Building modern finance teams: typical mistakes when building a finance data foundation

Finance teams are under increasing pressure to automate processes, improve reporting and prepare for a future where AI plays a much larger role in decision-making.  

Yet many organisations encounter the same challenges when trying to build a stronger finance data foundation. 

Across the market, businesses are investing in forecasting tools, reporting platforms and AI solutions. But success depends less on the technology itself and more on the foundations that sit underneath it.  

The quality of data, system architecture and user adoption ultimately determine whether these initiatives create value or add complexity. 

This article is the second in a joint series with Finthera, a Netherlands-based consultancy working at the intersection of finance and data. 

Drawing on their experience across finance, data engineering, business intelligence and AI, we explore the most common mistakes organisations make when building a finance data foundation and how finance leaders can avoid them.

Typical mistakes when building a finance data foundation 

Buying tools to solve symptoms rather than foundations 

The first reflex is almost always wrong: companies buy a tool to solve a symptom. Finance teams want to make progress and don’t want to wait while everything goes through a lengthy process. So they buy a forecasting tool, then a reconciliation tool, then a reporting tool, all separate from one another. 

The result is system sprawl, rising software costs and integrations that either don’t work properly or have to be maintained manually. 

Take a forecasting tool as an example. On the surface, it seems like a logical solution. It looks good in the demo and promises to make processes easier. But it also comes with a modelling suite and reporting functionality. 

It’s designed to pull data in from multiple sources, yet getting data back out can be more difficult because the vendor wants users to model and visualise information within its own platform. 

Before long, the organisation no longer has a tool that solves a specific problem. Instead, it has a platform that performs multiple functions when only one was needed. 

Relying on multipurpose tools that promise simplicity 

Compare that with invoice matching. It’s a single task and the tools that perform it generally integrate well with other systems. Data in, result out. 

That highlights an important distinction. Many platforms promise to handle modelling, reporting and data integration within a single solution. While appealing on the surface, this can reduce flexibility and create dependencies that become increasingly difficult to manage as requirements change. 

By contrast, specialist tools that perform a specific task well are often easier to integrate, easier to replace and less likely to create unnecessary complexity elsewhere. 

Getting the buy versus build decision wrong 

This ultimately comes down to knowing what to buy and what to build. 

Specialist tools that perform a single task well are usually worth buying. Core infrastructure is different. 

Your data warehouse, semantic layer and data pipelines form the foundation of finance reporting. This is where systems, data and business definitions come together, so it needs to be designed around your organisation’s environment rather than forced into a standard solution. 

Getting this balance right has long-term implications for reporting quality, scalability and the future capabilities finance teams will need.

Losing control as AI becomes embedded 

AI introduces another layer of complexity. 

Many providers package existing AI models within their own platforms, allowing organisations to choose different models for different use cases. While attractive, this also creates dependence on the provider’s decisions. 

The underlying model can change, limits can be reduced and output quality can deteriorate without any clear visibility into what’s changed. 

Consider a reconciliation agent running through such a platform. If the provider quietly reduces the context window, the agent may start missing transactions because it can no longer process the full dataset. The output is no longer correct, but the finance team has little visibility into why because it doesn’t control the model itself. 

As AI becomes more deeply embedded in finance processes, organisations will need greater control over how models are deployed, governed and monitored.

Assuming finance teams will automatically adopt new solutions

Even the most sophisticated architecture will fail if finance teams don’t trust it or understand it. 

People will return to Excel – not because it’s better, but because it’s familiar and the logic is visible. 

This remains one of the most overlooked aspects of finance transformation. Technology alone isn’t enough. Users need to understand how outputs are generated, what assumptions sit behind them and how new processes fit into their day-to-day work. 

Sustainable change depends not only on systems, but on transparency. Finance professionals need solutions they can understand, validate and feel confident using. 

Solution

Avoiding these mistakes means focusing on foundations before functionality: 

• Build a clear, scalable data foundation before introducing additional forecasting, reporting or AI tools 

• Buy specialist solutions that perform individual tasks well, and be cautious of multipurpose platforms that promise to do everything in one place 

• Design core infrastructure: data warehouses, semantic layers and pipelines around your organisation’s requirements rather than forcing it into a standard solution 

• Create a single source of truth so KPIs, financial statements and management reporting are based on consistent definitions 

• Retain control over how and which AI models are deployed, governed and monitored, so output quality doesn’t depend on decisions made by a provider 

• Invest in adoption from the start: make reporting logic transparent, documented and auditable, so finance teams can understand, validate and trust the outputs 
• Develop finance teams with the skills to work across finance, data and technology as these foundations mature 

How we can help

As finance functions adopt more automation, analytics and AI-enabled processes, demand for professionals who can bridge finance, data and technology is only likely to increase.  

Many of these skills aren’t part of traditional finance career paths, meaning the talent needed to support future finance teams can be difficult to find. 

Through our network across the Netherlands, we help organisations identify emerging skills gaps and attract the talent needed to support finance transformation and long-term growth.  

Through our partnership with Finthera, we can also connect finance leaders with specialist expertise in finance data foundations, helping organisations prepare both their technology and talent strategies for the future.