How I Got to 20 Percent

1. The seven years before I ever looked at a stock chart.

I did not come to this from outside the work. I have been a litigator since 2017, and nearly all of it has been injury cases.

My first two years were spent on the other side of them, doing insurance defense and then medical malpractice defense. I would not give those years back. Defending a file teaches you what a file costs in a way that bringing one never does, and cost is most of this document.

I also billed by the hour, and I was not built for it. I am not going to make a moral argument out of that, because a great many excellent lawyers bill by the hour and do it honestly. It is a question of what the unit measures. An hour measures how long I sat at a desk. It does not measure whether I was any good at what I did there, and it does not measure whether the person I did it for ended up better off. I spent two years being paid in a unit that had nothing to do with the outcome, and I was aware of it more or less constantly.

When I moved to the plaintiff side, the unit changed. The firm got paid when the client got paid and not otherwise. I was salaried, paid every two weeks, with a bonus structure on top of that, so the timing landed on the firm rather than on me. What changed for me was that nobody was counting my hours anymore. I do not think I understood at the time how much of my week-to-week life ran on that one structural fact. I noticed it mostly as relief.

From 2019 until I opened this firm I did nothing but plaintiff personal injury, first at Schiller Law Offices and then at Stewart & Stewart. I carried roughly a hundred active files at a time. I took and defended more than 250 depositions, including experts. I tried three cases to a jury, two of them as first chair. Every one of those files was litigated. I have never once carried a pre-suit caseload, which means nothing to a client and a great deal to a lawyer reading this. I have also been sole counsel on the briefs for the appellant in two appeals to the Indiana Court of Appeals: one a published opinion on a question of first impression under Indiana's Child Wrongful Death Statute, which was affirmed, and one that was reversed and remanded. All of it sits on public Indiana dockets with my bar number on it, so none of it requires taking my word.

Somewhere along the line the work became second nature. It stopped being something I reasoned my way through. I could read a file and know roughly what it was worth, what it would cost to work up, where it would fight me and where it would fold. I am not going to sit here and tell you I am better at this than the people reading it. I will tell you that after seven years of doing nothing else, I stopped being surprised.

I also liked it, and I want that on the record, because a document like this can read as though I left somewhere angry. I did not. Both firms were good places to learn how to actually try a case, I was trusted with real files early, and I represented people I still think about. Nothing in here is a complaint about either one.

What changed is that at some point I stopped only seeing the cases and started seeing the work behind them. How a file actually gets produced. What it costs to move one from the first phone call to a check. Which of those costs are the case itself, and which are just the way firms have always been assembled. Once I could see that, I wanted my own.

Wanting one and believing I could actually do it were two different things. I assumed I would be about 40 before I did anything about it. That was never a plan. It was an assumption I carried for years without ever taking it out and looking at it, because every version of opening a firm that I could actually picture required a kind of certainty I did not have. That is where I was stuck.

In early 2026 I started to think I might be able to open a personal injury firm sooner than that. The tools available now let one lawyer carry work that used to need a floor of salaried people underneath it. That is what makes a 20 percent fee possible. It is not generosity and it is not a loss leader. It is arithmetic that only recently started working.

What moved me off that assumption did not come from law at all.

2. Fall 2024, and the nine months of studying.

In the fall of 2024 I started studying the stock market. Not trading it. Studying it.

I did that for nine months.

I know how that sounds in trading, where the flex is moving fast. But it is one of the most important facts in this document, so I want to sit on it for a second. Nine months is a long time to read about something without doing it. It is long enough to get bored, long enough to watch other people make money while you are still reading, and long enough that several times I decided I had learned enough and then made myself keep studying.

What I was actually doing during those nine months was learning what a price is.

Not what a stock is worth. What a price is. A price is a piece of information produced by a lot of people acting on what they know, and the interesting question is never whether a price is high or low. The interesting question is whether the price is carrying information at all, or whether it is just sitting where it has always sat because nobody has had a reason to move it.

