficient allocation of time and resources in the direction of individuals more likely to comply and result in successful job outcomes: 'if they're claiming money and claiming benefit, then we don't wanna deal with them because we're not gonna get 'em a job' (worker two).
The pressure of financial targets, mediated through achieving job outcomes, was a constant for WP staff. This pressure could be amplified when claimants disengaged from WP by finding work of their own accord and either refused or forgot to declare a job outcome. This led Worker nine (welfare-to-work advisor; WP) to use 'fake social media profiles' to 'track "disengaged" clients' and find out whether they were working, so that those who had entered formal employment could be claimed and monetised as successful job outcomes. Approximately once a week, management in Worker two's office would ask staff to go door knocking in the communities of disengaged claimants: 'whether its your customer or not, you had to go with someone or drive to people's houses and knock on the door and say "where are ya?" : : : you had to get signatures and things like that if people had been in work or get them to sign something to claim the outcomes of money : : : Or, if they'd gone in work and you couldn't get hold of them, you had to knock on doors to get the information of where they were working, who was in charge and how many hours they were doing' The pressure of financial targets, perhaps combined with the service design flexibility afforded to WP providers, produced a range of surveillance practices in contracted employment services.
However, financial targets were not the only driver of frontline 'detective' work. JCP staff noted how, at certain points during their tenure, they had been subjected to informal sanctioning targets/expectations and formal off-benefit flow targets. When queried how they personally responded to sanctioning targets, manager one would train staff to become competent anti-work detectives. They would send staff to learn from decision-makers so they were aware of what to look for and how to carry out appropriate record keeping of claimant behaviour to achieve sanctions: 'training comes into it. If people weren't looking for work, you were looking at training your staff to spot it. They never made the decision, 'cause it went to independent decision-makers in a separate office. But you had to be able to report it. You had to be able to write up the findings of whatever your interview was, or whatever questions you'd asked. You'd do this impartially, but also know what the decision-makers were looking for to get sanctions.' It is known that some frontline staff will defy managerial pressure to sanction (Casey, 2019(Casey, : 1026. Worker three (active JCP work coach) resisted sanctioning on moral and normative grounds. They identified a 'spectrum' of attitudes and behaviour among their work coach colleagues, varying from 'policing' to 'coaching' dispositions (respondent terminology; cf. Zacka, 2017). Worker three said that some colleagues thought 'claimants [were] lazy, dishonest, not trying to get work and the stick was important', possessing tougher 'policing' dispositions towards the role. Conversely, Worker three placed greater emphasis on providing help and support to claimants, possessing softer 'coaching' dispositions. Worker three felt sanctions were antithetical to 'coaching' and would resist overt pressure to make sanction referrals in team performance meetings: '[the managers] line was it's inconceivable that out of all the people you've seen over the last week, that they were all trying very hard to find work'. Nonetheless, it is likely that managerial pressure would make others hyper-vigilant for non-compliance to make referrals. Manager one went on to say about their staff: 'they would love to prove that they could do their jobs, so sign offs, how many sign-offs, how many dis-allowances, or how many sanctions that were imposed'.
Off-benefit flow targets could also focus frontline attention towards probing for non-compliance: 'in order to achieve the off-flow targets we started using the fraud section more often : : : If, for example, a lone parent came in for a work-focused interview I've known colleagues refer them up to fraud because they've got a love bite on their neck which may indicate that they're living with their partner, which means they're not a lone parent so the claim would close. People look at what rings they've got on their fingers, have they got a ring, an engagement ring on their wedding ring finger? They'd look at the jewellery people are wearing, are they turned out well? That may indicate that they've got undeclared funds coming into the household. Their appearance would be monitored, looking for anything that might indicate that they're not actually a lone parent' (Worker ten) While sanctioning and off-benefit flow targets drove 'detective' work, this was not always done to ensure compliance. During the (post-2010) period reflected on in worker ten's excerpt, detection practices were also carried out to secure claim closures in JCP offices: 'because we switched to an off-flow target rather than an in to work target we started looking at ways to get claims closed down'. This is an important distinction because the outcome was on securing off-flows and not on ensuring compliance, it mattered less whether the evidence gathered was indicative of noncompliance: 'there was definitely an increase in the use of the fraud department and advisors were encouraged to make a fraud referral on the flimsiest of evidence' (worker ten). Performance metrics shifted moral concerns away from evaluating the veracity of evidence and the potentially harmful consequences of their actions, towards finding the most effective means of achieving the target. Two active respondents indicated that off-benefit flow targets are no longer in widespread operation. Nevertheless, this shows how detection practices, like other street-level practices, are acutely sensitive to top-down managerial pressures that filter vertically through the chain of command. Their precise form and frequency can fluctuate according to the specific priorities and orders laid out by ministers.
