AI vs Hiring: The Real Math for Small Professional Services Firms

The moment most small firms are about to hire is exactly the moment they should pause.

Not to cut costs. Not because hiring is bad. Because a £30K to £50K hire is the single most expensive decision most small firms make in any given quarter, and almost nobody runs the maths before they post the role.

A coordinator, paralegal, SDR, junior associate, operations assistant, or client service lead is not just the salary. It is the salary plus onboarding plus management load plus software seats plus the two or three months of output-below-break-even at the start. For a five to fifty person firm, a single bad hire can swallow a quarter of forward pipeline.

And most of these roles, inside most small firms, are now 70 to 85 percent pattern work. Not all of it. Not the judgement calls. Not the relationship bits. But the pattern parts, where someone is shuffling information between systems, replying to templated enquiries, chasing documents, scheduling, formatting, or following up, are now pattern work that systems can handle.

This is not a pitch for firing anyone. It is a pitch for stopping before you hire, running the filter below, and only then deciding whether the next hire is actually a person or actually a system.

The Hire You Are About To Post Is Probably 70 to 85 Percent Pattern Work

Look at the last three job descriptions your firm wrote. Pull them up on a screen.

Read each bullet under "responsibilities." For each one, ask a single question. Does this bullet describe a judgement, a relationship, or a pattern?

Judgement bullets sound like "decides which matters to accept" or "negotiates final fee with client" or "provides legal advice on complex cases." These are irreducibly human. You need a person.

Relationship bullets sound like "builds long-term relationships with key clients" or "represents the firm at industry events" or "mentors junior team members." These are also irreducibly human, for now.

Pattern bullets sound like "responds to incoming enquiries," "updates CRM records," "schedules calls," "prepares client reports," "chases outstanding documents," "formats proposals," "sends follow-up emails," "creates status updates," "onboards new clients," or "maintains the pipeline tracker."

Every pattern bullet in the job description is a candidate for automation, an AI agent, or a workflow redesign. Inside the small-firm roles I look at most often, the pattern bullets are between 70 and 85 percent of the listed responsibilities.

This does not mean you do not need a person. It means the maths of the decision is different than you think.

Before you post the role, work out the pattern-versus-judgement split. If the role is 80 percent pattern, you are not hiring a person. You are hiring a pattern-processor with some judgement attached. The pattern part can usually be built for a fraction of a year's salary, leaving the judgement work for a part-time person, an existing team member, or a smaller, more senior hire.

The Real Math, With Real Numbers

Run the numbers on a coordinator role. A £35K base salary in the UK is roughly £45K fully loaded once you add on national insurance, pension, software seats, desk, training, and management overhead.

That is year one.

If the role is 80 percent pattern work and 20 percent judgement, you are paying £36K per year for the pattern work and £9K per year for the judgement.

A workflow build that replaces the pattern work, properly scoped, for a small professional services firm, typically costs £15K to £30K as a one-off build. It then costs roughly £2K to £5K a year in software and maintenance.

Year one maths. You spend £45K on a person, or £25K on a build plus £3K ongoing. Saving is about £17K in year one.

Year two. The person costs £45K again, plus whatever rise you give them. The build costs £3K to £5K. The saving is £40K a year, recurring.

Year three. Saving is £40K plus raises.

But that is the boring version of the maths. The more interesting version is the capacity question.

A person handles roughly 1x of the volume you hired them for. A properly built pattern-processor handles 3x to 10x of the volume at the same cost. So the real calculation is not "salary versus build." It is "does this role scale with headcount, or does the system let me handle three times the throughput without hiring again for another eighteen months."

For most pattern-heavy roles, the build scales and the hire does not.

This is the bit firms miss when they frame it as AI vs hiring. The hire gets you linear capacity at a recurring cost. The build gets you multiplied capacity at a one-time cost. Over three years the comparison is not close.

The 3-Question Hiring Filter Before You Post The Job

Before you approve any hire, run the role through these three questions. This is the filter I walk firms through before they fund a headcount slot.

Question 1. What percentage of the role is pattern work?

Go line by line through the job description. Mark each bullet as judgement, relationship, or pattern. Count them honestly.

If pattern is under 30 percent, hire the person. You need a human with judgement.