That distinction is the whole rest of this document.

3. What I was looking for, and did not find.

Once I started trading, and then when I started building, what I was hunting was an inefficiency. A place where the price was not doing its job. Something the market had not priced correctly, that I could see and act on before it corrected.

I traded it myself first, with real money, for most of a year. I lost. That is the part I would leave out if I were selling you something.

At the beginning of 2026 I tried to automate the hunt. I built autonomous trading bots, several generations of them, and I ran them for months.

Here is how that went.

One hundred and two closed trades. A 19.6 percent win rate. An average result of negative 12.2 percent per trade. Zero profitable weeks. Not one. Every version I built finished negative, and the more sophisticated versions did not finish less negative than the simple ones, which is its own kind of lesson.

I am putting those numbers up front because everything after this is easier to dismiss if I open with a win.

The one thing that worked was a rule I wrote before I built anything, back when I had no idea whether any of it would function. No real money goes in until the system produces three consecutive profitable weeks on paper. Three. Consecutive. It never produced one. So it never got a dollar. What the bots cost me was a number in simulated accounts and a great many evenings. The money I actually lost, I lost by hand, the year before. That is where the rule came from.

That gate is the only reason this is a story about a law firm rather than a hole I dug.

And the honest conclusion from nine months of study and months of building is this: the equities market is extremely efficient. Thousands of very smart, very well-capitalized people are looking for the same thing I was looking for, with better information and faster systems. I did not find an inefficiency that survived contact with reality. The market was smarter than me, which is roughly what the market is supposed to be.

What I came out of it with was not a trading edge. It was a habit of mind: when you look at a price, ask what work it is doing.

4. The inefficiency I did find.

Then I looked at the market I had been working in for nine years.

Some background so you know I am not a tourist. Nine years practicing. More than 550 litigation files, and I mean litigation: filed cases, not intake counts. More than 250 depositions taken and defended. Cases in state courts across Indiana and in federal court in both the Southern and Northern Districts. I have tried cases to a jury, which teaches you more per hour about what a file is actually worth than anything else available.

In all of that time, the price never moved.

A case settles in six weeks on a demand letter. A case grinds through two years of motion practice, three depositions of treating physicians, a fight over a biomechanics expert, and a trial setting. Same firm, same year. The fee is a third of whatever comes in, and on the six-week case it is often a third of a much larger number, because the size of the policy has very little to do with the difficulty of the work.

I want to be careful here, because this is exactly where this argument usually goes wrong and makes people angry for no reason. Nobody is doing anything improper. The prevailing one-third fee is inside the rules, it has been the market rate for decades, and the lawyers charging it are, in my experience, working extremely hard for it. I am not saying anyone is overcharging. I am saying the number does not move with the work, and everyone in the profession knows it does not.

Go back to the question from section two. Is that price carrying information, or is it sitting where it has always sat because nothing has made it move?

There is now research on exactly this. A 2026 paper in the Georgetown Law Journal by Nora Freeman Engstrom at Stanford and Brianne Holland-Stergar at Montana, Competition and Contingency Fees, reviewed 500 personal injury firm websites. Twenty-two of them disclosed any fee information at all. Four point four percent. Not four point four percent disclosed a low fee. Four point four percent disclosed a number.

Their broader finding is that this market does not behave like a competitive one, and they document price clustering in a narrow band regardless of case complexity, quality of representation, or expected value.

Read that last clause the way a market person reads it. A price that does not respond to complexity, to quality, or to expected value is not a price. It is a convention.

The mechanism they identify is the one that made me finally take my own idea seriously. The client never writes a check. The fee comes out of a recovery that does not exist yet. So there is nothing for the client to compare, and they choose on advertising presence and perceived reputation instead of on price, because price is not information anyone has given them.

Their proposed remedy is disclosure, and they are explicit that fee caps are not the cure. I should also be explicit that they have never heard of me and are not endorsing anything. I am citing a finding, not borrowing a reputation.

