The essential guide to data analytics

The benefits of data analytics tools have long been hailed as a new dawn for the profession. But the accounting industry has been more cautious than some professions to adopt new digital tools. Since the start of the pandemic, however, there has been an explosion of interest from accountants for data analytics skills, as many realise the power of data and the insights it can provide to improve business efficiency.

The adoption of data analytics is expected to continue to grow. Gartner forecasts that 75% of organisations will shift from piloting to full implementation of artificial intelligence (AI) by the end of 2024, with public cloud services being essential for 90% of data and analytics innovation by 2022.

Many industry sectors have suffered amid the pandemic, and more businesses are likely to fail despite a bounce in economic growth. Business owners will want to ensure they remain competitive and efficient and will be looking to their accountants to provide faster insights, so that they in turn can make quicker, more accurate decisions. Businesses have also adopted automation and will expect accountants to be fluent in data and analytics in client discussions.

It is up to finance teams to harness these new skills. Accurate, meaningful data may mean the difference between business survival and failure. It’s not just about capturing and storing data, the narrative of the data you present is also crucial.

Understanding business goals is critical to analysing data

Finance teams are ideally placed to become a pivotal part in any data analytics project. Analysing data and presenting it to decision makers in an easily digestible format has long been a key skill in any accountant’s toolbox.

The speed of analytics tools can remove much of the heavy lifting, freeing finance teams up to spend more time on the analysis and presentation of data to gain crucial insights. 

There are different ways in which to analyse data. Predictive analytics, also known as business intelligence, is about understanding patterns in data such as customer churn or behaviour. For example, a retailer might want to understand whether sales were better in a certain part of the UK, and why.

Prescriptive analytics, also known as optimisation, is about the best course of action for the future. This kind of analysis allows you to understand how your plans worked out once you’ve defined your goal. For example, a retailer might look at how many pairs of jeans were sold.

Diagnostic analysis seeks to understand the root causes of events. For example, it will look at why sales rose or fell during a particular timeframe (such as changes in weather).

For these kinds of analyses, you’ll need to understand the source of the data, whether it’s trustworthy and whether it has gone through due process. No one has 100% perfect data, so it’s about understanding the quality of the data.

Defining the analytical approach is vital. Data scientists will build an excellent analytics model but may lack business knowledge to put the information into a useful context. Finance professionals are a natural go-between the data scientists and the decision makers. Accountants understand data, its governance and have the commercial acumen to explain the business benefits of analytics.

Cleaning data is the first step

Data is messy, especially when it is incomplete. How you prepare data and blend different data sources is an important first step. Most data sources are external, with data flows from entirely different systems, so formatting will vary widely. 

The initial step in data preparation is to translate the varying types of available data into a common file type. A comma-separated values (CSV) file might be the best as it is consistent. Common formatting in columns is important, especially when formatting dates to ensure consistency. It’s a matter of defining fields the same way – for example, using a consistent standard for which date is used as the transaction date in complex situations. 

An approach that is growing in popularity among firms is to create central data lakes or data warehouses to ensure a ‘single source of truth’. Data lakes and data warehouses are both widely used for storing big data. But there is a difference: a data lake is a source of raw data, while a data warehouse is a repository for structured data that has already been prepared for a specific task.

However, there may be missing data or duplication. Decisions made at this point will follow on downstream, potentially leading to problems, especially when you start to analyse the data. 

If you’ve got bad data, forecasting and reporting will be problematic. It’s essential to flag issues of missing data or duplication so that you can provide a clean data set for analysis.

Clean data is the baseline for any good investigation. Prepare a documented process for data cleaning so that if an employee leaves, the next person to deal with this has a written process to follow.

Obtaining data from clients’ databases, and then transforming that data into a useable format, may be challenging. The next stage – the lion’s share of the work – will focus on the analysis, which includes modelling and data visualisation. To do this, accountants will need to understand basic statistical testing techniques and concepts such as statistical distributions.

Choosing the right modelling tool

The ability to share data and visualisation in a format that everyone knows shouldn’t be underestimated. For example, Excel is user friendly and familiar.

But spreadsheets can struggle with bigger data sets. We are creating more and more data points and that is unlikely to change. Linking data sets can also be problematic with spreadsheets. The more time accountants save on manual tasks, the more time finance teams have for analysis and forecasting.

This is where modelling tools can help. Coding languages are an efficient way to manipulate data, using complex transformations that you may not be able to replicate with a spreadsheet. And you can repeat those transformations with new data. Coding languages can also help find errors quickly and can be rerun until the mistake is fixed.

Python and R are two of the most popular programming languages in this space. Python is more of a general purpose script language. It can scrape websites, carry out data analysis and complete low-level programming and statistical modelling. It is useful particularly for handling large amounts of data. It’s an all-round good tool. Although R can accomplish many of the same tasks, it is better for visualisation and building statistical models.

Choose a tool that fits your needs. Don’t be blinded by shiny new software. Define your problem and work back from that point. Ultimately, it’s about preference.

Taking a data modelling course is a good way to understand what the languages can achieve. However, data analytics is about continual learning and seeing where these languages can help organisations in their daily functions.

