Not long ago, an audit assistant testing journal entries would select a few dozen from a file, support them one by one and conclude on the sample. Today, in many firms, the same assistant imports all of the year’s entries into an analysis tool, applies risk filters to the whole population and only supports the entries that come out of it. This change, grouped under the name data analytics, alters the day-to-day work of engagements and what firms expect from candidates. This article explains what is changing, the most common analyses, the skills actually required at each grade, how to train without becoming a data scientist, how to show it in an application, and the limits of these tools.
From sampling to exhaustive testing of entries
Audit has always relied on sampling: you cannot check everything, so you select representative or risky items. Data-analysis tools now make it possible to process an entire population in a few minutes, provided a reliable extraction is available.
In practice, this changes three things:
- selection: instead of a random sample, the auditor targets items with risk characteristics, across the whole population;
- coverage: some tests, such as searching for duplicates or missing invoice sequences, become exhaustive;
- documentation: the tool keeps a record of the processing carried out, which allows a reviewer to retrace the steps.
The French professional auditing standards do not impose any tool, but they require auditors to gather sufficient appropriate audit evidence and to take the entity’s IT environment into account. Firms have made data analysis a way of meeting these requirements more efficiently. The large networks also recruit technical profiles into their audit teams, as shown by careers pages presenting their tech roles or their programmes combining financial audit with a technology specialism.
The most common data analyses on engagements
Journal entry tests. This is the most widespread use. You import the journal entry file, check that it is complete (reconciliation with the trial balance), then apply criteria: manual entries, entries posted after year-end or at weekends, round amounts, unusual users, rarely used accounts, vague descriptions. The selected entries are then supported.
Detailed analytical review. Analysis of margins by product, site or month; trends in customer and supplier payment terms; comparison of staff costs with headcount. The aim is to identify the variations that require an explanation.
Reconciliations between systems. Reconciling the invoicing system with the accounts, payroll with social declarations, the inventory management system with the trial balance.
Duplicate and sequence tests. Searching for supplier invoices paid twice, missing customer invoice numbers, suppliers created twice with different bank details.
Access rights analysis. In connection with IT audit, identifying users with incompatible rights, such as creating a supplier and approving its payment.
Skills expected of an assistant and a senior
Expectations vary by grade, and you need to distinguish an auditor’s skills from a specialist’s.
For an assistant, firms expect:
- excellent command of spreadsheets: pivot tables, lookup functions, Power Query;
- the ability to check that an extraction is complete before analysing it;
- an understanding of the standard tests offered by the firm’s tool and how to interpret their results;
- rigour in documenting the processing.
For a senior, add:
- designing analyses suited to the client and the risks identified;
- working with the client’s IT department to obtain reliable extractions;
- the ability to explain results to the team and the client;
- coordination with the firm’s data or IT specialists.
For a specialist (audit data analyst, IT auditor), firms look for more technical skills: SQL, Python or R, data-preparation tools such as Alteryx, visualisation, and sometimes knowledge of ERP database structures. The line between these roles and data jobs is thin: the article distinguishing the data analyst from the data scientist 🔗 helps place these profiles.
Training without becoming a data scientist
An auditor does not need to become a developer to make use of data. A few steps are enough to gain autonomy.
- Strengthen your spreadsheet skills. Power Query and pivot tables cover a large share of an assistant’s needs. This is the first investment to make.
- Learn the firm’s tool. Firms train staff on their analysis tools; asking to join engagements where they are used is the best way to learn.
- Get started with a specialist tool. IDEA, ACL or Alteryx offer vendor training and certifications; some master’s programmes include them.
- Discover SQL. Understanding a select query, a join and an aggregation is enough to talk with an IT department and check an extraction.
- Visualise. Power BI or Tableau make it possible to produce readable analyses for the team and the client.
The most important thing is to practise on real cases, on engagements, rather than piling up theoretical training.
Showing these skills in your application
On the CV, each data skill should appear in an experience line describing a real analysis:
-
Before: “Data analytics skills.”
-
After: “Exhaustive journal entry testing for 6 clients in IDEA (completeness check of the extraction, filters on year-end manual entries), 40 entries selected and supported.”
-
Before: “Proficient in Power BI.”
-
After: “Power BI dashboard reviewing margins by site, reused on engagements in the sector.”
Our article on audit software on a CV details how to present the tools section, and the IT auditor CV example shows a profile combining audit and data analysis.
At interview, expect a question such as: “How would you test a client’s journal entries?” A good answer starts with the completeness of the extraction, cites a few risk criteria, and ends with supporting the selected entries. Our article on the case study in an audit interview offers a suspicious-entry case to work on.
Limits and the role of professional judgement
Data tools do not replace the auditor. Three limits should be kept in mind, and expressed at interview.
Data quality. An exhaustive analysis on an incomplete extraction gives false assurance. Checking the completeness and accuracy of the data is the first test, not a formality.
The meaning of results. A filter brings out unusual entries; it is the auditor who judges whether they are abnormal, based on their knowledge of the client.
Areas of judgement. Accounting estimates, provisions, impairments or going concern cannot be tested by an algorithm: they require professional scepticism, discussions with management and a reasoned conclusion.
A candidate who masters the tools and can explain their limits is more reassuring than one who presents them as a solution to everything. To showcase this profile, the CV editor structures your experience on one page and produces a PDF readable by screening software, which picks up the tool names and methods you cite.
Sources
- KPMG France, tech roles at KPMG 🔗.
- Deloitte France, Excellence Track Program, financial audit with a tech or ESG specialism 🔗.
- Haute Autorité de l’audit, the professional standards framework for statutory auditors 🔗.
Frequently asked questions
Do you need to know how to code to work in audit today?
Not for most auditor positions. Excellent command of spreadsheets and of the firm’s analysis tools is enough. Code becomes useful for specialist profiles.
Do exhaustive tests replace sampling?
Not completely. They replace some samples, especially on journal entries or duplicates, but many procedures (confirmations, stock counts, estimates) are still based on selections and judgement.
Is data training valued by firms?
Yes, especially if it comes with a solid grounding in accounting. Firms also recruit engineering or data analyst profiles for their specialist teams.
How can you talk about data analytics without firm experience?
By describing a concrete project: analysis of published accounts, a dissertation, a master’s project on a real dataset. What counts is showing the approach, from extraction to conclusion.