Research and analysis

HMRC Costly Channels usage amongst Small Businesses: Technical Report

Published 29 May 2025

1. Glossary

Word Definition
VAT Value Added Tax
PAYE Pay As You Earn
SA Self Assessment
IDBR Inter-Departmental Business Register
HMRC HM Revenue and Customs
CATI Computer-Assisted Telephone Interviewing
CAWI Computer-Assisted Web Interviewing
ISBA Individuals, Small Businesses and Agents

2. Introduction

HM Revenue and Customs (HMRC) commissioned BMG Research to undertake research to ascertain robust evidence on the customer experience of small businesses, and how this is linked to ‘Costly Behaviours’.

The core objectives of this research were:

  • to develop and implement a questionnaire to measure customer experience and ‘Costly Behaviours’
  • ensure that survey design, procedures and materials are of robust quality, and appropriate pre-testing methods are incorporated into the methodology
  • consider the fixed list of behaviours within a pre-existing theoretical model and validate these with empirical research
  • conduct survey fieldwork and report results to a wide range of stakeholders, including HMRC senior leadership teams

This research aimed to answer the following questions:

  • what is the prevalence amongst small businesses of behaviours costly to customers and HMRC?
  • what causes ‘Costly Behaviours’ amongst small businesses?
  • how does the customer experience of small businesses influence their behaviour, and what is the link to ‘more costly’ customer behaviours?
  • are there potential opportunities for customers to change this behaviour, and so reduce costs?

3. Survey

3.1 Questionnaire

The questionnaire was developed by HMRC and BMG to address the research objectives.

The questionnaire was developed by using a pre-existing behavioural model as a guide (COM-B model for ‘Costly Behaviours’) and, where relevant, used questions from the HMRC Individuals, Small Businesses and Agents (ISBA) Survey.

3.2 Costly channels

HMRC defined ‘costly channels’ as customer interactions that result in additional or avoidable costs for HMRC. Channels that are advisor lead were defined as costly in this research. Channels that may be costly to customers were not considered in this research.

Channels that were costly were defined as:

  • telephone
  • advisor-led webchat
  • post
  • face to face (carried out by Customer Compliance for a compliance check, enquiry or to prevent criminality)

Channels that were non-costly were defined as:

  • online — to search for information on the HMRC webpages on gov.uk
  • online — HMRC digital assistant
  • online — HMRC app
  • online — use the Business Tax Account/Personal Tax Account accessed via Government Gateway
  • online — HMRC YouTube tutorial
  • online — non-HMRC webpages
  • telephone HRMC automated phone system
  • ask a tax agent or accountant
  • ask a friend, family member or colleague

3.3 Scenario approach

To address the research objectives, BMG Research used a scenario-based approach. Respondents were presented with scenarios to identify specific actions and preferences in set situations to gain a reflection on their potential behaviour. 

Eight scenarios were created based on problems or questions HMRC small business customers may encounter. The scenarios were categorised into types of interactions: payment; searching for information; and making a query.

Each respondent was shown two scenarios at random. No respondent was shown two scenarios within the same category.

For each scenario, respondents were asked where they would go first to answer their query and were provided with a list of options with different contact channels.

Respondents were then asked their reason for selecting the first channel.

The respondents who selected a costly channel as their first choice, were then shown a non-costly contact channel (selected at random), and asked why they had not chosen that channel. The non-costly channel options given to respondents were searching online for information on the HMRC webpages, the HMRC digital assistant, the HMRC app, the online HMRC business or personal tax account and HMRC YouTube tutorials.

Finally, respondents were asked the length of time they would take to look for an answer to their query before using a different channel to answer it.