'Monitoring appearances' and behaviours, as highlighted above, was common practice in the 'detective' work respondents did during face-to-face interactions. Some scrutinised the condition of claimants' hands. Dirty hands could be indicative of informal work: 'if they're not working but why are they covered in oil or dirty finger nails?' (worker two). Staff would pay attention to the availability and attendance patterns of their caseloads: 'we would put them on courses that last two weeks or three weeks : : : and they was like "well I can't do that" : : : you could see them trying to think, like what can I say? And it's like, well, we kind of know that you're probably doing a bit of cash in hand' (Worker eight; welfare-to-work advisor; WP) Staff were also attentive to clothing: 'they'd be turning up in new clothes, brand new latest fashionable things. Well I can't afford that and I'm working, so how are you?' (Worker eight). It is possible that new clothes indicate engagement in illicit activity. However, it is not uncommon for individuals enduring poverty, who are excluded from mainstream consumption norms, to occasionally prioritise purchase of luxury items with cash-benefits (McKenzie, 2015: 109). Negligible consumptive pleasures are one way claimants locate comfort, feel valued (among peer groups) and 'get by' when existing on a low-income (McKenzie, 2015: 115; see https://covidrealities.org/ experiences/themes). Furthermore, numerous studies have shown how frontline staff harbour stigmatising perceptions of claimant appearances and behaviours, which can either be cultivated and legitimised through the chain of command (Kaufman, 2021: 177;Redman and Fletcher, 2022) or based on 'perceptions and beliefs that staff themselves bring to their jobs' (Wright, 2003: 215). It is therefore also possible that suspicions aroused from observations made on claimant appearances were unsubstantiated and more accurately reflected the persistence of (classed) prejudices on the frontline. Unsubstantiated or otherwise, the arousal of suspicions could encourage staff to adopt a direct approach to detective work by subjecting individuals to periodic interrogation: 'I always wanted to get to the bottom of why don't you want to work? What is it? What's your reason behind it? If you have got another income stream, then stop committing fraud and let's spend the money on people who need it. I'd try and just be very direct in that way and find out.' (Worker six; welfare-to-work advisor; WP) Welfare-to-work advisors were explicitly instructed to classify claimants according to a traffic light scheme (green = job ready; amber = some help/pressure; red = multiple barriers to work; Kaufman, 2020: 210). Consequently, gathering information about the appearances and behaviours of caseloads-and what this might reveal about their willingness to adhere to work-related instructions/activities-featured in reaching calculative decisions about whether and how to intensify compliance-focused policies/practices. For example, worker two reflected on the dilemma of claimants perceived as both work-resistant and employable: 'people that maybe were red or amber, that maybe didn't want to work : : : that clearly could work. You'd book 'em on courses : : : it was a test for them'.
Alongside direct and indirect 'detective' work during face-to-face interactions, some frontline staff would police claimant behaviour outside the office, in the wider community, beyond official working hours. Across the sample, it was not uncommon for respondents to recall instances where they or a colleague had witnessed a claimant working.
'I had to tell someone in my [anonymised location] once. I found out they were working and claiming benefit because she told me. I don't think she'd really thought through what she was saying. So I had to say to her, "If you don't stop claiming or declare it. I can no longer speak to you. I can't be friends with you."' (manager one)

Correction
Where non-compliance was detected, it was not always guaranteed that frontline workers would take further action. A range of 'moral, normative and calculative' rationales were again in operation here. For example, staff in worker three's office would make calculative judgements about completing fraud referrals, due to the low likelihood of successful outcome: 'even my colleagues who've got negative views about claimants don't generally bother making fraud referrals very much, because they don't think anything's going to happen'. Workers may also make moral and normative judgements about the magnitude of non-compliance, allowing some to take 'small liberties' (Dubois, 2010: 151): 'If I don't think their job search is good enough, I might just ignore it' (worker three).