If pattern is between 30 and 60 percent, hire the person but scope a workflow build alongside them. The first six weeks they are in the job, build the pattern layer out from under them. They will thank you. Nobody likes doing the pattern work. The judgement work is what people actually want to spend their day on.

If pattern is over 60 percent, do not hire. Build first. Then decide if you still need the hire, and if so, at what level.

Most coordinator, SDR, intake, and operations roles inside small professional services firms score in the 70 to 85 percent range. Most junior associate and paralegal roles score 50 to 70 percent. Most senior and partner-level roles score under 30 percent.

Question 2. Does the role live inside a workflow you have mapped?

You cannot automate a workflow that does not exist. If the role is being created because "things are chaotic and we need someone to own it," the role will fail, and so will any automation.

If the workflow is mapped, you can decide where the human goes and where the system goes. If the workflow is not mapped, the role turns into a general-purpose human sponge that absorbs everything nobody else wants to do. That is a terrible role for the person, and a terrible spend for the firm.

If your answer to this question is "no," the honest next step is not the hire. It is the workflow audit. Then the hire question answers itself.

Question 3. What happens to this role if your volume doubles in 18 months?

Most small professional services firms are growing. That is the reason you are hiring.

If the role scales linearly with volume, you will be in this same meeting in eighteen months, hiring another one. Then another. Each hire comes with onboarding, management, and turnover risk. By the time you are on your third coordinator, you are running a small department whose entire existence is pattern work.

If the role is backed by a system, doubled volume is a software line item, not another hire. The system takes the volume. The judgement-heavy human you did hire is still the one human in the loop, which is exactly where you wanted them.

This is the question that changes the shape of firms. Firms that answer it early scale to forty people with ten headcount. Firms that do not answer it, scale to forty people with thirty headcount and wonder why margin is thin.

What AI Actually Replaces, And What It Does Not

AI does not replace people. That framing is wrong, and it leads to bad decisions in both directions.

AI replaces the pattern parts of roles. It replaces the drafting, the chasing, the formatting, the routing, the summarising, the triaging, the scheduling, the extracting, and the updating. When it is built properly and wired into the firm's real workflows, it handles those parts at 3x to 10x the volume a person would, at a fraction of the cost.

AI does not replace the judgement parts. It does not decide whether to take a case. It does not negotiate with a difficult client. It does not navigate a partner disagreement. It does not build a relationship. It does not notice the thing that is slightly off in a conversation that makes a deal different from how it looked.

Which means the right shape of a small firm in the next five years is fewer pattern-heavy hires and more judgement-heavy ones. Fewer coordinators, more senior associates. Fewer SDRs, more closers. Fewer junior paralegals, more mid-level ones with better tools underneath them.

The firm ends up smaller, more senior, more profitable, and more scalable. That is the direction the maths points. Most firms have not done the maths yet.

The Hiring-Signal Moment Worth Paying Attention To

Inside the 5 to 50 person firms I work with, there is a specific moment worth paying attention to.

It is the moment a managing partner is about to sign off a £35K to £50K admin, coordinator, or operations hire. Sometimes it shows up as a LinkedIn job post. Sometimes as a brief to a recruiter. Sometimes as a slot in the next budget.

That is the moment to pause for ninety minutes and run the filter above.

In most cases, the answer is not "do not hire." The answer is "hire, but also build." Or sometimes, "build first, then decide."

Firms that pause and run the filter save between 30 and 70 percent of the fully loaded cost of the next three hires, without losing output. Firms that skip the filter and hire reflexively end up with departments full of pattern-processors and then wonder why they cannot get their margins to move.

The difference between those two firms, three years out, is not a rounding error. It is the difference between a profitable practice and a fragile one.

Before You Post The Role

If you are about to post a role, or just approved one, spend the next hour doing this.

Pull up the job description. Count the pattern bullets. Work out the three-year maths, not the one-year maths. Ask whether the workflow around this role exists on paper or only in someone's head.

If the answer is uncomfortable, do not post the role yet. That discomfort is the data.

The hire might still be right. But the version of the hire that is right is probably smaller, more senior, and sitting on top of a system you have not built yet.

AI is not here to replace your team. It is here to make your team smaller, more senior, and more leveraged. The firms that figure this out first will hire less and deliver more. The rest will keep scaling by headcount and wondering why it never feels like the firm is getting easier to run.