Is it cheaper to automate than to hire?

Usually, for the specific slice of work that is repetitive and rule-shaped, because automation has a build cost and then a very low running cost. It is not cheaper for work that varies, needs judgement or requires someone to be accountable to a client. Most firms get the best result by hiring for the judgement and automating the admin that surrounds it.

When should we buy software instead of building an automation?

When the problem is common, a vendor already solves it properly, and your version of the process is not a competitive advantage. Practice management, accounting, e-signature and scheduling are all cheaper to rent than to recreate. Build only the connections between those tools, which is where off-the-shelf products tend to stop.

What are the signs a process should not be automated yet?

It runs only a few times a month, it changes every quarter, nobody can write down the steps without arguing, or the whole thing exists because a decision was never made. Automating any of those locks in the mess and makes it run faster. Document or fix the process first, then reassess.

Can automation replace an administrator?

Rarely, and firms that aim for that are usually disappointed. What it reliably does is remove the repetitive slice of an administrator's week, which either gives you capacity you already needed or delays the next hire. Treat it as buying back hours, not as replacing a person.

Automate, buy software, or hire someone?

What actually decides this?

Volume, stability and judgement, in that order. Everything else is detail.

Volume sets whether the effort is worth it at all. Something that happens forty times a week is worth an investment that something happening four times a year is not. Be honest about the count rather than the annoyance level, because the two are unrelated. Rare tasks feel enormous precisely because they are unfamiliar.

Stability sets whether the investment survives. A process that has changed twice this year will change again, and an automation built on it becomes maintenance work rather than saved time. Stability is worth more than volume here, because an unstable process quietly converts a one-off build cost into a permanent one.

Judgement sets the ceiling. Work where the right answer depends on knowing the client, reading a room, or accepting professional responsibility does not become mechanical because you put a model in front of it. You can automate the preparation and the follow-up around a judgement call. The call itself is a person's job.

When is buying software the right answer?

When the problem is common, a vendor maintains a good answer to it, and your version of the process is not a competitive advantage. Practice management, accounting, document signing, scheduling, payroll: rent these. The vendor amortises development across thousands of firms and you never will.

The test is whether a client would notice or care that you do this differently to your competitors. If not, it is plumbing, and plumbing should be bought.

What you give up is fit. Off-the-shelf software makes you adopt someone else's shape of the process, and that is usually the right trade, because their shape has been refined by more firms than yours. Where it stops being the right trade is when you find yourself paying for a product and then paying a person to work around it every day.

The gap that buying leaves is almost always the same one: the connections between the products you bought. Vendors build good tools and mediocre bridges. That gap is where automation earns its place, not in recreating the tools themselves.

When is hiring the right answer?

When the work needs judgement, when it varies constantly, when someone must be accountable to a client, or when the volume has grown into a genuine role rather than a set of tasks. Also when you simply need capacity now and cannot wait for a build.

Hiring is the flexible option and the expensive one. A person absorbs exceptions, notices when something is wrong, handles the case nobody anticipated and improves with experience. No automation does any of those things. The cost is fixed, ongoing, and comes with on-costs and management time that firms routinely underestimate.

The failure pattern is hiring to absorb a mess. If the new person spends their first six months doing repetitive work that a workflow would handle, you have converted a fixable process problem into a permanent salary. Sort the repetitive slice first, then hire for what is left, and the role you write will be a better one.

When is automation genuinely the right answer?

When the work is frequent, the process is stable and written down, the inputs are already digital, and the consequences of an error are either low or catchable by a human review step.

The workflows that pay off in professional firms are consistently unglamorous:

An enquiry arrives and gets logged, categorised and a draft reply prepared for approval.
A quote is accepted, so the invoice is raised and the payment follow-up is scheduled.
A matter closes, so the review request goes out and any unhappy reply is routed to a partner first.
Friday reporting assembles itself from the systems that already hold the data.
A shared inbox is triaged and routed instead of read by three people.

Each of those is high frequency, low judgement, and sits between two products you already pay for. That is the shape to look for.

When is automation the wrong answer?

More often than the market admits. These are the situations where the honest recommendation is to spend the money elsewhere:

Low frequency. A monthly or annual task rarely earns back a build, no matter how tedious it is.
An unstable process. If it has changed twice this year, wait until it settles.
Nobody agrees how it works. Three people describing the process differently is not a technical problem.
The task should be deleted. A surprising share of firm admin exists because a report nobody reads was requested in 2019.
High consequence, no reviewer. If the output goes straight to a client or moves money and nobody will check it, do not build it that way.
It is really a decision, not a process. Automation cannot resolve a partner disagreement about who owns intake.
One person's undocumented expertise. If the knowledge lives in someone's head, capture it first. That is a documentation project, and it is worth doing on its own merits.

How do the three compare?

Criterion	Automate	Buy software	Hire
Best when	Frequent, stable, mechanical	The problem is common and solved	Judgement, variety, accountability
Cost shape	One-off build, low running cost	Ongoing subscription per seat	Fixed ongoing, plus on-costs and management
Time to value	Weeks	Days, then a slower adoption period	Months, including notice and ramp-up
Handles exceptions	Poorly, by design	Only the ones the vendor anticipated	Well
Improves over time	No, unless you invest again	Yes, at the vendor's pace	Yes
Main risk	Built on an unstable process	Poor fit and switching cost	Hiring for a mess instead of a role

What order should you work through this?

Run the options in cost order, cheapest first, and stop at the first one that holds.

Delete it. Ask what happens if this stops. Do it for the reports first.
Simplify it. Remove the approval steps that exist out of habit rather than risk.
Buy it, if a vendor already solves it properly.
Automate the joins between the things you bought, starting with one workflow rather than five.
Hire for the judgement, the relationships and the exceptions that remain.

Most firms do this list backwards, which is why they end up with a capable person spending Thursday copying data between two systems.

What to do next

Take the three things currently annoying you most and score each one on frequency, stability and judgement. The scores will usually sort themselves into obvious answers, and the awkward one in the middle is normally a process that needs writing down before anything else happens.

Then pick a single option and do it properly rather than doing all three partially. If the answer turns out to be automation, scope one workflow, insist on a fixed price, and check the payback arithmetic before you sign. If the honest answer is that you should hire, or buy, or simply delete a report, that is a perfectly good outcome from a fit call and worth hearing before you spend anything.