Nevertheless, in cases of more serious non-compliance the general feeling across the interviews was that, where practicably possible, action should be taken: 'I think most people felt, if they thought their customer was working [they would] would want them off' (manager one). This included those with self-identified 'softer' dispositions, such as worker three: 'I was responsible for dealing with communication from their WP and other providers who would communicate with us from time to time about claimants. So there was a couple of occasions where I got information sent to me by these providers which indicated some fraud : : : I just completed a fraud referral' There were a range of policies and practices respondents could and had previously deployed to correct claimant non-compliance. This could be a matter of processing sanctioning and fraud referrals, or articulating their power/authority to set punitive apparatus in motion and using it as a threat. Worker seven would do this by 'depersonalising' (cf. Wright, 2003: 240) the process and emphasising that they, much like their caseloads, were powerless to the bureaucratic operation: 'I didn't let them get away with anything. It wouldn't be right. It wouldn't be fair to let them get away with it, but I used to set the stall out and say, "This is not my choice. It's my job." : : : "you do realise that it's not an entitlement, it's only an entitlement if you fulfil the criteria." That was always laid out in very clinical terms : : : then if it went to a decision maker, it wasn't my fault, it was their fault' (Worker seven; JCP work coach) The spatially separated division of labour that exists within the sanctioning process between sanction referrals and independent decision-making can disperse and therefore minimise moral responsibility for any harmful consequences triggered by adverse sanction outcomes. Emphasising powerlessness to the independent decision-making process enabled worker seven to enforce compliance while also establishing a measure of cognitive dissonance, by relinquishing moral responsibility, when executing aspects of the role known to inflict harm. Doing this allowed worker seven to placate a tension that had emerged between their personal conviction that claimants should demonstrate compliance and their incongruous discomfort at the prospect of utilising some compliance-focused policy tools.
Where staff felt like individuals were not demonstrating sufficient compliance to their work-related obligations, they could improve compliance by undertaking emotion work to re-adjust the general mood and feel of face-to-face interactions. Managerial pressure would occasionally encourage Worker three to alter their approach: 'I've got my manager sitting there and they don't come up with a good story about their job search : : : so I give them a bollocking : : : they [claimant] were exposing me as a softy : : : they were putting me at risk' This could have the desired effect: 'they can pick up when you're pissed off with them, and they will generally respond by trying a bit harder' (Worker three).
Alternatively, if a claimant was persistently late to attend appointments, then JCP staff could use their discretionary powers to impose an informal benefit delay by rescheduling their signing on date. Frontline staff can face intense pressure to schedule high volumes of face-to-face appointments. Consequently, late attendees can disturb the smooth flow of bureaucratic routine and frustrate workers already under pressure to meticulously manage high-volume caseloads. This could encourage retribution: 'if it's an individual that's always late sometimes you have to cause them a little bit of inconvenience in order to get them back on track. Because there's so much pressure on our diaries if somebody is 5 or 10 minutes late it can completely screw your day up and it messes all your diary schedule up. If somebody's doing that on a regular basis then every now and again you'll say "you're not signing today, come back in the morning" which then delays their payment by a day. So that's the punishment if you like for not coming to your appointment on time.' (worker ten) This case shows how formal sanctions are not the only tool that some frontline staff have at their disposal (cf. Soss et al., 2011: 227). Imposed benefit delays can also be used as a negative incentive to encourage the desired level of compliance. It also shows how administrative pressure can influence compliance-focused practices (cf. Brodkin, 2013b: 152-154).
In cases where individuals were suspected to be working informally but more evidence was needed to legitimate a fraud referral, a common response would be to use 'More Frequent Attendance' or 'MFA' (JCP staff terminology): 'Using more frequent attendance was just usual course of action that everyone did. There'd be certain triggers on a claim in terms of somebody's behaviour, missing appointments, coming in late, phoning up to rearrange, where you'd get an idea that there's a reason behind this. They came in work boots the other week, that type of thing that might trigger a referral.' (worker ten) Kaufman (2020) unearthed similar practice in his research, where frontline staff could intensify claimants' personal experiences of workfare by increasing their number of appointments. However, whereas Kaufman frames this as a tool to 'intensify' compliance, the present research found that MFA is also practiced to act on aroused suspicions, to smoke-out and disrupt non-complianceor to 'shake the tree' (cf. O'Sullivan et al., 2021: 157). This was done across JCP and WP offices.

The inclusion of new staff groups in compliance enforcement
Since the 1990s, British Jobcentres have been transformed by the deployment of private security officers to control shopfloor access and monitor claimant behaviour. It is in this context that claimants have reported Jobcentre attendance as an increasingly hostile and intimidating experience (Wright et al., 2020). Consequently, the authors contend that it is important to ascertain the views and experiences of security officers as they have become a significant and relatively new actor in employment service delivery.