But that finding shows a market where 95 percent of sellers will not state a price. That is not a market that has already optimized itself. It is a market that has never been asked to.

5. What an injury file actually costs to produce.

So I did the thing I had spent nine months learning to do. I stopped asking what the price was and started asking what the thing actually costs to make.

Here is the production line of a personal injury file, start to finish.

Answering the phone when an injured person calls. Taking the intake and getting the facts down accurately the first time. Running the conflict check. Calculating and calendaring the limitations date. Sending representation letters. Requesting records and bills from every provider, then chasing those requests, then chasing them again, because roughly nothing arrives on the first ask. Matching inbound faxes and mail to the right file. Filing and indexing what arrives. Keeping the client informed without making them ask. Tracking deadlines. Getting the crash report. Identifying coverage. Drafting correspondence. Then reading the medical chart. Building the treatment chronology. Deciding what the records actually say about causation. Finding the gap in treatment before the adjuster does. Valuing the case. Making the demand. Filing suit. Written discovery. Depositions. Motion practice. Mediation. Trial.

Now sort that into two piles.

The first pile is judgment. Reading the chart. Causation. Valuation. Strategy. What to concede and what to fight. When to file. Whether the offer is good. Every deposition. Every argument.

The second pile is everything else, and the second pile is enormous. It is most of the hours in the file and almost none of the lawyering.

At a conventional firm the second pile is staff, and that is a completely rational answer. Hire people, train them, supervise them, and pay them every two weeks whether or not a case resolved that month. Once you carry a salaried floor, the fee has to clear it. That is the actual reason a third is a third. It is not greed. It is arithmetic performed on a cost structure everyone inherited.

I want to state the strongest objection fairly, because it was put to me directly and it is a good one. It goes: the reason nobody runs at 20 percent is not that nobody thought of it. It is that a lower fee works while you are tiny and breaks the moment you have to add people, and anyone claiming otherwise is early enough to still be enjoying the honeymoon.

That is a real argument, it may turn out to be right, and I will come back to it at the end rather than pretend it away here.

6. Which lines I removed, which I kept, and the one I added back.

The second pile is machine work now, and I built the machines. Intake, including after hours. Records requests and the chasing that follows them. Matching inbound documents to the right file. Filing, indexing, and making the file visible to the client without anyone assembling an update. Status. Deadline tracking. Conflict and limitations gates. A follow-up cadence that stops the second a human answers.

I am not going to name a single vendor, and not because I am protecting anything. The tools are the least durable part of this answer. Anything I named would be replaced within a year, and if what you took away was "he uses this product," you took the wrong thing. The transferable part is the sort: decide which pile each task belongs in, then refuse to turn the second pile into payroll.

The first pile is mine and it is staying mine. No system I run reads a medical chart. Nothing builds my chronology. Nothing decides causation, finds the treatment gap, values a file, or tells me whether an offer is good. I do that on every case, and I intend to be doing it on every case ten years from now.

None of that means the work gets done without people. I will need help as the caseload grows, including from other lawyers, and I do not know yet what those roles look like. That is a near term problem rather than a prediction. When the clerical floor was never there to begin with, the jobs you turn out to need are not automatically the jobs firms hire for today.

This is where the AI objection arrives, and it deserves a straight answer rather than a defense. The objection is that automation on injured people's cases carries risk that is not worth whatever it saves.

I agree with the premise, and that is precisely why the line sits where it does. A system that mishandles a records request costs me an afternoon. A system that misreads a chart costs a client their case, and there is no version of that I get to file under efficiency. Rule 1.1 has no efficiency exception. So the machines touch work where a mistake is visible, recoverable, and mine to catch, and they do not touch work where a mistake is none of those things. That is not a risk tolerance I arrived at. It is the only allocation I could defend.

