Haldenmoor Coaches Limited operates scheduled bus services under contract to two local transport authorities alongside a private coach hire business, running 96 vehicles from three depots in the North West and employing 420 people, of whom 184 are drivers. The author is Operations Manager at the largest depot. Driver turnover reached 31 per cent in the last financial year against a sector benchmark of around 20 per cent, and the resulting reliance on overtime and agency cover has produced both cost pressure and service cancellations. Organisational detail is illustrative and anonymised.
Table of Contents
Introduction
This report presents a management project undertaken to establish why driver turnover at Haldenmoor Coaches has risen to 31 per cent and what the business should do about it. The project was conducted over fourteen weeks and used a mixed methods approach combining an all-driver survey, semi-structured interviews with leavers and current drivers, and analysis of existing exit and rota data.
A note on what follows. The research described here was designed and conducted in this organisation, and the findings are specific to it. Where the report explains method rather than findings, the reasoning is transferable; the conclusions are not.
Learning Outcome 1: Know how to plan a management project
AC 1.1 Develop the aim, objectives and scope of the proposed management project
The business needs. Driver turnover of 31 per cent means the depot replaces roughly one driver in three each year. Each replacement costs an estimated £4,800 in recruitment, licence acquisition support, training and supervised driving before the driver is productive. Beyond cost, the resulting shortfall drives rest-day working and, on 47 occasions last year, service cancellations that attract contractual penalties from the transport authority.
Aim. To identify the principal causes of driver turnover at Haldenmoor Coaches and to recommend evidenced interventions capable of reducing turnover to below 22 per cent within twelve months.
Objectives. Five, each written to be assessable:
- To establish the pattern of driver turnover across depots, tenure bands and shift types using existing organisational data.
- To identify the reasons drivers give for leaving, distinguishing stated from underlying reasons where possible.
- To identify the factors current drivers report as most likely to prompt them to leave.
- To compare the findings against published evidence on retention in transport and comparable operational sectors.
- To develop and cost recommendations, prioritised by expected effect and feasibility.
Developing the objectives. Each was tested against whether it could be completed within the time available and whether achieving it would contribute to the aim. An earlier draft objective concerning national driver shortage was removed, since establishing it would consume effort and would not tell Haldenmoor anything it could act on.
Scope. The project covers drivers employed at all three depots, both scheduled service and private hire, over the preceding 24 months. It examines pay, rota design, management relationship, working conditions, training and progression.
Exclusions, stated explicitly. The project does not cover engineering or office staff, does not extend to agency drivers supplied through third parties, does not evaluate the national labour market, and does not consider vehicle specification. Recruitment practice is examined only where it bears on subsequent retention. Stating exclusions matters because a project that expands to cover everything adjacent to the problem finishes late and concludes little.
Testing the aim against the SMART criteria. The aim is specific in naming turnover at a defined organisation, measurable through a stated target of below 22 per cent, achievable on the evidence that the sector benchmark sits around 20 per cent, relevant to a documented cost and service problem, and time-bound at twelve months. Setting a target below the sector benchmark would have been unachievable and setting none would have made the project unassessable.
Constraints. Fourteen weeks, no dedicated budget beyond the author’s time and a small allowance for interview cover, and access to drivers limited by the operating pattern, since drivers are on the road for most of their shift.
AC 1.2 Specify a research approach for the proposed management project
Philosophical position. Saunders, Lewis and Thornhill (2023) describe research design as a sequence of decisions moving from philosophy through approach and strategy to methods. Specifying the position taken here, the project adopts pragmatism. The question is practical rather than theoretical, and the appropriate test of a method is whether it produces information the business can act on rather than whether it satisfies a single philosophical tradition.
Specifying why the alternatives were rejected, a purely positivist approach would treat turnover as a measurable phenomenon explicable through quantitative analysis alone, which would establish patterns without explaining them. A purely interpretivist approach would generate rich accounts from a small number of drivers without establishing how widely those accounts apply.
Reasoning approach. Specifying this element, the project is primarily inductive. It begins with organisational data and driver accounts and builds an explanation from them, rather than testing a hypothesis formed in advance. A deductive element enters at objective four, where findings are compared against published evidence.
