The key to success with low code is finding the right project for the platform, according to Rebecca Parsons, chief technology officer at ThoughtWorks, a global software consultancy. Parsons said that it’s crucial to have a thorough understanding of the problem before deciding to build a low-code solution for it.

“We are really trying to instill that notion that you have to have that level of understanding: What is the 80% that I want to optimize for and does that fit this low-code platform,” she said.

SEE: Everything you need to know about using low-code platforms (free PDF) (TechRepublic)

IT teams should be testing platforms such as Amazon Honeycode to build simple business applications, according to ThoughtWorks’ biannual Technology Radar report. Well-defined workflow processes, such as those that need read-only copies of data, make the best pilot programs.

“If it was that simple to solve, it wouldn’t take that long to build a solution in the traditional way,” she said. “We’re clearly pushing toward more complicated workflows.”

Processes that have numerous decision points or those that require information from multiple systems are not a good fit.

Parsons said companies should incorporate low-code platforms into the overall software development strategy. Adopting an agile way of working is crucial to success at digital transformation but many organizations still struggle to adopt this approach. Parsons thinks that business leaders need to truly accept the consequences of the increasing pace of change.

“People still cling to this need for predictability and there’s only so much predictability you can get,” she said. “The surface area of impact for this rate of change is increasing, and business leaders are not used to re-evaluating business models on a three- to nine-month basis.”

SEE: Why 2021 will be the year of low-code (TechRepublic)

If executives can get comfortable with pushing forward despite high levels of uncertainty, other changes flow from that shift in mindset shift.

“We have to be able to accept that we might have to change direction very quickly and figure out what do you have to do as an organization to get decision making to work in that context,” she said.

Parsons said that low-code platforms are not necessarily new but instead the current manifestation of what people tried to do with fourth-generation languages.

“The newer ones are trying to bring in context and widen the footprint and bring in more domain context in customer service or relationship management,” she said.

Here are Parsons’ suggestions on how to make low-code projects successful.

Five tips for success with low code

Just as software developers have learned the basics of the business operations, department leaders need to understand at a high level how software works.

“What do we have to do to equip people who know the business but not software a basic working knowledge?” she said.

Parsons also recommends having a person on the low-code development team who understands the limitations of the tech and can provide high-level guidance for the project. That way the individual can provide high-level guidance but not be involved in day-to-day work.

Low-code platforms are good for prototyping a new solution.

“Software engineers hear all the time from business colleagues, ‘You gave me exactly what I asked for but not what I want,'” she said.

Building a prototype in low code helps spot those disconnects early on in the process. She said companies have to make sharp distinctions between production code that is deployable and secure and proof-of-concept code that will be limited to that phase of development.

SEE: The CIO’s guide to low-code platforms (TechRepublic Premium)

“People get so reliant on something that was a prototype and all of a sudden you have a business critical function that is a prototype,” she said.

It’s also important to make a plan for exceptions that a low-code platform can’t handle.

Parsons described an automation project that helped a client realize how rare a particular edge case was. Writing code to address that edge case would have taken two-person years of work time.

“The client did a data warehouse query and found that over the last five years, it only happened three times,” she said.

The solution was to take a week to write code that could spot one of those exceptions and flag it for special handling.

Companies also should consider how low code fits into the overall strategy of continuous delivery and consider how testable the code is and whether deployment can be automated, Parsons said.

Parsons said that small- to midsize companies can take a different approach to modernizing workflows: Change the process to match the way the software wants to do it.

“If you make the problem fit the tool,” she said, “it could mean you have more exception cases but you can significantly expand the footprint of where these things apply.”