And here is the line I added back, days before writing this. I signed an advertising agency. For a while my honest answer to "how do you afford it" included carrying no meaningful client acquisition cost. That was true when I said it and it is no longer the whole picture, and I would rather put that here than have somebody find it and call it a reveal.

7. The part nobody has a job title for.

"I built the machines" is the kind of sentence you should make me back up, so here is what that actually means.

There is a client portal, where a client can see their own file, their documents, and where the case stands without calling me to ask. There is a mobile app on the App Store and on Google Play that does the same thing from a phone. There is a separate portal for investigators that shows them only the matters they are assigned to and nothing else. There is an intake pipeline that answers the phone at two in the morning. There is a document system that takes an inbound fax, works out which file it belongs to, and files it. There is a deadline tracker that does not get tired.

Now the part I want to be precise about, because it is the one claim on this page you could test in a five-minute conversation.

I specified all of it. I directed it, I made every product decision, I decided what it must never do, and I am accountable for every line of it the way I am accountable for a brief with my signature on it. The code was written with AI.

That is not an asterisk on the argument. It is the argument. I am not a software engineer and I am not going to pretend to be one in a document about honesty. What I am is the person who knows which forty seconds of a records chase actually matter, and that is the scarce input.

This is also why I do not think a vendor could have built this for me. Not because vendors are bad, but because the hard part is not the software. The hard part is the specification, and the specification requires somebody who has taken 250 depositions and knows that the treatment gap is the thing the adjuster is going to hunt for. Hand that spec to a good engineer and you get a good system. Skip the spec and you get a product that automates the wrong forty seconds beautifully.

I think this turns into a job. Somebody who understands both how a file actually moves and how to make software move it is going to be a role that firms hire for, and I do not think it is a paralegal with a scripting habit or an engineer who read a practice guide. It is closer to a practicing lawyer who learned to specify systems. I have not seen a good name for it yet. Legal engineer is the one I use.

Which raises the obvious objection, and it is a fair one: if this is a hire, then any firm can make it, and whatever I have is gone by next year.

I do not think so, and the reason is the whole point of the next two sections. Hiring a legal engineer does not remove a salaried floor. It adds a salary on top of one. A firm that automates the second pile while still employing the second pile has spent money, not saved it, and the savings only arrive at the end of a long and genuinely painful transition. The advantage was never the technology. It was not having the floor in the first place.

8. Why the fee does not step up.

Most contingency agreements step up when a lawsuit is filed. A third before suit, more after, sometimes more again near trial. Mine does not. Twenty percent, settled or filed, through trial and appeal.

The reasoning behind the step-up is not a mystery and it is not greed. Litigation is where the hours are, so the fee follows the work. It is internally consistent and I understand it completely.

But it is consistent given the cost structure underneath it. When the clerical hours inside litigation cost a fraction of what they used to cost to produce, there is nothing left for the step-up to pay for. The escalator was pricing something I no longer buy.

There is a second reason and it is the one I actually care about. A fee that rises when you file puts the lawyer and the client on opposite sides of the table at the exact moment the case needs a clean decision. If filing is right for the case, the client should hear that from somebody whose fee does not move either way. That is not generosity. It is the only version of that conversation I can have with a straight face.

Now the concession, before anyone has to force it. A contingency fee prices risk, not hours. That is the entire point of it, it is legitimate, and a lawyer who advances costs and carries a case for three years against a real possibility of zero has earned a number that has nothing to do with a timesheet. Anyone arguing fees should simply track effort has not thought about who funds the losses.

So my argument is narrower than "fees should match work." It is about the two case types where everyone knows in advance that the risk was not there.

The first is the small policy case. Twenty-five thousand in coverage, nothing else recoverable, liability clear. Everyone involved knows on day one what that file is worth and what it will take. A third of it, plus costs, and the injured person can net less than the people who worked on it. The percentage bites hardest exactly where the client has the least cushion.

The second is its opposite. A catastrophic injury with clear liability and a policy that tenders on a demand letter in 45 days. A third of a very large number for six weeks of excellent but not especially risky work.