Specifying the practical reason for inductive reasoning, management had already formed a hypothesis, namely that pay was the cause. Designing a study to test that hypothesis would have found evidence for it, because leavers offer pay as a reason whether or not it is the operative one. An inductive design allows a different explanation to emerge.
Methodological choice. A mixed methods design, sequential and explanatory in structure: quantitative data first to establish the pattern, then qualitative work to explain it, then a survey to test how widely the explanations hold.
Specifying the justification, mixed methods suit a question with both a distributional and an explanatory component. Bell, Harley and Bryman (2022) note that combining methods allows the weaknesses of each to be offset, and here the numerical data establishes scale while the interviews establish meaning.
Time horizon. Cross-sectional, examining the position at a point in time while drawing on 24 months of historical turnover data. Specifying why a longitudinal design was rejected, it would produce better evidence and would take two years, by which time the business would have absorbed a further two years of turnover cost.
Research strategy. Specifying the strategy within the approach, the project uses a single-organisation case study incorporating a survey. Saunders, Lewis and Thornhill (2023) distinguish strategy from method, and the distinction matters here because a case study permits depth in one setting while conceding that findings may not generalise to other operators. Specifying why that concession is acceptable, the business needs to know why its own drivers leave rather than why drivers leave generally.
Ethical position. The project involves employees discussing their intention to leave and their view of their managers, which carries obvious risk to participants. The approach therefore rests on voluntary participation, informed consent, confidentiality and reporting that cannot identify individuals.
AC 1.3 Specify the research methods to be used to collect data and information
Secondary data analysis. Existing organisational records covering leaver dates, tenure at leaving, depot, shift pattern, exit interview notes where completed, absence records and rota data showing rest-day working by individual. Specifying the value of this method, the data already exists, costs nothing to obtain and covers the whole population rather than a sample.
Specifying its limitations, exit interview notes were completed for only 62 of 118 leavers and are of uneven quality, and the recorded leaving reason is a single category selected by the manager conducting the interview rather than the leaver’s own account.
Semi-structured interviews. Twelve interviews of approximately 45 minutes: six with drivers who left in the preceding six months, contacted by telephone, and six with current drivers selected to span tenure bands and both depots.
Specifying why semi-structured rather than structured, the format allows a prepared set of topics while permitting the interviewer to follow what the participant raises. Specifying the risk this creates, the interviewer is the Operations Manager, and drivers may moderate what they say to someone in authority. Specifying the mitigation, leaver interviews were conducted first on the reasoning that former employees have less to lose, and current driver interviews used the themes leavers raised rather than introducing them fresh.
Survey. A structured questionnaire issued to all 184 drivers, using closed questions with rating scales on the factors identified through interview, plus two open questions. Specifying the design decision, the survey was built after the interviews rather than before, so that it tested explanations grounded in driver accounts rather than in management assumptions.
Specifying distribution, paper copies were distributed at sign-on with a sealed return box at each depot, because drivers do not routinely access company email and an online survey would have produced a response bias towards office-based staff.
Sampling. The survey was a census of all drivers rather than a sample. Interview participants were selected purposely to span tenure, depot and shift type rather than randomly, since the aim was range of experience rather than statistical representativeness.
Ethical procedures. Participation voluntary and stated as such; consent obtained in writing for interviews and implied by return for the survey; no individual identifiable in any output; interview recordings held securely and destroyed after transcription; survey responses anonymous; participants informed how findings would be used. Specifying the position on data protection, personal data was processed under the Data Protection Act 2018 with a stated purpose and retention period.
Document analysis. A fourth method applied to existing internal material: rota policy, the induction programme, training records and the previous two employee survey reports. Specifying its contribution, this establishes what the organisation intends to happen, which can then be compared against what drivers describe actually happening. Specifying a finding this produced, the induction programme on paper runs to eight weeks, and no driver interviewed described anything beyond the first fortnight.
Reliability and validity considerations. Specifying how these were addressed, the survey questions were piloted with four drivers to test comprehension, the interview themes were derived from the data rather than assumed, and triangulation across three sources allows a finding supported by all three to be held with more confidence than one appearing in only one.
AC 1.4 Analyse the role of stakeholders involved in the management project
Drivers. Analysing the primary stakeholder, drivers are simultaneously the subject of the research, the source of its data and the people its recommendations will affect. Their role is participatory rather than passive, and the project depends entirely on their willingness to answer honestly. Analysing what determines that willingness, drivers who believe the exercise is genuine participate; those who believe it is a formality do not, and a 31 per cent turnover rate suggests a population with reason for scepticism.
