IE 11 is not supported. For an optimal experience visit our site on another browser.

Why retailers might offer you a job before you even know you want one

From software that finds candidates before they even apply to chatbots and one-way video interviews, retailers are leaning in to AI to streamline their recruiting processes.
Illustration of a line of applicants holding resumes as digitized shadows fall behind them.
The labor market has reached a competitive fever pitch, pushing some companies to embrace artificial intelligence more quickly than before the pandemic.Eric Petersen for NBC News

As the retail industry continues to lose workers, companies are turning the table on traditional hiring practices. Instead of waiting for candidates to come to them via traditional job postings, retailers are leaning on artificial intelligence software to search out people who would be good fits for jobs — before candidates even consider applying.

The pandemic economy left retail companies with thin employee rolls as workers left the industry for higher pay, improved benefits and better working conditions elsewhere. More than 1 million of the 11 million jobs currently available in the U.S. are in the retail sector, according to the Federal Reserve Bank of St. Louis.

Usually automation happens when we see a bottleneck. Now the bottleneck is ‘I don’t have enough applicants.’

The number of job applicants in the sector fell by about 40 percent last year, said Al Smith, the chief technology officer of iCIMS, a recruiting platform used by companies such as Target, Dick’s Sporting Goods and Foot Locker. Last year, the company got an average of 23 job applications for one retail job on its platform. By the end of 2021, it received an average of 14 applicants. 

“The pandemic hit retail hard — and resignations hit hourly workers pretty hard,” Smith said.

For retailers, that means the labor market has reached a competitive fever pitch, pushing some companies to streamline their recruiting processes and embrace artificial intelligence at a faster pace, recruiters said. Before the pandemic, it would take days to hear back from a store manager to schedule an interview, said Kevin Parker, the CEO of the recruiting technology firm HireVue. Now, it could take minutes.

“It’s 24/7,” he said. “So that always-on recruiting and always-on hiring make a big difference.”

Many of the artificial intelligence tools — such as chatbots, interviews over text and instant video interviews — use natural language processing and machine learning to help speed up the early back-and-forth calls and messages that slow the recruiting process, when an applicant could spring to another retailer who responds more quickly. With an AI-enabled chatbot, the software uses natural language processing to analyze a candidate's text and respond with information about open roles or specific benefits, for example.

Oracle uses an advanced technique that uses natural language processing and machine learning to post job opportunities in front of potential candidates as they browse their social media pages, said Nagaraj Nadendla, the company's senior vice president of product development. It can also use automation to immediately send notifications to anyone who “likes” their company page.

"The AI here plays the role of adviser, making recommendations but leaving the final decisions up to the humans running the decision," Nadendla said in an email. "Because the system has built-in machine learning, it can analyze how those recommendations are used over time to make continually better ones."

Fetcher, an outbound recruiting platform startup founded in 2015, searches the internet for potential candidates who could be good fits for certain jobs — but who have not yet applied. While the technology has traditionally been used to fill open positions for engineers or finance analysts, in the last year it has increasingly been used to fill hourly retail and hospitality jobs, said Andres Blank, Fetcher’s CEO. 

“Usually automation happens when we see a bottleneck,” Blank said. “Now the bottleneck is ‘I don’t have enough applicants.’”

Some recruitment tools use natural language processing to determine whether candidates' email responses suggest they are interested in an opportunity.

The company’s software uses natural language processing to analyze and extract the most important keywords from a job description, such as the location and the necessary skills and years of experience. It then uses what are called relational algorithms to expand the keywords to include job titles and the skills that are similar to those sought by the retail company performing the search. For example, a store clerk and a cashier may serve similar roles but have different titles.

Machine learning-enabled tools then search and sort its database of potential job candidates built by crawling through LinkedIn and job recruiting profiles to find and rank the most relevant people. It then uses natural language processing to determine whether candidates' email responses suggest they are interested in the opportunity. 

In such tight competition for labor, retailers are more open to "onboarding" people who may have adjacent experience in call centers or customer service, Blank said.

They've also become more adept at poaching workers from competing companies, Blank said. Retailers often provide Fetcher with market analyses of their top competitors. The software then analyzes which companies are downsizing and identifies employees at those companies. It also analyzes how long workers may stay in particular positions to get a sense of whether they may be open to jumping ship. Within hours, Blank said, the company can compile a list of potential candidates.

While sales at Target, Walmart and Amazon have increased during the pandemic, sales at Dollar General, Bed Bath & Beyond, Rite Aid and other retailers decreased leading into the busy holiday season as the retail industry continued to battle widespread worker shortages. Some retail workers even struggled to get through their shifts amid understaffing or said they were expected to work just days after having tested positive for Covid.

One retailer trimmed the time it takes to screen and schedule job applicants to just seven minutes.

Over the last year, Carter’s Inc., Tractor Supply Co. and other retailers have used natural language processing-enabled chatbots to speed the recruiting and hiring process to fill their employee ranks. Carter’s, the children’s apparel retailer, trimmed the time it takes to screen and schedule job applicants for interviews to seven minutes using the recruiting chatbot software designed by Paradox AI, said Jennifer Frazer, Carter's senior director of talent acquisition. Melissa Kersey, Tractor Supply's executive vice president and chief human resources officer, said about 30 percent of applicants choose to interact with its chatbot to apply for positions. On average, candidates spend about 12 minutes applying for jobs — down from an average of 29 minutes, she said.

“We knew the war on talent had just kind of come out, so we knew we had to provide the right experience,” she said. 

As AI-enabled recruiting tools become more in demand, some lawmakers, academics and labor advocates have called for closer scrutiny of their use in recruiting and assessing job candidates.

Joseph Fuller, a professor at Harvard Business School, raised concerns in a research paper published in September about how AI tools are designed to rank candidates based on how closely their résumés or cover letters include terms in job descriptions, leaving out thousands of potential job candidates.

The concerns go beyond text. HireVue dropped its facial analysis of interviews in March 2020 after the Electronic Privacy Information Center filed a complaint against the company with the Federal Trade Commission alleging that the tool harmed candidates’ privacy. In November, the New York City Council passed a bill requiring employers and staffing agencies to conduct annual bias audits if they use AI tools to screen job candidates.

Relying on imperfect AI-enabled hiring tools like those means “employers artificially and significantly constrain the pool of people they consider,” Fuller said.

“Then they’re left with either very few candidates or candidates that might be borderline overqualified so they are more likely to turn over,” he said.