Those two are the ends of the barbell, and they are where the risk justification runs out, because in both of them the risk was known to be small before anyone started. Most files sit at neither end. But a pricing structure should be judged partly on what it does at the ends.

9. Why this is a moat and not a discount.

Here is the part that is a finance argument rather than a law firm argument.

A discount is a price cut inside an existing cost structure. It is unsustainable by definition, because sooner or later the structure demands its money back. Every "we will beat any fee" pitch in any industry is this, and the people telling me my math fails are, quite reasonably, hearing a discount.

What I am describing is not that. The number is lower because the file costs less to produce. The 20 percent is an output, not a promotion.

Which raises the obvious question: if the inefficiency is real, why has nobody closed it?

Because an inefficiency anyone can exploit closes overnight and makes nobody any money. The ones that persist are the ones where seeing it and acting on it are different problems. Any firm in the country can read section five. Acting on it means dismantling a salaried floor that people's livelihoods depend on, rebuilding a production process while carrying a live caseload, and cutting revenue per file by forty percent during the transition. That is close to impossible for a firm that already exists, and nearly free for a firm that never built the floor in the first place.

That is the whole moat. Not technology. Sequencing.

Now the condition this entire thesis hangs on, and I would rather state it than have it found. It works as long as the prevailing fee stays roughly where it is. If the market compresses toward 20 percent, my advantage evaporates and I am just another firm competing on the same terms as everyone else. The bet is not that nobody will follow. It is that following is slow, and that the firms best positioned to follow are the ones for whom the transition is most expensive.

And one more limit, because a finance person will raise it and they will be right to. Legal services are a credence good. Clients cannot evaluate quality before they buy, and frequently not after. In credence markets, price does not behave the way it does for commodities. A lower number can read as a signal of lower quality and repel the buyer instead of attracting them, which is an objection I have had put to me more than once and have no clean answer to.

So the mispricing does not close simply because a cheaper option appears. It closes only if the cheaper option is visibly capable first. Which is why I would never ask anyone to hire me because I am less expensive. Hire the lawyer who can do the work. If two of them can, the number is a tiebreaker, and it should be.

10. What I do not know, and the date I will know it.

Here is the honest state of things.

I opened this firm in July of 2026. I have not resolved a case. I have not collected a fee. I have no employees. Every number in this document about what things cost is a claim about producing files, not about a year of operating results, because I do not have a year of operating results.

And one more thing belongs on that list. I built this firm around litigated files, because litigated files are the only ones I have ever watched a firm actually make money on. I have never run a pre-suit book and I have never really worked one. When a file was going to be filed I filed it, usually sooner than most people would have. So when I argue in section eight about the case that tenders on a demand letter in 45 days, I am reasoning about a phase I have never operated, and a lawyer who has run one for ten years knows things about it that I do not.

So when someone says the model breaks at scale, I cannot refute it, and a well-constructed argument is not the same thing as evidence. The strongest version of that objection is this: the cost base is less variable than I think. Somewhere between here and a real caseload sits a file that needs a person I do not have, and once I hire that person the salaried floor returns and takes more than the margin the model leaves. If that is true the whole thing folds, and it folds quietly rather than dramatically, which is worse.

I cannot rule it out. Neither can anyone arguing with me, which is why the argument is unresolvable in a comment thread. Both sides are reasoning from structure. Neither side has the number.

So I am going to produce the number.

One year from the publication of this piece I will post what a file actually netted and what it actually cost to produce. Not impressions, not case counts, not a testimonial. The two figures that decide whether any of this was real. And I will post them if they are bad, which is the only part of this commitment that means anything.

If the cost base turns out to be fixed, I will have been wrong in public with my name on it, and the people who told me so will be entitled to say so.

That is a better way to settle this than either of us talking. It is also, for whatever it is worth, the same rule I wrote down in 2024 before I put a dollar into anything: decide in advance what would prove you wrong, and then go look.