The challenges of integrating SaaS applications


In 1999, Salesforce created the first true software-as-a-service (SaaS) package. Now there are more than 10,000 cloud-based applications aimed at enterprises and SMEs, covering everything from core business functions, such as finance or enterprise resource planning (ERP), to niche tasks.

This proliferation of SaaS technology has helped organisations to modernise, digitise and automate business processes. These applications also enabled IT to reduce overheads including hardware, infrastructure and licensing.

But firms now frequently operate multiple SaaS applications and so need ways to connect workflows and ensure consistent, accurate shared data.

Companies such as Salesforce and NetSuite (now owned by Oracle) pioneered the model for subscription-based, or pay-as-you-go, enterprise software. Today, it is easy to forget how revolutionary the model of paying a simple fee, per user, per month was. And there are now businesses – and not just startups – that own hardly any enterprise IT infrastructure at all, running everything they can in the cloud.

However, using multiple SaaS applications in a business can cause a range of issues, partly because of the way they are designed and operate, and partially because customers chose multiple SaaS applications for best-of-breed functionality.

SaaS packages are, for the most part, “functional” or foundational applications, designed to address the needs of a specific part of the business. These include areas such as customer relationship management and sales automation – where Salesforce.com made its name – and fields such as human resources (HR).

But the applications were designed, at least initially, to work on independently. Part of the appeal of SaaS was that departments or functions could source their own technology, independently of IT, and often in areas that IT had overlooked or were under-resourced.

“Most places have SaaS applications, few large organisations don’t have any,” says Andrew Larssen, an enterprise IT specialist at PA Consulting Group.

“We see a lot of these in HR, as it is an internal function that can be a bit forgotten as it doesn’t generate money. That’s not where the big spending is.

“The other area where you see a lot of SaaS is sales and relationship management. A lot of companies have something in that space to manage their contacts and leads…but even when you buy from a single vendor, often the challenge is getting everything to work together.”

And this challenge becomes even more difficult to address, as the number of separate, “siloed” applications increases, and as organisations attempt to automate their workflows.

SaaS sprawl

Today’s enterprises are likely to run dozens, if not hundreds, of separate SaaS applications.

The ease of acquiring and setting up SaaS packages, their self-service nature – often with no need to involve IT – and the understandable desire of departments to buy best-of-breed technology has led to “SaaS sprawl”. And, although departments might see themselves as siloes with their own, specific IT needs, the business as a whole needs common workflows and consistent, shared data.

“These SaaS applications often only specialise in one domain, such as ERP, CRM [customer relationship management] or HR,” notes Steve Deng, senior director analyst at Gartner. “But they don’t work in isolation. Common integration use cases include data synchronisation and orchestration of business processes across these applications.”

As Deng points out, firms need to keep data consistent across all their SaaS applications. This is easier said than done, as applications will typically hold data in their own data stores and on their own cloud infrastructure.

Enterprises need to ensure they use consistent data formats so they can exchange data between SaaS applications and applications run by suppliers and customers. Enterprises need to “gain visibility and insight from data that are distributed across these SaaS applications”, he adds.

“SaaS applications and on-premise applications may represent or store their data under different data models and formats,” cautions Deng. “Many organisations struggle with keeping data consistent across applications in a timely manner.”

Poor quality or inconsistent data can all too easily undermine some of the benefits of buying a best-of-breed SaaS application – and inefficient or broken workflows only add to the problems.

“Enterprises do need to integrate their various systems to be able to produce management reporting or transfer data between systems,” says PA’s Larssen. “If you have a sales system or manufacturing system, you need to transfer data between them. Without being able to do that, you end up with business silos or rekeying data. Data is the lifeblood of a business and it needs to flow around.

“The alternative is the ‘swivel chair’ approach, with users re-keying information between systems. That is time consuming and even more error prone,” he warns.

And integrating SaaS applications is potentially even more important, where organisations want to automate their processes, or use advanced analytics, machine learning or AI.

Integration options

Fortunately, the options for integrating SaaS applications are improving. Application software providers are making their products easier to integrate. As the SaaS market has grown, new entrants have competed, in part, on their ability to work with other software providers and their datasets and workflows. This also allows smaller software firms to offer more specialist, niche applications which can interoperate with the larger, enterprise applications and suites.

“The more forward-thinking vendors would rather have a smaller slice of the pie than none of it, so are much happier to integrate with all sorts of products,” says PA’s Larssen.

Then there are the technical advances in application and data integration. SaaS applications have gone from supporting – often fairly limited – data export to in-depth support for application programming interfaces (APIs), allowing firms to build connections between tools.

Meanwhile, a greater choice of SaaS applications – and the need to integrate with on-premises applications, legacy workflows and external data sources – has created its own marketplace for integration tools.

A number of software providers now offer low-code, no-code or even pre-built, drag and drop methods. These integration-platform-as-a-service (iPaaS) tools are now an industry in their own right, and firms can either use them directly to create their own bespoke workflows, or bring in consultants who have experience of building similar workflows, often in a similar industry.

Some software tools companies specialise in adding events engines and streaming, which allow looser connection between applications than APIs. This can be useful in cases where a delay or disruption to one application would disrupt another, or even cause data loss, in a closely coupled, API-based environment. A further option still is to use the cloud itself, as a basis for integration, with tools such as AWS’ Lamda.

Just as there is only rarely a one-size-fits-all SaaS application that covers all parts of a business, so there is no single best way to integrate different applications.

For some, for example, where applications frequently share data and form part of a closely coupled workflow, API integration may be best. For others, where applications are more independent, and workflows need resilience so that failure or disruption in one SaaS application does not cause disruption to others, an event or streaming based connection might be better.

In some cases, where data sharing is less time sensitive, a conventional batch-based or extract, translate and load (ETL) process could be more appropriate. The cost and effort of connecting APIs might not be justified in a workflow that only needs to run once a week.

However, as PA’s Larssen points out, CIOs may have to pick the best integration method to serve a range of requirements as it makes little sense to run multiple forms of integration between the same applications. “If you have two products you are trying to integrate, it is less work to integrate them all the same way,” he says.

Taking an integration between an HR and an IT system as an example, he points out that although a batch process would work well for adding for new joiners, and an API for locking and unlocking accounts, the smarter option is to use just one method. “You may as well use the APIs for everything,” he says.

Firms’ choices will also be driven by the time, skills and resources they have to integrate their SaaS applications. A high-value, critical or highly custom workflow merits custom coding; a common, utility workflow may well be better achieved through an off-the-shelf tool or iPaaS application.

“Unless you source your applications from a single vendor, sooner or later you will need to integrate applications and data,” says Gartner’s Deng. But whichever option CIOs choose, it should be a more reliable, accurate and efficient option than the “swivel chair” and rekeying data.



Source link