How to Tell If Your AI and Software Spending Is Paying Off

The problem: You can see exactly what you spent on software and AI tools last year, and nobody in the business can tell you what any of it returned.

The solution: Connect the spending to the work it was supposed to change, so the next investment decision is made on your own numbers instead of a feeling.

Your bookkeeper can tell you to the dollar what you spent on software last year. Ask what it returned and the room goes quiet. Somebody will say the note-taker is popular. Somebody else will say the proposal tool never really caught on. Nobody has a number, and nobody has a way to get one.

That is a strange position to be in, because you would never buy a truck or add a project manager this way. You would at least know what the last one did. With technology spending, the decision keeps getting made and the result never comes back. So the next renewal gets approved because cancelling feels risky, and the next tool gets bought because a competitor mentioned it. That is not investing. That is guessing with a purchase order attached.

You can see the cost. You have never seen the return.

The cost side is easy. It is a line in the accounting, a credit card charge, a subscription that renews. The return side is the problem, and it is a problem of assembly rather than of measurement.

A proposal tool is supposed to shorten the time between a request and a signed engagement. To know whether it did, you would need to know how long that took before, how long it takes now, and how many more of them closed. Those three facts live in three different places: an inbox, a spreadsheet somebody keeps, and the accounting system. Nobody has ever put them next to each other, so the tool gets judged on whether people say they like it.

This is why technology spending stays unexamined while everything else in the business eventually gets a number on it. Not because owners do not care about the return, but because getting the return would take a week of somebody's time, and by the time they finished, the renewal would already have gone through.

What it takes to know

The answer comes from connecting the spending to the work it was meant to change. Not a report about the tool, but a line from what you paid to what actually happened afterward in hours, in jobs, in invoices, in closed work.

That means the things your business runs on need to be joined up: your clients, the work you do for them, the people on it, the hours logged, and what got billed. Once those sit in one connected picture rather than in separate tools that never speak, the question stops being a research project. You can ask what the proposal work cost before and after, and see it.

Keeping that picture current is where automation earns its place. Nobody has time to rebuild the comparison by hand every quarter, and a number that takes a week to produce will never be produced. When the routine collecting and matching happens on its own, the return on a tool is simply visible, the way revenue is visible, and your people stay on client work instead of assembling evidence about software.

Where the next dollar should go

The point of measuring the last investment is not to feel better about it. It is that the next decision is usually bigger.

An owner at this size is choosing between a $95k hire, a $40k tool, and a marketing spend, and the honest position is that all three feel plausible. With the work visible, those choices stop being equally plausible. You can see that the hours going into proposals are worth more than the hours going into scheduling, so you know which one to spend against. You can see that a tool nobody adopted cost you nothing but the license, while a process nobody automated is costing you real hours every week.

That is the shift. Not better tools, but the ability to tell which spending bought something and which only felt like progress.

A look at a professional services firm

Take a professional services firm doing about $8 million a year, with around thirty-five people billing their time.

They were early and enthusiastic. A writing assistant, a note-taker, a pilot to draft proposals, a scheduling add-on. Roughly $40k went out across a year in licenses and the internal time to set them up. Two years on, the owner cannot say whether any of it changed a single number in the business. The proposal tool could not see a client's history, the note-taker did not know which engagement it belonged to, and nothing connected to what got billed. The tools were not obviously bad. They were simply unjudgeable.

Now suppose the firm connected the picture first: clients, engagements, the people on them, hours logged, and what was invoiced, joined up with a clear owner for each piece, using the tools they already had rather than replacing them. Then suppose agents went back onto the repetitive parts, drafting from real client history, matching time to the right engagement, and flagging unbilled work before it aged out.

What the firm would gain is not mainly the recovered hours, though you would expect those. It is that the next $40k would be checkable. They would be able to see what a category of work cost before and after, and decide the following year's spending on that rather than on enthusiasm. Set against the pattern they are in now, where $40k a year of tool spending sits beside a $95k hiring decision and a renewal cycle nobody can evaluate, five years of that adds to close to a million dollars committed without a single look back. Being able to look back is the return.

How to start

1. List what you spent last year. Every tool, every license, every pilot, with the internal time it took to set up. Most owners are surprised by the total.
2. Name what each one was supposed to change. Faster proposals, fewer unbilled hours, less re-entry. If you cannot name it, that is the finding.
3. Connect the things that would show it. Join your clients, work, hours, and billing into one picture, so the before and after can actually be compared.
4. Automate the upkeep, then judge one decision. Let the collecting and matching run on its own, then take a single tool and say plainly whether it earned its cost. Do the next one the same way.

The takeaway

The reason you cannot say whether your last technology spend paid off is not that you failed to track it carefully enough. It is that the evidence sits in pieces that were never joined, and assembling it by hand costs more than the answer is worth. Connect the work into one picture, let the routine collecting run on its own, and the return stops being an opinion. Then the next decision, the hire or the truck or the tool, gets made on your own numbers, and being wrong costs you once instead of every year.