To this end, it can be seen how security officers too play an active role in detecting and correcting alternative forms of claimant non-compliance. Situated on the floor(s) and entrances/exits of JCP offices, security officers are tasked with marshalling the office space. This includes maintaining vigilance for signs of agitation and aggression in waiting rooms or during work coachclaimant interactions and, in the event of a disturbance, repressing claimant antagonisms: 'if they come in raging we're on the job as soon as they come in' (Worker eleven; active G4S security officer). Hostile claimants are recorded, their behaviour is then reviewed by senior department officials and they are deemed 'potentially violent' or placed under more intensive 'monitored behaviour' during future appointments. In the case of the former, 'potentially violent' claimants can be required to attend appointments in a separate screened room, or they may be issued a fixed-term ban from the premises. In the case of the latter, security officers may be required to escort the individual to their work coach and stand in close proximity during interviews to encourage deference and deter hostile interaction: 'you have to have somebody watching them all the time' (Worker twelve; active G4S security officer).

Conclusion: Competing perspectives on the workfare project
This article has shown how frontline employment service staff enforce behavioural compliance by deploying a range of practices to detect and correct variations of noncompliance (e.g., from social media surveillance and scrutinising claimant appearances/behaviour, to imposed benefit delays and 'more frequent attendance'). It has also been shown how these practices are not homogenous or evenly implemented across the frontline. Rather, there are one or a combination of drivers-pertaining to the compliance-focused policy framework, managerial and administrative pressures, workers' discretionary power and/or stigma-laden perceptions of claimant behaviours and appearances-which converge to produce inconsistently implemented detection and correction practices.
These drivers of inconsistent frontline practices have already been unearthed and robustly theorised in existing street-level research (Lipsky, 2010). Nevertheless, the present article also explores a feature of employment service delivery that is largely ignored in existing street-level research. Although many policies and practices central to enforcing behavioural compliance have already been extensively discussed, corresponding street-level analyses exploring how frontline staff deal with claimant non-compliance are comparatively thin on the ground. Dealing with non-compliance, we consequently contend, is also a central but neglected dimension to enforcing compliance in the daily delivery of employment services. Practices of detection and correction are a neglected dimension to what Kaufman (2020: 216) has described as the 'disciplinary axis of discretion' currently operational in employment services across the world. By shedding new empirical light on these practices, the present article deepens our understanding of how the workfare project is constituted in daily, situated practice.
Claimant non-compliance is a politically contested subject and this has implications for how these findings may be interpreted. Those (political and intellectual actors) who have played an instrumental role in unfolding the workfare project have often maintained that exhibitions of claimant non-compliance signify the persistent presence of a behaviourally dysfunctional 'underclass' in need of workfare policy solutions (Mead, 1986;Fletcher and Redman, 2022). From this perspective, detection and correction practices will likely appear as coherent street-level manifestations of a reform agenda seeking to tackle poverty by ensuring poor, unemployed populations comply with mainstream behavioural standards.
However, recent decades have seen the growth of a critical scholarship contesting the dominant orthodoxy that the workfare project is designed to tackle poverty. This scholarship claims that workfare is instead a project of governing and perpetuating the symptoms of 'advanced marginality'-a 'new regime' of post-Fordist poverty wrought by the simultaneous decline of stable, middle-income employment, the shrinking of the social welfare state and the augmented shame of occupying residence in the decaying nether regions of urban social space (Wacquant, 2008). The workfare project is said to do this by bullying poor people into permanently precarious poverty jobs while criminalising (stigmatising, disciplining, punishing) the struggles they often undertake to make ends meet under daily duress of advanced marginality. This perspective is partially supported by quantitative research reporting that in-work, deep and relative poverty have all risen in tandem with the unfolding workfare project (Edmiston, 2022).
It is also supported by a body of qualitative research, which has examined variations of non-compliance from the perspectives of people engaged in such behaviour. This has consistently found that variations of non-compliance are frequently articulated by individuals as functional and necessary efforts taken to ensure survival and minimise personal disadvantage (MacDonald, 1994). This often takes shape in tacit, non-confrontational struggles within and against a milieu of precarious, low paid job opportunities in the formal economy, inadequate cash-benefit support, and (narrow and broad) workfare policies perceived by marginal individuals as failing to reflect their personal needs and interests (Redman, 2021). This is additionally supported by research which has shown how workfare policies can produce contradictory effects, pushing marginal individuals further away from employment service provisions and towards (re-)engaging with the informal economy or criminal endeavours (Fletcher and Flint, 2018).
As such, viewpoints on the workfare project ostensibly look very different depending on the (dominant or marginal) perspective you look from (cf. Fletcher and Redman, 2022). The latter perspective suggests that detection and correction practices may be less concerned with correcting the behavioural dysfunctions of an 'underclass' and far more concerned with criminalising daily efforts to minimise the material and symbolic miseries of existing on the margins in the post-Fordist present. this begs the question: what privileges the assumptions embedded in his proposed solutions over those found in the solutions of the 'liberal' reformers?