Table 1: Scenarios used

Scenario group Scenarios Question text
Payment I am not sure that I have paid the correct tax Please imagine you want to check you have paid the correct amount of tax. In this scenario, where would you go first to answer your query?
Payment I am making a payment to HMRC Please imagine you are making a payment to HMRC. In this scenario, where would you go first to make a payment to HMRC?
Payment I want to challenge a penalty I have received Please imagine you have received a penalty notice that you think is a mistake. In this scenario, where would you go first to challenge a penalty you have received?
Information I am a start-up small business. When do I need to register to pay tax? Please imagine you are a small start-up business, starting your business for the first time and you need to know when to register to pay tax. In this scenario, where would you go first to answer your query?
Information I am doing my Self-Assessment tax return and I want to check which expenses I can claim for. It’s getting close to the deadline. Please imagine you are doing your Self-Assessment tax return, it is close to the deadline and you want to check which expenses you can claim for. In this scenario, where would you go first to answer your query?
Information I want information on how to register for VAT Please imagine you want to find information on how to register for VAT. In this scenario, where would you go first to answer your query?
Querying I have received my tax bill and I believe it is incorrect Please imagine you have received a tax bill but believe it is incorrect and you want to query it. In this scenario, where would you go first to answer your query?
Querying I have just paid HMRC but the payment isn’t showing on my tax account Please imagine you have just paid HMRC but the payment isn’t showing on your tax account. In this scenario, where would you go first to answer your query?

3.4 Questionnaire testing

Cognitive testing was conducted with 20 respondents to ensure new questions and the scenarios were interpreted by the respondents consistently and in the intended way. A pilot survey was also conducted to test the survey and the fieldwork approach.

3.5 Questionnaire structure

The questionnaire began with screening and profiling sections. Respondents were then asked about their general perceptions of HMRC and their experience of contacting HMRC in the last 12 months. The questionnaire ended with each respondent being shown two scenarios and asked where they would go for each one and why. The topics included in the questionnaire are detailed below.

The questionnaire was tested by using cognitive testing of some questions with customers, followed by a pilot survey of customers.

Profiling questions

  • selfemployed status
  • partnership status
  • years business has been operating
  • years in current role

General Perceptions

  • overall experience of HMRC in the last 12 months
  • perception of HMRC
  • trust in HMRC
  • confidence dealing with HMRC
  • knowledge of taxes they pay
  • worry about the consequences of making a mistake on their taxes

Contact Experience

  • contact method
  • taxes contacted about
  • contact experience

Scenarios

  • where they would go first to answer their query
  • reason for selecting contact channel
  • reason for not selecting a non-costly channel
  • length of time they would search for an answer through a non-costly channel before using a different method
  • whether agree that HMRC took circumstances into account
  • whether HMRC made any errors with respondents’ tax dealing

The average survey length was 19 minutes and 27 seconds.

3.6 Quality Control

A minimum of 5% of all CATI fieldwork is monitored either in a real time environment or by listening to recorded interviews with a minimum of 75% of the interview listened to or monitored. Electronic monitoring sheets are used to document details of the call monitored with a systematic method used for selecting interviewers to be monitored. 

4. Fieldwork

This survey consists of a random probability online and telephone survey. The pilot was conducted between 31 May 2023 and 7 June 2023 and the main fieldwork was conducted between 4 July 2023 and 22 August 2023. A total of 2,001 interviews were achieved.

4.1 Data collection methods

This survey used Computer-Assisted Telephone Interviewing (CATI) and Computer-Assisted Web Interviewing (CAWI). All contacts sampled from the HMRC Self-Assessment (SA) database were sent a pre-survey letter which invited them to participate in the survey via an online link (with access using a survey reference number). All respondents were able to request to do an online survey during the telephone introduction which was administered by e-mail using e-mail addresses provided by the respondent.

The survey, and information letters, were offered in Welsh for those in Wales.

4.2 Achieved surveys

The research comprised a quantitative survey of 2,001 interviews, 1,798 surveys completed by telephone and 203 completed online.

Of the Inter-Departmental Business Register (IDBR) sample 1,434 completed the survey via telephone and 90 completed the survey by e-mail invitation following a request to complete the survey online.

In total, 5,554 SA respondents were sent a pre survey letter with the option to complete online. 72 completed it online via the pre survey letter which represents a 1.3% response rate. A further 41 completed the survey by e-mail invitation following a request to complete the survey online and 364 completed a telephone survey.