Crucially, if required, data analysis can be carried out in either of these languages and then transferred back to an Excel spreadsheet for further analysis. Accountants are used to manipulating Excel spreadsheets so arguably coding is the next step up.

Defining good practices 

There are numerous ethical considerations to be aware of, which should be built in from the start, particularly as we begin to use more advanced technologies in predictive and prescriptive analytics. 

There are ethical requirements to make sure that you only collect the data you need to collect. Accountants will need to consider the General Data Protection Regulation (GDPR), as well as other industry requirements. Conforming with regulations such as GDPR is the absolute minimum, but individuals need to do more to meet ethical obligations. It is vital that organisations instil an ethical culture around data privacy and confidentiality that goes beyond the legal requirements.

Once you’ve collected the necessary data, make sure you're measuring accurately. Measuring incorrectly can lead to unethical and sometimes devastating outcomes, whether from bad data or incorrect analysis. Failure to put appropriate controls in place can increase the probability of error, so it is vital to check that systems are accurate and controls are effective.

Selling the story 

Knowing how to present your data using the most effective dashboards could be the key to securing board buy-in and meeting business goals.

Around 200 quintillion bytes of data were produced each day in 2021, according to figures from Statista. So, how can finance teams effectively cut through mountains of data to present key insights to decision makers? 

Effective visualisation design can often be taken for granted.  The best designs consider the audience and why they need the data. Everyone must be able to understand the data, not just technical experts. 

Know what your reports and graphics can do and what they should not. It is not always possible to meet everybody's needs.

Data visualisation should not necessarily provide the solution, but the means for the decision makers to resolve specific problems.

In any data-based project, it is important to understand the shape, quantity and quality of available data. Consider how you can label data in design to make it easy to read. Consistent labelling will improve users’ understanding. You can’t include everything, so you must refine and prioritise the content. Detail matters in design.

Andy Kirk, freelance data visualisation specialist, uses the London Tube map as an example of how design is fundamental to users’ needs in data visualisation. Listen to Andy Kirk’s webinar and other recordings here.

Adopting data analytics

Armstrong Watson, a firm serving clients in the North of England and Scotland, has been using data analytics in its corporate finance division for several years. The tools have improved the firm’s capabilities in advising its clients in areas such as business expansion, new product development and succession planning, among others.

Analytics tools have also helped increase efficiencies internally, enabling partners to provide clients with real-time data to help with quicker, more targeted decision-making.

“Being able to create targeted reports to clients that enable them to visualise complex financial information simply, is the key to providing great service. Modern data visualisation tools like Power BI make this so much easier than it used to be,” says Toby Woodhead, head of technology at Armstrong Watson.

The firm is looking into how to use these tools more widely internally to improve processes and systems, but also in client services such as compliance accounting. Currently, the firm is using Fathom and Spotlight, among others. 

“Suppliers are trying to have their own USP, which makes it a bit tricky because they don't necessarily do everything we want. With a lot of technology and accounting, what we're really trying to do is raise the bar for the entire practice, and ultimately for the entire sector,” Woodhead says.

Another area where the firm is looking to develop bespoke analytics tools is in audit. 

Armstrong Watson is working with accounting data specialists Engine B to develop a common data model (CDM) that can be applied widely across the profession.

“We're really looking for data analytics to do ideally more than one but at least one of these three things: improve the quality and efficiency of our audits, and to add value to the client,” Woodhead says.

Outside of the Big Four firms, barriers to the adoption of data analytics in audit include the cost of new technology, ensuring clean, standardised data, and audit methodologies not reflecting the changes in technology.

However, data scientists and accountants are working together so that in the future data analytics will allow auditors to do 100% testing rather than a small sample as is currently the case. 

Once the challenge to refine a common data standard is resolved, it is hoped that the change will increase competition leading to competitive pricing and wider adoption.

Further learning

Employers are increasingly looking for data analytics and data science skills. Finance professionals can become the bridge between decision makers and the growing amount of data organisations are producing.

There are two routes you can take via the ICAEW Data Analytics Certificate qualification: the analyst pathway and the management pathway.

Analyst pathway

The analyst pathway will furnish you with a solid grounding in data analytics and develop your hands-on data analytics skills.

The qualification will teach you to recognise different data types and structures, how to code and manage data, and identify and resolve data quality errors. You’ll also learn to apply more sophisticated data analytics techniques, forecast modelling and the ethics of data analytics.

By the end of the qualification, you’ll be able to effectively communicate your analysis through narrative and visualisations, as well as present findings and recommendations to decision makers.

Management pathway

The management pathway caters for finance leaders who want to understand the business benefits of applying data analytics.

The course will also show you how to develop operational data analytics capabilities and how to interpret data analytics outputs.

The comprehensive modules range from data strategy, governance and integrity, exploratory data analysis, forecasting and risk management to influencing through data visualisation.

The course shows how and why data-driven models can improve decision-making, and the benefits of finance leaders owning or exerting influence over analytics within the business. 

Useful resources

• How Eurobank got to grips with data analytics in internal audit
• Data analytics community
• Data analytics and big data