.svg)
The AI strategy trap: Why most talent teams are building on quicksand
.svg)
.svg)
Most AI projects undertaken by talent acquisition teams fail. And, it’s not technology that is causing so many AI implementation failures.
We talked about this with Matt Staney, former talent acquisition leader at SAP, Glassdoor, and Twilio.
“Why do most TA AI projects fail?” he asks. “Simple. The data’s a mess. AI in hiring is often sold like a magic wand, but most TA teams aren’t ready for it. Years of legacy ATS spaghetti, siloed systems, incomplete profiles, and inconsistent feedback loops mean that the models don’t have enough clean, connected information to learn from. So what you get instead is glorified keyword matching dressed up with a chatbot and called ‘AI.’”
Keep reading to understand why AI technologies aren't scaling to production and how to avoid common talent acquisition adoption failures.
Why 80% of AI projects fail in talent acquisition
A variety of strategic and organizational failures are causing about 80% of recruiting-related AI projects to miss their goals, according to Harvard Business Review.
There are multiple reasons for these AI implementation failures. Checkr cites pressure to adopt AI quickly, budget constraints (18% cite budget limitations), and internal resistance to change (17% face internal pushback).
One of the biggest challenges can be summed up in two words: data quality. Futurism Technologies says that “70% of AI integration projects fail due to poor data quality, outdated infrastructures, and scalability issues.” Similarly, according to a McKinsey report, 70% of AI projects fail to meet their goals "due to issues with data quality and integration.”
Whatfix said it well: “Artificial intelligence feels magical at times, but the truth is that AI is far from supernatural — it requires quality data to facilitate something useful.”
Companies so often try to implement AI technologies before they have ensured their data is correct, secure, and unbiased.
Understanding common data strategy gaps
Here are some of the AI challenges that talent leaders and their teams are experiencing with AI implementations.
Misaligned use cases and poor communication
As Whatfix writes, “Teams are often misaligned on what problem they are solving with an AI initiative and why they are solving it in the first place.” Someone views a demo or hears about a demo, and leadership gets excited about it. They plunge into the new tool before they have fully decided it solves their challenges or before they've established the right data foundation.
Data quality issues that sabotage AI success
Many AI systems for talent acquisition have been built on poor data foundations. For example, these systems will use only one organization’s employees, or even its high-performing employees as training data. This can produce a poor quality data foundation that only exacerbates bias.
Pete Tiliakos, a leading talent technology analyst, recently asked an HR leader about her plans to adopt AI technology. She said her data was housed in an old on-premise PeopleSoft system. “We’re not ready,” she told him.
Tiliakos says that “it doesn’t matter what process you’re putting into AI — talent acquisition, payroll, learning, whatever, it’s really about the health and healthfulness of that data.”
Siloed systems and fragmented data
Psicosmart mentions a recent study by the Talent Board showing that “only 38% of organizations effectively use candidate data to enhance their recruitment strategies. This disparity highlights a critical juncture in the hiring journey — while data is abundant, the ability to integrate and leverage this information remains a complex puzzle that many companies are struggling to solve.”
McKinsey finds that “organizations that fully harness the power of data integration can increase their productivity by 20-25%.” At one technology company, data was siloed in a 12-year-old ATS and adjacent CRM, requiring manual processes and time-consuming analysis through external tools. Teams frequently restarted searches from scratch in external platforms rather than leveraging existing internal databases, leading to duplication and inefficiencies.
Findem automated many of the manual processes that slowed recruiters down at the tech company, giving the team more time to focus on actually engaging candidates. Sourcing Analytics Essentials dashboards provided instant insights into talent attributes like skills, diversity, and network connections, empowering recruiters to make informed decisions without switching systems. Recruiters ultimately got visibility into trusted data, paving the way to successful technology implementation.