Table 2: Method of data collection by sample type

Method of data collection IDBR sample SA Sample Total
CAWI (via letter) 0 72 72
CAWI (via e-mail) 90 41 131
CATI 1434 364 1798

4.3 Sample source and definition of Small Businesses

For the purpose of this survey Small Businesses were defined as having a turnover under £10m and between 0 and 19 employees. Following the same process as the HMRC ISBA Survey the sample of Small Businesses was sourced from two databases:

  • establishments listed in the IDBR
  • sole traders or partnerships that had made a SA tax return

4.3.1 The Inter-Departmental Business Register (IDBR)

This is a comprehensive live list of all UK businesses that are VAT registered and or have a PAYE scheme. It is obtained from five administrative sources (HMRC VAT, HMRC PAYE, incorporated businesses registered at Companies House, DEFRA farms and the Department of Finance and Personnel, Northern Ireland (DFPNI).

This sample was restricted to those businesses in which the enterprise had fewer than 20 employees and an estimated turnover below £10m.

4.3.2 HMRC Self Assessment database (SA)

The SA database provides coverage of smaller businesses that are not VAT registered (or with a PAYE scheme) and therefore missing from the IDBR. However, as the SA tax return annual deadline is 31 January of the subsequent tax year, and the database is not considered up to date for another 6 months, the SA database is representative of the population as it was (at best) 24 months previously.

This sample was restricted to those sole traders or partnerships that have a turnover above £9,170 but below the VAT threshold or was identified as an Agent (Trade Classification Number of 6201 or 6615 or with an SA description which is used to identify Agents).

4.4 Sample design

Following the same process as the HMRC ISBA Survey, Small Businesses were sampled and surveyed at the Enterprise level. An Enterprise can be defined as the smallest combination of legal units (generally based on VAT and or PAYE records) that is an organisational unit producing goods or services, which benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. An enterprise carries out one or more activities at one or more locations. An enterprise may be a sole legal unit. A local unit is an enterprise or part thereof (such as a workshop, factory, warehouse, office, mine or depot) situated in a geographically identified place. 

The target for the overall sample size for this study was 2,000, aiming to conduct 500 surveys amongst the SA Sample and 1,500 amongst the IDBR sample. The sample design included some deliberate disproportionate sampling by turnover size amongst the SA sample and employee size amongst the IDBR sample. This was to ensure there was a sufficient sample size within each cohort to undertake sub-group analysis. It was recognised within this sample design that more weighting would be required (than if it was to be representative) to weight the sample back to the population which would diminish the weighting efficiency of the total sample. If the sample was truly representative, and no weighting was required, our population estimates would have a maximum margin of error of ±2.2% (at 95% confidence level) of the true population. With the reduced sampling efficiency, we would expect our population estimates to have a maximum margin of error of ±3.4% (at 95% confidence level) of the true population.

4.5 Sample stratification

BMG Research used data from IDBR and HMRC SA database to inform the sample design which sought to deliberately disproportionately sample by turnover size amongst the SA sample and employee size amongst the IDBR sample.

The primary sample frame for this research was the IDBR. The IDBR sample was stratified by major SIC category and a systematic sample was drawn within the following cells:

  • 0 employees
  • 1 to 4 employees
  • 5 to 9 employees
  • 10 to 19 employees

There are many businesses not listed on IDBR (e.g., those not registered for VAT and without PAYE). HMRC’s SA database was used to draw a sample from this population. Eligible SA records were sorted by region and, within region, sorted by income from sole trading and or partnership activity, and then a systematic sample was drawn within the following cells:

  • partnership £9,170 to less than £20,000 turnover
  • partnership £20,000 to less than £40,000 turnover
  • partnership £40,000 to less than VAT threshold turnover
  • sole trader £9,170 to less than £20,000 turnover
  • sole trader £20,000 to less than £40,000 turnover
  • sole trader £40,000 to less than VAT threshold turnover

Sample was selected with the aim of achieving approximately 75% of interviews with IDBR sample and approximately 25% with SA sample.

4.6 Opt Out

Under Data Protection Act requirements, an opt-out stage was included for the HMRC sample drawn from the SA database. This involved sending a letter to all sampled respondents and providing them with the opportunity to either call a free phone number, or email if they did not want to take part in the survey. Any Small Businesses sampled in Wales were sent a letter with the text in both English and Welsh languages.

