Headcount and headaches: a simple test for what to automate first

Short answer: before building any team-based AI automation, ask two questions: how much genuine pain is this team dealing with, and how many people are in it? Nick Harding, CEO of Fifty One Degrees, calls this "headcount and headaches": worth automating if a team has real pain and real headcount, not worth it if the team is one or two people, because you'll spend more time building the automation than it will ever save you.

The framework, in Nick's own words

This came up during a conversation between Nick Harding and Julian Tan of BT, when the topic turned from general AI use to specific process automation, customer service bots, workflow automation, anything more built-out than someone using ChatGPT for their own work. That's a different question with a different answer, and Nick's framework is how he decides.

"I have this quite simple framework that I sometimes apply, which I call headcount and headaches," he says. "Ultimately, if you have a team where you're experiencing a lot of headaches and that team also has material headcount, then it's worth considering whether you can do something there to try to relieve one or both of those things. But vice versa, if you've got a team of one or a team of two, you're probably going to spend more time building the AI automation than it's going to save you in time."

Julian Tan's own answer, just before Nick introduced this, sets up exactly why the test matters. Automation isn't free, even when it works: "There will definitely be specific use cases where AI will do it, and do it better. But there are other areas where... AI may create a lot more content that you then need to go through, so the productivity and efficiencies you'd see get offset." Nick's framework is a quick way to sort which situation you're actually in before you commit time to finding out the hard way.

Why this matters: two very different questions get conflated

A lot of small businesses treat "should we use AI" as one question. It isn't. There's a clear line between everyday, general AI use, someone on your team using Claude or ChatGPT to draft something faster, and building a dedicated automation for a specific process. Nick draws this distinction explicitly: "Generic AI platforms, using Claude, using ChatGPT, is a totally different ball game, everybody should be doing that, everybody could benefit from doing that. But when it comes down to team-based automations... the ROI point is so important."

General use is close to a no-brainer for everyone. Building a bespoke automation is a real investment decision, and it deserves a real test before you start.

Running the test on your own business

Pick a team, department, or recurring function in your business and ask the two questions honestly:

• Headaches. Is this genuinely a painful, repetitive, error-prone part of the business, not just mildly annoying, but a real drag on time or morale?
• Headcount. Is there enough volume of people doing this work that fixing it once pays off many times over?

If both answers are yes, it's worth exploring. If either answer is no, especially headcount, Nick's advice is blunt: the automation will likely cost you more in build time than it ever saves. That doesn't mean do nothing. It means the fix for a one- or two-person team is more likely to be a better use of general AI tools, not a bespoke build.

What to try this month

List every team or function in your business, and score each one honestly against headaches and headcount. Whichever team scores highest on both is your genuine first candidate for a real automation project, everything else is probably better served by general AI use, not a build.

To think through where automation actually makes sense for your specific business, the AI for Growth community gives free access to the AI Strategy Generator.