Depot managers and supervisors. Analysing this group, they provided access, rota data and context, and they are also implicated in the findings, since management relationships emerged as a theme. Their role is therefore both enabling and exposed, which required careful handling: findings about management quality were reported at group level rather than by named individuals.
The Operations Director as project sponsor. Analysing the sponsor role, they authorised the project, provided access and will decide whether recommendations are implemented. Analysing the risk this creates, the sponsor had a prior view that pay was the cause, and a project producing a different answer requires the sponsor to accept evidence against their own position.
Human resources. Analysing their role, they supplied leaver data and exit interview records and hold responsibility for several likely recommendations. They also function as a check on the ethical handling of employee data.
Trade union representatives. Analysing an important stakeholder in a unionised operation, representatives were briefed before the survey was issued. Analysing why this mattered practically, a survey about working conditions issued without union awareness would have been read as management gathering evidence for a purpose it had not disclosed, and response rates would have suffered.
Transport authority clients. Analysing an external stakeholder, service cancellations arising from driver shortage attract contractual penalties and affect the authority’s own performance reporting. They are not participants in the research and they are affected by whether it succeeds.
Analysing the difficulty of researching one’s own department. Buchanan and Huczynski (2023) observe that insider research trades distance for access, and that participants adjust what they say according to the researcher’s position in the hierarchy. Analysing the evidence that this occurred, two current-driver interviews were noticeably guarded while leaver interviews were not, which is precisely the pattern that observation predicts and a limitation the findings must be read against.
Analysing stakeholder influence collectively. These groups hold different kinds of influence. Drivers determine whether the data exists; the sponsor determines whether findings are acted on; representatives determine whether the process is trusted. Whittington et al. (2023) observe that stakeholder management fails where influence is assumed to be uniform, and the practical consequence here was that each group required a different kind of engagement rather than the same briefing.
AC 1.5 Examine the factors that need to be considered when developing a plan for the management project
Time and the sequencing of dependent activities. Examining the principal planning constraint, the mixed methods design is sequential, so the survey cannot be written until interviews are analysed, and interviews cannot begin until secondary data has identified who to approach. Each stage depends on its predecessor, which removes the option of running activities in parallel to recover time.
Access and operational reality. Examining a factor specific to this setting, drivers are on the road for most of their shift and available only at sign-on and sign-off. Examining the planning consequence, data collection had to be built around the operating pattern rather than around the researcher’s convenience, which extended the survey window to three weeks to cover all rota cycles.
Resource and cost. Examining what the project required, the principal cost is time rather than money: the author’s own time, interview cover for participating drivers, and printing. Examining a hidden cost, twelve driver interviews of 45 minutes each require cover arrangements that themselves consume supervisory time.
Researcher position and objectivity. Examining a factor that must be planned for rather than wished away, the author is the Operations Manager researching their own department. This carries advantages in access and context and creates real risks of participant reticence and researcher bias. Examining the mitigations planned, these included interviewing leavers first, using an independent colleague to code a sample of survey open responses, and stating the position openly in the report.
Ethical approval and data protection. Examining a procedural factor, the project required agreement on how employee data would be handled, what would be reported and to whom, agreed with human resources before any data was collected rather than afterwards.
Quality of existing data. Examining a factor that materially shaped the plan, exit interview completion at 53 per cent meant secondary data alone could not answer objective two, which is why interviews were necessary rather than merely desirable.
Stakeholder expectations and the risk of a predetermined answer. Examining a political factor, the sponsor’s prior view that pay was the cause created a risk that a different finding would be resisted. Examining the planning response, the method was agreed with the sponsor in advance, including the possibility that pay would not emerge as the principal factor, so that the finding could not later be attributed to a flawed design.
Reporting and dissemination. Examining a factor often left out of project planning, how findings will be communicated must be decided before data is collected, because participants will ask. The plan commits to a summary for all drivers, a full report to the sponsor and a briefing to representatives. Examining why this matters beyond courtesy, an organisation that surveys its workforce and never reports back has taught that workforce not to respond next time, which is a cost borne by whoever runs the following project.
