One corporate job listing attracts 250 applicants, on average. This high application volume can be overwhelming for recruiters, and leaves them with only 7.4 seconds to review each resume they see. In a high-unemployment climate like today’s, this problem is exacerbated. Recruiting requires heavy lifting, and there’s no denying that various forms of AI (such as machine learning, a subset of AI) and automation are playing much bigger roles these days for hiring teams. AI can help bring exponential efficiency and equity to strengthen the process if wielded in the right way. Here's what to consider before implementing AI and automation in recruiting.

Best Practices For Making The Decision To Implement An AI Recruiting Platform

The first thing you need to ask yourself is, “What problem am I trying to solve?” Perhaps you are trying to lower costs, improve your candidate experience or reduce bias in the hiring process. Write down your answers, and then tie them to quantitative goals so you can measure the success of AI in your recruiting process after implementation.

Use the same mental model you’d use for when you’re hiring a new (human) teammate: What are the input and output goals for the role? You may even want to write a rough job description for what you want your new technology to solve for. After asking the right questions, you may decide to hire an additional recruiter, continue business as usual or use basic automation, or you may determine that AI is the right solution for your team.

What To Look For In An AI Recruiting Platform

Every business has its own recruiting needs, so here are five things to look for in an AI recruiting platform.

Integrations: It’s imperative that the solution you purchase fits seamlessly into your existing workflow and even maximizes the impact of other tool investments in your ecosystem. Dive deep with vendors to understand how their integrations work and what implementation and updates look like. Often vendors will say, “We integrate with XYZ tool,” but the devil is in the details.

Conversational design with bots: I’ve spoken to many companies that have implemented bots to help engage and screen candidates, configured them with an initial set of questions and then not gotten the promised results. In today’s world, it’s very easy to implement a bot, but a bot alone is just a shell. What’s important is the content, data and transactions delivered through the bot and, for machine learning solutions, how the bot learns over time. Not getting this right will leave you with the difference between a seasoned recruiter asking the right questions at the right time to the right candidate, and someone who has been asked to interview a candidate but has never spoken to one in their life.

When vendors promise a bot, it’s also important to understand who will help with conversational design and how exactly the bot’s knowledge base, ability to understand candidate intent and responses evolve over time.

Candidate experience: With cost savings at play, it’s easy to jump to a solution. If you’re looking at a solution that directly touches candidates, ask vendors if they can provide you with data around their candidate experience scores and accuracy numbers associated with their algorithms. Virgin Media, for example, discovered that bad candidate experiences cost their company $5 million a year in rejected candidates switching to competitors. If implemented well, AI can ensure no candidate goes ignored and their questions are answered quickly. It can also be used to keep candidate pipelines engaged so silver medalists are on deck when the next opportunity arises.

Timeline and costs: How much time are you willing to allocate to setting up the platform, and how much are you willing to pay? These two factors are often underestimated, particularly when it comes to AI solutions, which require large sets of training data to tune their algorithms. Remember that not all platforms come ready to use out of the box, so it takes time and resources to get these platforms fully tailored to your needs.

Data ownership: This isn’t always top of mind, but in today’s age, where data is invaluable, make sure you read the terms of use for the tools you're considering and understand your data rights. Where is the data stored? What are the security features that ensure that data remains private? What happens to your data if you are no longer using the service?

Potential Pitfalls

Some AI platforms require extensive work, while others don’t, and it’s important to know which are which and what you get from each. Some solutions are lightweight and easy to implement but come at the cost of being limited in their functionality. If you’re willing to put in additional time and effort in allowing machine learning algorithms to be trained with data over time, you can reap deeper benefits, for example.

One way to accomplish all these exercises is to whiteboard your current candidate journey. See where AI or automation falls into the different steps in the journey — which tasks you would always want backed by a human, and which tasks you would prefer to automate or use AI for. Then map out what a candidate journey would look like if you were to implement an AI recruiting platform. Once you’ve done that, you’ll get a more concrete picture of whether you need an AI solution or even a tech solution at all.

AI in recruiting is there to help streamline the process and take administrative tasks off the shoulders of recruiters. However, before deciding to implement an AI recruiting platform, consider what business problem you’re trying to solve and how this technology would fit into your long-term business strategy. Sometimes, all it takes is to take a step back and a map of your candidate journey to fully understand whether you truly need to implement an AI recruiting platform.

Start with designing your candidate journey. If you need help, please set-up time with me to do a virtual white board session.

[Original article first appeared on Forbes.com on June 9, 2020]