Two weeks were allowed between mail out of the letter and the start of fieldwork.

Under one per cent of Small Businesses opted out of the research following receipt of the letter.

4.7 Respondent selection

Where a named contact was available interviewers were asked to speak to the named contact. Where there was no named contact, interviewers were asked to identify the owner or the person who has responsibility for accounts, tax or payroll at the business. The objective was to find the person most informed about the businesses’ interactions with HMRC to reflect the views of the business. 

4.8 Eligibility

At the start of the survey there was a series of questions to establish eligibility as a Small Business. All Small Businesses were asked to confirm the number of employees and annual turnover to establish that they have less than 20 employees and a turnover between £9,170 and £10 million. Any businesses that did not meet these criteria were screened out of the study.

4.9 Sample productivity

The sample did not (for the most part) include telephone numbers so a look up process was undertaken. This included an automated process, followed by a manual look up stage for the remaining sample. Table 3 shows the number of sample records obtained and the rates of number lookup success.

Table 3: Sample order and number lookup rates

Sample Initial selection % Valid numbers after lookup
SA Partnership £9,170 to less than £20,000 turnover 631 57%
SA Partnership £20,000 to less than £40,000 turnover 633 60%
SA Partnership £40,000 to less than VAT threshold 604 57%
SA Sole Trader £9,170 to less than £20,000 turnover 3,471 76%
SA Sole Trader £20,000 to less than £40,000 turnover 3,512 78%
SA Sole Trader £40,000 to less than VAT threshold 3,372 77%
IDBR 0 employees 13,067 24%
IDBR 1 to 4 employees 13,018 36%
IDBR 5 to 9 employees 11,174 65%
IDBR 10 to 19 employees 11,178 72%
Total 60,660 54%

The number of sample records issued for fieldwork and the conversion to interview is given in Table 4.

Table 4: Sample conversion rates

Sample Issued for fieldwork Interview conversion Interviews achieved
SA Partnership £9,170 to less than £20,000 turnover 333 10% 34
SA Partnership £20,000 to less than £40,000 turnover 333 11% 36
SA Partnership £40,000 to less than VAT threshold 333 6% 21
SA Sole Trader £9,170 to less than £20,000 turnover 1,522 8% 117
SA Sole Trader £20,000 to less than £40,000 turnover 1,522 9% 138
SA Sole Trader £40,000 to less than VAT threshold 1,511 9% 131
IDBR 0 employees 3046 8% 241
IDBR 1 to 4 employees 4,552 6% 252
IDBR 5 to 9 employees 4,921 9% 456
IDBR 10 to 19 employees 5,543 10% 575
Total 3,616 8% 2,001

4.10 Response rate and eligibility calculations

Replicating the same process as the HMRC ISBA Survey, BMG Research have estimated response rates and estimated eligibility (used for weighting calculations) based on the American Association for Public Opinion Research (AAPOR). Table 5 shows the final fieldwork outcomes and estimated response rate and eligibility rates for Small Businesses. This web link takes you to the AAPOR website where further information can be found on response rates: https://aapor.org/response-rates/.

Achieving interviews with businesses proved challenging. The important factor appeared to be the ongoing effect that COVID-19 had on the way businesses were operating. Specifically, working from home (or not regularly working from an office) had an effect on availability for contact via telephone or availability during fieldwork. 

Table 5: Response rate — Interviews

Outcome
Total sample issued 23,616
I = Complete Interviews 2,001
R = Refusal and break off 3,527
NE = Ineligible or non-working numbers 4,257
U= Unknown eligibility 13,831
e = estimated eligibility 56%
Response Rate: ( I / I + R + (e(U)) ) 15%

5. Weighting approach

The weighting approach followed the weighting approach used in the ISBA survey and comprised 2 steps. The first step was the calculation of a design weight based on the estimated sampling probability. The design weight was applied to the dataset, and in the second step the weighted data was calibrated so that it was representative of the population being sampled.

5.1 Estimation of the population

The populations from which the samples were drawn outlined previously are shown in Table 6.