Researcher capability. Examining a factor rarely planned for, the project required competence in survey design, thematic coding and basic statistical description. Most managers reach their role without formal grounding in research method (Chartered Management Institute, 2023), and the honest position taken at planning was that the survey and descriptive analysis were within reach while the thematic coding required guided reading and independent review of a sample to be defensible.
Contingency. Examining what could go wrong, low survey response was the principal risk. Examining the contingency planned, a second distribution round and a stated intention to report findings back to drivers regardless of outcome, since visible follow-through is what makes a second attempt credible.
Learning Outcome 2: Be able to conduct a management project
AC 2.1 Conduct research to deliver the management project
Secondary data collection. Leaver records for 118 drivers over 24 months were extracted and coded by depot, tenure at leaving, shift pattern, recorded reason and rest-day working in the three months before leaving. Rota data was matched to individuals to establish actual hours worked rather than contracted hours.
Interviews. Twelve semi-structured interviews were completed against a target of twelve. Six leavers were contacted from a list of the most recent thirty; eleven were approached and six agreed, a response rate consistent with expectations for former employees. Six current drivers were interviewed at the two larger depots, spanning tenure from four months to nineteen years. Interviews were recorded with consent, transcribed and coded.
Survey. Distributed to all 184 drivers over a three-week window covering all rota cycles. 121 usable responses were returned, a response rate of 66 per cent, which is high for an employee survey and reflects both the paper distribution method and union endorsement.
What went as planned and what did not. Reporting honestly on the conduct of the research, three departures from the plan occurred. Leaver interviews took longer to arrange than expected, delaying the survey by nine days. Two current-driver interviews produced noticeably guarded responses, which is consistent with the researcher-position risk identified in planning. And the rota data proved less complete at the smallest depot, where rest-day working had been recorded inconsistently, which limited that part of the analysis to the two larger depots.
Response bias assessment. Assessing whether the 121 respondents represent the 184 drivers, response rates were compared across depots and tenure bands. Depot two returned 61 per cent against 71 per cent at depot one, and drivers with under twelve months’ service were slightly under-represented at 58 per cent. Reporting this honestly, the under-representation of short-tenure drivers matters because they are the group the findings most concern, and it means the first-year findings rest more heavily on interview evidence than on survey evidence.
Data handling. Interview recordings were destroyed after transcription. Survey responses were anonymous and aggregated. No individual is identifiable in this report.
AC 2.2 Analyse and interpret research findings using relevant tools and techniques
Descriptive analysis of secondary data. Turnover was analysed by the tenure band, which produced the project’s first substantive finding.
| Tenure at leaving | Leavers | Proportion of all leavers |
|---|---|---|
| Under 6 months | 41 | 35% |
| 6 to 12 months | 26 | 22% |
| 1 to 3 years | 22 | 19% |
| Over 3 years | 29 | 25% |
Analysing this distribution, 57 per cent of leavers depart within twelve months. Interpreting the finding, this is not a general retention problem distributed evenly across the workforce; it is concentrated in the first year, which points towards recruitment, induction and early experience rather than towards long-run pay competitiveness.
Cross-tabulation of turnover against rest-day working. Drivers who worked more than two rest days per month in the quarter before leaving accounted for a disproportionate share of leavers among those with over twelve months’ service. Interpreting this with appropriate caution, the relationship is associative rather than causal, and the direction is not established: excessive rest-day working may prompt leaving, or drivers about to leave may take more overtime to maximise earnings before departure.
Thematic analysis of interview data. Interview transcripts were coded following the approach set out by Braun and Clarke (2022), moving from familiarisation through initial coding to the generation and review of themes. Four themes were developed:
Rota predictability. Raised in ten of twelve interviews. Drivers described difficulty planning life outside work because rotas were published with limited notice and changed frequently.
Early experience and supervision. Raised in eight interviews, exclusively by shorter-tenure drivers and leavers. Accounts described being placed on difficult routes early, limited support after training ended, and uncertainty about who to approach with a problem.
Relationship with the immediate supervisor. Raised in nine interviews, with sharply divided accounts between depots.
Pay. Raised in seven interviews, and notably raised as a comparison against local competitors rather than as an absolute complaint. Two leavers who cited pay in their exit interview described other factors as decisive when interviewed.