Balancing the push to innovate with practical reality
AI challenges will probably sound familiar. They include the following:
Budget constraints vs. executive expectations
Every talent acquisition department, and HR department for that matter, seems to be tasked with doing more with less. Mass-apply technologies are resulting in hundreds and often thousands of job applications for open jobs, making it all the more challenging to find the best-fitting candidates for a job. And while jobs need to be filled with limited people and money, many corporate executives want to simultaneously transform into digital, AI-driven, efficient, and partially automated organizations as soon as possible.
Skills gap in data literacy among HR teams
Deloitte found that in 47% of organizations, a lack of data literacy among employees is a significant barrier to using data-driven insights. Deloitte says that “many organizations face resistance from employees who are comfortable with conventional recruiting methods.” The challenge is to create a data-driven culture where talent teams can use data to make major improvements to their hiring processes.
People-related challenges
Deloitte also notes that the “common narrative” is that “AI improves our productivity and well being by reducing our workload.” At the same, Deloitte points out that “77% of employees say AI has increased their workloads and decreased their productivity, and 61% say it will increase burnout.”
In readying their organizations for an AI implementation, leaders should help their CEO and other executives understand what is possible and realistic, and what is not. They should also have a plan for developing the skills necessary in the company for AI success. And they should never forget the importance of the candidate and employee experience.
Establishing a data-first approach to AI strategy
To avoid an AI implementation failure, here’s what successful organizations suggest.
Don’t buy a tool simply because it’s shiny and cool
The AI strategy has to be clear to the entire talent acquisition department and the broader company. Whatfix writes, “An AI initiative is like any other product development project: it must meet a user need, and all stakeholders have to understand what that means to implement something that provides value to users and helps your team meet its business goals.”
Sarah Smart, a consultant who previously worked as in house talent leader, experienced this firsthand. At one job, she arrived to find that the company had selected a supposed “AI tool” from a large, legacy HR technology company. The vendor had said the product would streamline their talent department and reduce the need for so many recruiters. Smart realized the technology was little more than keyword matching based on people’s past titles and education, was not AI, did not expand the talent pool, and didn’t work well.
“There was a fundamental lack of education on the buying side,” she said. “They lacked the understanding of AI to prove it wasn’t working."
Build your AI on a solid data foundation
Informatica says that AI implementations fail not because of “a lack of data, but a lack of AI-ready data.” But, “there is no magic bullet to deliver AI-ready data. It will not come from flipping any single switch. It is the result of a rock-solid data foundation that can handle current and future unknown workloads, use-cases and capabilities with ease and efficiency.”
Staney agrees. “If you want AI that actually improves quality of hire, retention, or DEI outcomes,” he says, “start by preparing your data. That means mapping your entire talent tech stack and consolidating where possible, auditing the consistency and completeness of your candidate and employee data, and most importantly, integrating post-hire outcomes into your feedback loop, because AI is only as good as what you feed it.”
You should reach a certain level of recruiting maturity before taking on an AI implementation. For example, you must be ready to ensure data privacy and data security. You should plan on consolidating your tech stack, not adding on AI software that often duplicates what other tools do, produces confusingly contradictory metrics, and complicates your talent acquisition processes. You should also outline the criteria for selecting AI technology. Know who will be in charge of areas such as training as well as ethics.
Take a data-first approach to planning
Make sure you’ve established what metrics, such as “leads” or “manager reviews,” are important to you at the outset. Staney says to “align on shared definitions: What is a ‘quality hire’ to your org? How do you define potential? Culture fit? DEI?”
Decide on realistic goals, such as moving 1 out of 3 people from the screening stage to the manager review stage. Set annual targets, quarterly targets, and then weekly benchmarks. Monitor them and adjust them regularly. Work on these kinds of goals before turning to your technology selection.
Avoid AI Implementation failures and set the stage for success
Building a strong data foundation, planning ahead, and avoiding buying tools only because they are the newest and “shiniest” will help you avoid AI implementation failures and position you for success.
“The truth is,” says Staney, “AI won’t fix your hiring. But it can scale what already works, if your data tells a clear story.”
Here is some additional information to help as you build a data foundation for AI success.