Table 6: Sample group populations

Sample group Total
SA Partnership £9,170 to less than £20,000 turnover 19,856
SA Partnership £20,000 to less than £40,000 turnover 32,604
SA Partnership £40,000 to less than VAT threshold turnover 49,375
SA Sole Trader £9,170 to less than £20,000 turnover 1,021,435
SA Sole Trader £20,000 to less than £40,000 turnover 867,497
SA Sole Trader £40,000 to less than VAT threshold turnover 405,903
IDBR 0 employees 261,361
IDBR 1 to 4 employees 1,699,441
IDBR 5 to 9 employees 278,202
IDBR 10 to 19 employees 146,805
Total 4,782,479

Although the sample frame approach was designed to minimise overlap, as there is a lag time in the SA database, some businesses will have multiple chances of selection. The degree of sample frame overlap, and thereby the total population distribution, was estimated from the data by:

  • calculating the survey eligibility rates (Table 7) and adjusting the sample group population to reflect these rates (Table 8)
  • estimating the sample frame overlap by using the respondent data from the HMRC SA database for those who stated their enterprise was registered for VAT or had a PAYE system, the conditions required to be in the current IDBR (Table 9).
  • Recalculating the final population estimates taking into consideration the estimated eligibility and overlap (Table 10).

Table 7: Observed survey eligibility rates

Sample group % Eligible
SA Partnership £9,170 to less than £20,000 turnover 74%
SA Partnership £20,000 to less than £40,000 turnover 83%
SA Partnership £40,000 to less than VAT threshold turnover 74%
SA Sole Trader £9,170 to less than £20,000 turnover 84%
SA Sole Trader £20,000 to less than £40,000 turnover 82%
SA Sole Trader £40,000 to less than VAT threshold turnover 87%
IDBR 0 employees 82%
IDBR 1 to 4 employees 84%
IDBR 5 to 9 employees 87%
IDBR 10 to 19 employees 82%

Table 8: Estimated survey eligible population

Sample group Total
SA Partnership £9,170 to less than £20,000 turnover 14,693
SA Partnership £20,000 to less than £40,000 turnover 27,061
SA Partnership £40,000 to less than VAT threshold turnover 36,538
SA Sole Trader £9,170 to less than £20,000 turnover 858,005
SA Sole Trader £20,000 to less than £40,000 turnover 711,348
SA Sole Trader £40,000 to less than VAT threshold turnover 353,136
IDBR 0 employees 214,316
IDBR 1 to 4 employees 1,427,530
IDBR 5 to 9 employees 242,036
IDBR 10 to 19 employees 120,380
Total 4,005,043

Table 9: Estimated overlap between SA and IDBR

Sample group % on IDBR
SA Partnership £9,170 to less than VAT threshold turnover 41%
SA Partnership £20,000 to less than £40,000 turnover 53%
SA Partnership £40,000 to less than VAT threshold turnover 57%
SA Sole Trader £9,170 to less than £20,000 turnover 40%
SA Sole Trader £20,000 to less than £40,000 turnover 41%
SA Sole Trader £40,000 to less than VAT threshold turnover 45%

It should be noted that the overlap is estimated from the survey and there are therefore margins of error around the estimates.

Table 10: Estimated survey population distribution

Sample group Total
SA Partnership £9,170 to less than £20,000 turnover 8,643
SA Partnership £20,000 to less than £40,000 turnover 12,779
SA Partnership £40,000 to less than VAT threshold turnover 15,659
SA Sole Trader £9,170 to less than £20,000 turnover 498,670
SA Sole Trader £20,000 to less than £40,000 turnover 422,685
SA Sole Trader £40,000 to less than VAT threshold turnover 188,698
IDBR 0 employees 214,316
IDBR 1 to 4 employees 1,427,530
IDBR 5 to 9 employees 242,036
IDBR 10 to 19 employees 120,380
Total 3,151,396

In line with the ISBA survey, for the purposes of weighting, the estimated population has been recoded into the following 2 variables (Table 11 and Table 12).