Survey analysis. Responses were analysed using descriptive statistics, with mean scores and distributions reported for each factor and results cross-tabulated by tenure band and depot.
| Factor | Mean importance (1 to 5) | Percentage rating 4 or 5 |
|---|---|---|
| Rota predictability | 4.4 | 81% |
| Relationship with supervisor | 4.1 | 74% |
| Pay level | 4.0 | 71% |
| Early support after training | 3.8 | 63% |
| Vehicle condition | 3.1 | 41% |
| Progression opportunity | 2.6 | 29% |
Analysing distribution rather than mean alone, supervisor relationship showed the widest spread of any factor, with depot two scoring materially below depots one and three. Interpreting this, an organisation-level mean of 4.1 conceals a depot-level problem that the aggregate figure would have hidden entirely.
Triangulation. Interpreting the three sources together, rota predictability appears in all three: it is the highest-rated survey factor, the most frequently raised interview theme, and consistent with the rest-day working pattern in the secondary data. Pay appears strongly in the survey and exit records and weakly in interviews once probed, which is the classic pattern of a socially acceptable stated reason concealing a different operative one.
AC 2.3 Use research findings to analyse options to meet the project aim
Option one: increase driver pay to match the highest local competitor. Estimated cost £310,000 annually across 184 drivers.
Analysing this option against the evidence, pay rated third in the survey and weakened under interview probing. Analysing the likely effect, a pay increase would probably reduce turnover somewhat, since pay is not irrelevant, and it would not address the first-year concentration that accounts for 57 per cent of leavers, because new drivers are leaving before pay comparison becomes their principal frame of reference. Analysing cost-effectiveness, this is the most expensive option and the one least supported by the findings, and it is the option management had assumed.
Option two: redesign rota publication and stability. Publishing rotas four weeks ahead rather than ten days, and limiting in-period changes to defined circumstances. Estimated cost £40,000 in additional planning resources plus reduced flexibility.
Analysing this option, it addresses the factor rated highest in the survey and raised most frequently in interviews, and it is supported by all three data sources. Analysing feasibility, it requires planning capability the business partly lacks and a negotiated change with representatives. Analysing the trade-off honestly, longer rota notice reduces the operator’s ability to absorb short-notice changes, which has a real operational cost the recommendation must acknowledge.
Option three: restructure the first six months of employment. A structured 26-week programme with a named mentor, protected route familiarisation, defined check-ins at weeks 2, 8 and 20, and route allocation that does not place new drivers on the most difficult work. Estimated cost £55,000 in mentor time and supervisory capacity.
Analysing this against the findings, it targets the 57 per cent of leavers who depart within twelve months, which is where the concentration lies. Analysing the evidence base, early experience rated fourth in the survey overall and considerably higher among drivers with under twelve months’ service, which is the population it addresses.
Option four: supervisory capability development at depot two. Targeted development for the six supervisors at the depot scoring materially below the others, estimated cost £18,000.
Analysing this option, it addresses a depot-specific finding that the aggregate data concealed, and it is the cheapest option under consideration. Analysing the constraint, capability development produces effects over months rather than weeks and depends on the individuals concerned being willing.
Option five: do nothing beyond current practice. Analysing the baseline, at 31 per cent turnover the annual replacement cost is approximately £176,000 in direct recruitment and training, before cancellation penalties and overtime premium. Analysing why this option belongs in the appraisal, it establishes the cost of inaction against which the others are judged.
Comparative analysis. Analysing the options against expected effect, cost and feasibility, options two, three and four together cost approximately £113,000, address the factors the evidence identifies, and target the population where turnover is concentrated. Option one costs nearly three times as much and addresses the factor the evidence ranks third.
AC 2.4 Draw conclusions and propose recommendations
Conclusion one: turnover is concentrated in the first year, not distributed across the workforce. With 57 per cent of leavers departing within twelve months, this is an early-experience problem rather than a general retention problem. Concluding what follows, interventions aimed at the whole driver population will spend most of their cost on people who were not going to leave.
Conclusion two: rota predictability is the most consistently supported factor. It is the highest-rated survey item at 81 per cent, the most frequently raised interview theme at ten of twelve, and consistent with the rest-day working pattern in the leaver data. Concluding on the strength of evidence, this is the only factor supported by all three sources.
