A hiring manager gets 12 automated updates from the ATS in a single day. They ignore all of them, and by the end of the week, they’ve gone off-system and hired someone through their network.
This is happening in your business right now. It might feel like a behavior problem, but its root cause is actually poor design.
Most TA tech transformations don’t fail because of the platform. They fail because no one is clear on the intention behind the changes.
I’m not talking about the usual slide deck lines about “efficiency” or “candidate experience.” The real question is: what are you trying to accomplish?
Because in reality:
The system tries to do all of it, and ends up doing none of it particularly well.
A recruiter opens a new req.
The intake form is half-complete, a bit messy, and no one’s really agreed what “good” looks like for the hire because nobody updated the knowledge articles.
But the search starts anyway, as it always does.
At the same time:
Individually these are manageable, but together, they are a real problem.
So people do what people always do:
I’ve seen this firsthand in a healthcare ATS implementation. About a year after go-live, the team still was not seeing the efficiency gains it expected. Data quality was poor, reporting was unreliable, and recruiters had developed too many different ways of using the system.
Amongst other things, the data coming out of the system was wonky at best and the team, while having embraced the system, were not really advocating for or pleased with how things were.
After spending dozens of hours interviewing recruiters, I realized that the root problems were straightforward: there was no clear process definition and no real governance around system use. So the recruiters were creating dozens of workarounds.
Once we introduced clearer best practices, process guidance, and training, usage became more consistent.
With this cornerstone in place, we were able to start analyzing a cleaner data set which in turn allowed us to better drive adoption and continual improvement through listening groups with the users.
Within a few months, adoption improved significantly, data quality became more reliable, and the team moved from inconsistent usage to a much more standardized operating model.
Meanwhile, the people actually doing the hiring are just trying to get through the day. A recruiter logs in and half the fields do not match how they actually assess candidates, so they ignore them. A hiring manager gets asked to give structured feedback in a format that does not reflect how they think, so they send a message instead.
The process may still look intact on paper, but the data becomes unreliable. Workflows get bypassed. And everyone still pretends that the system is still working.
If you’re lucky, that breakdown is slow and quiet. More often, it is not. It turns into that moment in Shrek — angry villagers, pitchforks out, everyone blaming everyone else, and no one entirely sure how it got that bad.
I walked into this exact scenario recently after the rollout of a new global ATS and hiring manager portal.
I arrived after a fantastic project team had recently launched Avature as a new global ATS. The project launch had been carefully thought through and well executed, but it became clear after going live that the Hiring Manager community in particular was struggling with adopting the new Hiring Manager Portal.
The problem was not that the technology lacked capability. It was that the people expected to use it could not yet see how it helped them. Many of the hiring managers were happy with the previous system, so the change already faced a higher bar. The new portal offered better functionality, but from their perspective it felt more complicated, not more useful.
So the first step was not more training. It was simplification.
We made design concessions, simplified workflows quickly, and stayed in close contact with the hiring manager community to understand where the friction was. As the experience became clearer and more usable, adoption began to improve. From there, we kept listening and refining, adjusting workflows and language as we learned more about what was and was not landing.
A second issue was the relationship between the client and Avature teams, which had started to strain under the pressure and made recovery harder. Part of the work became rebuilding a more functional partnership, gathering feedback from both sides and creating a more stable way of working together.
Through a combination of alignment, adoption, and alliance, the portal and its user community moved into a much healthier operating model. What started as a serious bump in the road ended up creating a stronger foundation for how the team worked going forward.
Right now, most organizations are accelerating into AI and there’s real pressure from the top to move fast — to adapt rapidly, to show progress, to not be seen as falling behind.
When leaders push teams to adopt AI quickly, most organizations respond the only way they can: they layer it onto whatever is already there. Faster job descriptions. Faster screening. Faster updates. Faster everything.
But speed doesn’t fix a bad process. It just exposes the cracks more quickly.
I saw this with one team that introduced an AI job description tool. At first, recruiters loved it. It saved time, reduced pressure, and made it easier to move quickly when hiring volumes were high.
Then the downside started to show.
Without clear guardrails and consistent human review, the content became flatter and less distinctive over time. Roles started sounding too similar. Important differences between teams got lost. Recruiters started relying on the tool too heavily instead of really thinking about the story they were trying to tell in the market.
Give a stretched recruiting team a tool that promises speed, and of course people will lean on it. But if the underlying process is already unclear, AI does not solve that problem. It scales it. What starts as efficiency can turn into generic messaging, weaker market positioning, and less disciplined decision-making.
Eventually, the team had to move back towards a much more blended approach: using AI to accelerate the process, but not outsourcing judgment, positioning, and critical thinking altogether.
The technology was not the problem. The problem was dropping it into a process that was not ready to use it well.
You are never adding a tool into a clean environment. You are adding it into existing systems, habits, workarounds, and politics. That is why so many transformations struggle. If the process does not fit how recruiters and hiring managers actually work, people will go around it.
That gets even harder on a global scale. A global process may look tidy on a slide, but hiring does not work the same way across markets. Standardization only works when there is room for local reality.
The transformations that succeed usually get three things right: alignment on what the process is meant to do, adoption by the people expected to use it, and alliance between TA, tech, and the business.
Before adding more AI, ask a simpler question: does this actually work for the people using it? If not, you are not transforming anything. You are just scaling the problem.
