This article originally appeared on CFO.com.

From improving accounts receivable and tax preparation to tracking far-flung routers for telecom providers, automation is changing how companies operate and people work across a broad range of business processes. Automation has advanced to the extent that companies lacking an ambitious and rigorous automation agenda risk falling behind their competition.

Over the next two years, a recent Bain & Co. survey of nearly 800 executives worldwide found, the share of companies scaling up automation technologies will at least double. The fallout from the coronavirus will likely accelerate adoption. These technologies include low code automation, optical character recognition (OCR), robotic process automation (RPA), and conversational artificial intelligence (AI).

But the path to realizing the benefits of automation has been more challenging than many executives anticipated. Nearly half (44%) of survey respondents, for instance, said their automation projects had not delivered the expected savings.

The benefits are clear, with labor time-savings leading the list. It either frees up employees to do higher-value work, or can be recouped as cost savings, or both. Survey respondents reported cost savings of roughly 20% on average over the past two years across processes in finance, human resources, supply chain and procurement, and service delivery and operations.

Microsoft’s finance group, for instance, has consolidated and simplified disparate reports, tools, and content in an automated, role-based personalized portal. The group also uses bots in financial operations, credit and collections, management reporting, and tax. As a result, Microsoft finance has reduced by 20% the time spent compiling and validating data, which saves over 150,000 hours of work every quarter. In addition, chatbots reduce support costs by 30%.

The median payback for labor-cost savings takes 13 to 18 months, the survey found. Fertile territory consists of high-volume, rules-based processes, such as scraping data from the web, copying and pasting data from one system to another, doing heavy data manipulations, or opening emails and attachments—as companies do in providing status updates to customers or documenting a loan.

Cost savings, moreover, is just the start. On the revenue side, automation can stem revenue leakage or bring revenue in the door faster.

Telecommunications carriers, for example, have many business customers that each has hundreds or thousands of routers throughout their locations. As routers break or get replaced by new models, or customers start or stop service, keeping track of the assets has proved difficult, and use of the assets does not always flow through to billing.

One major telecom in the Asia-Pacific region dealt with the issue by building scripts to identify products across all its customers, using analytics to flag the gaps, and deploying automation to reconcile and recover the leaked revenue. Automation also handled updating and billing for future assets. In this way, the carrier was able to bring in millions of dollars in previously lost revenue.

Companies have realized other benefits as well, notably improved process quality, accuracy, reduced cycle times, and improved compliance. All of these make for a better experience for customers.

In the oil and gas industry, companies long had to rely on manual inspection of pipelines for corrosion and leakage. Now companies can send pictures of a pipeline to an AI engine, compare the pictures with an existing database, and alert the right employees to possible problems. This has increased pipeline uptime, reduced manual inspections, improved safety, and helped pipeline crews prioritize the repair of the most important kinds of damage.

Dealing with execution barriers

Yet, significant barriers to realizing value from automation persist—and the major barriers all involve execution: competing business priorities, lack of necessary funding, and a lack of skills or centralized coordination. Addressing these execution barriers requires a clear-eyed self-appraisal of the organization. Our analysis of the survey responses, combined with experience working with companies to apply automation strategically, suggests three key principles to keep in mind to overcome these execution barriers.

First, ground automation in the customer’s experience.

Automation should further the broader redesign of the customer experience, in which a cross-functional team defines the desired future state of the experience, working backward on how to achieve it through changes to people, policies, processes, and systems.

Take billing of customers, for instance. Customers might not care if their provider reduces billing costs through automation; they want bills to be accurate, prompt, and easily reviewed. Automating billing needs to not only reduce costs but also deliver a billing experience that customers value.

Because automation is just one tool, it often works best in tandem with other tools and as part of a deliberate overall strategy. A global consumer products manufacturer mapped the end-to-end experience of customers, partners, and employees so that it could understand where automation would best play a role. Rather than letting the magnitude of cost savings solely inform initial automation, the company focused on the episodes that mattered most to these stakeholders. The executive team set guidance based on the big goals that automation and digital technologies could enable. The company even combined the IT department with the global business services organization to accelerate the adoption of automation.

Second, spend as much if not more time on what comes after automation.

Beyond developing, testing, deploying, and maintaining the technology, what really matters is achieving the expected business outcomes from the automation. In this respect, it’s similar to outsourcing. Beyond selecting the right partner, negotiating a competitive contract, and transitioning the work to the provider, the retained organization left behind must be sized correctly and fit for purpose, or else the savings from outsourcing won’t materialize.

The risk is even higher for automation, because much of the work automated consumes only a fraction of each employee’s time. Companies that fail to redesign the jobs impacted will not achieve their goals.

Third, treat automation as a major change to be actively managed from the start.

Part of strong C-level sponsorship involves effectively managing change with the business, so that automation remains a priority. The senior team must demonstrate how automation will change the experience of employees and customers for the better, painting this picture in detail before automation goes live. Without specificity on how the experience will improve, many employees naturally will resist.

The most effective change programs clearly communicate which employee populations will be affected by automation and specifically how. Open, direct, and early communications about the impact of automation, together with reskilling programs, help people start thinking about or acting on new career paths. In addition, some effective programs involve employees in identifying which points of a process are broken or inefficient.

A look in the mirror

Companies committed to a sustained, holistic approach to automation should answer a set of tough questions:

• Do we have genuine sponsorship by the senior executives, business unit leaders, and functional leaders, not just from shared services leaders?
• Is automation a visible and prominent part of the corporate strategy and operational plan?
• Are we thinking boldly enough with a vision of how to improve a process and the customer experience?
• Have we picked the right processes to automate?
• How will we persuade the business to embrace automation and change behavior?

These questions help build a framework for realizing the value that automation can generate. Without it—if a company simply sets up automation projects and lets them run—the efforts are bound to stall out.