Table 11: Weighting targets — Number of employees

Number of employees N %
IDBR 0 employees 214,316 6.8%
IDBR 1 to4 employees 1,427,530 45.0%
IDBR 5 to 9 employees 242,036 7.6%
IDBR 10 to 19 employees 120,380 3.8%
Sole Trader (not on IDBR) 1,130,111 35.6%
Partnership (not on IDBR) 37,081 1.2%

Table 12: Weighting targets — Industry sector

Industry sector N %
Sole Trader or Partnership (not on IDBR) 1,167,192 36.8%
ABDE — Primary 116,172 3.7%
Manufacturing (including repair and installation of machinery and equipment) — C 95,688 3.0%
Construction (including electricians and plumbers) — F 299,037 9.4%
Wholesale and retail trade (including repair of motor vehicles and motorcycles) — G 306,662 9.7%
Hotels and restaurants (including accommodation and food service activities) — I 126,937 4.0%
Transport, storage, and communication (including information) — H,J 237,232 7.5%
Finance and insurance, Real estate, renting or business activities (including legal services, administrative and support services) — K,L,N 283,515 8.9%
Education — P 31,634 1.0%
Health and Social work — Q 75,120 2.4%
Other services — M, O, R, S, T OR U 432,265 13.6%

5.2 Calculation of the design weight

In line with the ISBA survey, an estimated sampling probability was computed for each case:

p(HMRC) + p(IDBR)

Where:

  • p(HMRC) = probability of being sampled from the HMRC SA database
  • p(IDBR) = probability of being sampled from the IDBR

Both p(HMRC) and p(IDBR) were calculated as the product of 2 separate components for each stratum:

  • proportion of sample group population that was sampled
  • proportion of usable sampled cases

Usable cases were those cases for which a telephone number was found.

All cases had a sampling probability for the database from which they were originally selected. Cases which were identified in the survey as also being on the other database were assigned a sampling probability for both.

Cases selected from SA were identified as also being on IDBR if their business was registered for PAYE and or VAT.

Cases selected from IDBR were identified as also being on SA if they met all the following conditions:

  • self-employed
  • turnover of below the VAT threshold
  • they (or their partner) paid some Class 2 National Insurance and or Income Tax through SA

The design weight was calculated as the inverse of the estimated sampling probability (i.e. one divided by the sampling probability).

5.3 Calculation of the final weight

The design weight was used as the input weight in a calibration matrix designed to force the respondent profile to match the population profile. Two dimensions were used for this survey; number of employees for IDBR sample, turnover for HMRC SA sample. These were based on the estimated survey eligible population shown in the preceding tables (Table 11 and Table 12).

As shown in Table 13 and Table 14, the weighted sample margins match the population margins along with the design effect — estimated at 2.35 (an effective sample size of approximately 850).

Table 13: Sample conversion rates — Number of employees

Number of employees Estimated Population % Final Weighted Sample %
IDBR 0 employees 6.8% 6.8%
IDBR 1 to 4 employees 45.0% 45.0%
IDBR 5 to 9 employees 7.6% 7.6%
IDBR 10 to 19 employees 3.8% 3.8%
Sole Trader (not on IDBR) 35.6% 35.6%
Partnership (not on IDBR) 1.2% 1.2%

Table 14: Sample conversion rates — Industry Sector

Industry sector Estimated Population % Final Weighted Sample %
Sole Trader or Partnership (not on IDBR) 36.8% 36.8%
ABDE — Primary 3.7% 3.7%
Manufacturing (including repair and installation of machinery and equipment) 3.0% 3.0%
Construction (including electricians and plumbers) 9.4% 9.4%
Wholesale and retail trade (including repair of motor vehicles and motorcycles) 9.7% 9.7%
Hotels and restaurants (including accommodation and food service activities) 4.0% 4.0%
Transport, storage, and communication (including information) 7.5% 7.5%
Finance and insurance, Real estate, renting or business activities (including legal services, administrative and support services) 8.9% 8.9%
Education 1.0% 1.0%
Health and Social work 2.4% 2.4%
Other Sectors (B or D or E or M or O or R or S or T or U) 13.6% 13.6%
Estimated design effect   2.35
Estimated effective sample size   850