Conclusion three: pay is a stated reason more often than an operative one. Pay appears strongly in exit records and survey responses and weakens when probed in interview, with two leavers who recorded pay in their exit interview describing other factors as decisive. Concluding carefully, this does not mean pay is irrelevant; it means pay is the reason drivers give and not, on this evidence, the reason most of them leave.
Conclusion four: supervisory quality varies materially by depot and is masked by aggregate reporting. Concluding on the wider implication, the organisation’s existing reporting would never have surfaced this, since it reports at group level.
Conclusion five: the sponsor’s initial hypothesis was not supported. Stating this plainly is necessary, since a project that quietly declines to contradict its sponsor has not been worth conducting.
Conclusion six: the organisation collects retention data it does not analyse. Exit interviews were completed for only 53 per cent of leavers, the recorded reason is a single category chosen by a manager rather than the leaver’s account, and rest-day working was recorded inconsistently at one depot. Concluding on what this means beyond the immediate project, the business held most of the information needed to identify this problem two years ago and had not looked at it.
Recommendations.
Recommendation one: publish rotas four weeks in advance and restrict in-period changes. Owner: Head of Resource Planning. Timescale: six months, phased by depot. Cost: approximately £40,000. Justification: the only factor supported by all three data sources, and the highest-rated item in a survey with a 66 per cent response rate.
Recommendation two: introduce a structured 26-week programme for new drivers. Owner: Depot Managers, with human resources support. Timescale: design in three months, first cohort at month four. Cost: approximately £55,000. Justification: targets the 57 per cent of leavers departing within twelve months.
Recommendation three: targeted supervisory development at depot two. Owner: Operations Director. Timescale: immediate, six-month programme. Cost: approximately £18,000. Justification: addresses the widest variation found in the data, at the lowest cost of any option.
Recommendation four: review pay position annually rather than increasing now. Owner: Head of Human Resources. Justification: pay is not the operative factor on this evidence, and committing £310,000 to it would consume the budget the other recommendations require.
Recommendation five: re-measure at twelve months, disaggregated by depot and tenure. Owner: the author. Justification: the aggregate figure concealed the depot variation, and repeating the measurement in the same form is what tests whether the interventions worked.
Recommendation six: improve exit data capture. Owner: Head of Human Resources. Timescale: three months. Cost: negligible. Justification: exit interviews at 53 per cent completion with a single manager-selected reason cannot support analysis, and a structured leaver questionnaire completed by the leaver would have made much of this project unnecessary.
Implementation sequencing. Recommending an order rather than a list, recommendation three should begin immediately since it is cheapest and addresses the widest variation found; recommendation two should follow at month four once designed; and recommendation one requires negotiation with representatives and should be scheduled to conclude by month six. Maylor and Turner (2022) note that implementation failure more often follows from attempting everything simultaneously than from any individual action being wrong.
Limitations of this project. Stating these plainly, the interview sample of twelve is small and purposive rather than representative; the researcher is the Operations Manager, which may have moderated what current drivers said; the rest-day analysis excluded the smallest depot owing to incomplete records; and the study is cross-sectional, so associations identified are not causal relationships. Bell, Harley and Bryman (2022) note that acknowledging methodological limitation strengthens rather than weakens a research report, because it allows the reader to weigh the conclusions appropriately.
References
Bell, E., Harley, B. and Bryman, A. (2022) Business research methods. 6th edn. Oxford: Oxford University Press.
Braun, V. and Clarke, V. (2022) Thematic analysis: a practical guide. London: SAGE.
Buchanan, D.A. and Huczynski, A.A. (2023) Organizational behaviour. 11th edn. Harlow: Pearson.
Chartered Management Institute (2023) Taking responsibility: why UK plc needs better managers. London: CMI.
Data Protection Act 2018, c. 12. Available at: https://www.legislation.gov.uk/ukpga/2018/12 (Accessed: 26 August 2026).
Maylor, H. and Turner, N. (2022) Project management. 5th edn. Harlow: Pearson.
Saunders, M.N.K., Lewis, P. and Thornhill, A. (2023) Research methods for business students. 9th edn. Harlow: Pearson.
Whittington, R., Regnér, P., Angwin, D., Johnson, G. and Scholes, K. (2023) Exploring strategy: text and cases. 13th edn. Harlow: Pearson.