You have won $197m of federal work since 2019 and you are still winning at the same rate. Three quarters of what is left on your books runs out within eighteen months, and 93% of it belongs to one customer.

You are winning new work as fast as you ever have. That is the good news, and it is genuine.

Sorted by the year each contract started, your record shows no decline in the ability to win. 2024 and 2025 were your two strongest years in seven.

So nothing in this report is about a business losing its edge. A firm that wins $21m of new work two years running is doing the hardest part right. The question is what happens next, and that lives in a different column of the same records.

The uncomfortable half

Each of those contracts also carries an end date. Group the 40 that are still live by when they finish, and three quarters of the value lands inside the next eighteen months.

A recompete is not a loss. Incumbents win most of them. But 35 of them inside eighteen months is a workload, and it arrives whether or not the team has capacity for it that quarter.

99.8% of every prime award you have won since 2018 comes from the Department of Defense. The sub-agencies look like four customers and are four doors into one.

Read by sub-agency, as above, that looks like a spread of customers. It is not. Army, Navy, Air Force and Special Operations Command are all the Department of Defense, and together with the rest of its components they are 99.8% of $197.1m across 355 awards. Everything outside Defense, across all 4 departments that have ever awarded you, comes to less than a quarter of one percent.

Concentration is not automatically a weakness. Knowing one customer deeply is usually why a contractor wins repeatedly, and your win rate suggests exactly that. The risk is narrower and worth naming precisely: a single customer changing how it buys reaches 93% of what you hold.

That could be a decision to bundle several small contracts into one larger vehicle, a shift to a different contract type, or a budget change at one installation. None of those is a judgement about your work. All of them would arrive at the same time as the recompete window in Part 1.

Most of the obvious answers assume a business that is failing to win work. Yours is not, which removes them.

Each play below was checked against your own award record and against what the public solicitation files show about how this customer buys. The point of publishing the rejections is that they are the expensive part to work out.

If you win recompetes at the rate incumbents normally do, this is a busy year. If you win them at your own historical rate, it is a comfortable one. The gap between those two is the whole question.

There is no honest way to forecast this from outside, and we are not going to pretend otherwise. What the record supports is a range, and the shape of what falls inside it.

Your new-award record makes the top rows more plausible than the bottom ones. You won $21.7m in 2024 and $21.4m in 2025 without any of these recompetes landing. A firm winning at that rate is not usually one that loses most of its recompetes.

The strongest case against acting on any of this is that the pattern is normal for your kind of contract, and that we cannot see the difference from outside.

1. Short contracts with option years look like a cliff and are not one

Many federal contracts run one base year with several option years the customer can exercise. The published end date is often the end of the current period rather than the end of the relationship. If most of your 35 expiring contracts are mid-option rather than genuinely ending, the cliff in Part 1 is an artifact of how end dates are recorded and the correct response is to do nothing. This is the single most likely reason this report is wrong, and you can settle it in an hour with your own contract files.

2. Concentration in fire alarm work at military installations is close to unavoidable

You do a specialised thing at a specific kind of site. The customer that owns most of those sites in your region is the Army. Diversifying customers may mean diversifying geography or trade, which is a much larger decision than a portfolio adjustment and could easily be a worse business than the one you have.

3. You may already know all of this

If you track your own end dates in a schedule, none of Part 1 is news, and the value of this document is limited to the competitor picture. We have no way to see your internal systems from the public record, and a report that assumes you are not paying attention would be a poor one.

Build the recompete calendar first, before deciding anything about customer mix. It is the cheapest thing on this list and it determines whether the rest matters.

The recommendation is deliberately the smallest one available. Line up all 40 live contracts by end date, mark which are true expiries and which are option exercises, and put the resulting workload against the weeks available. That single view decides whether anything else in this report deserves attention.

The first one can end this exercise by lunchtime, which is the point of doing it first.

Day 1. Separate real expiries from option years

Take the 35 contracts expiring inside eighteen months and mark each as a genuine recompete or an option the customer can simply exercise. If most are options, Part 1 is an artifact of how end dates get published and you can stop here. That result would be worth knowing on its own.

Week 1. Put the real ones on a calendar against your capacity

Recompetes cluster. Count how many fall in the same eight-week windows and set that against who actually writes your proposals. The constraint in this business is rarely the work and often the number of people who can write while the current jobs continue.

Weeks 2 to 6. Ask about bundling before it is announced

The single event that would hurt most is several of your smaller contracts being combined into one larger vehicle you are too small to bid alone. Contracting officers will usually discuss acquisition strategy ahead of a solicitation. That conversation is free, and it is far more useful eighteen months out than eighteen days out.

Weeks 6 to 12. Only then look at the customer mix

If the calendar turns out to be manageable, the 93% concentration becomes the next question rather than an urgent one. Handle it in that order. Doing it the other way round spends the scarce months on the slower problem.

We build the analysis a business would get from a good outside team, from public records, and we send it before anyone asks.

We are Scalable OS. We work with public records: federal contract awards, agency records and public solicitation files. From those we reconstruct from them what is actually happening inside a business and the market around it. Then we send that to the business, unsolicited, before there is any relationship at all.

The reason is straightforward. The analysis in this document is the kind that normally arrives after a retainer, a discovery phase and a scoping call, which means most contractors never see it at any point in their working lives. It is not expensive to produce, because the underlying records are public and free. It is that nobody has a reason to produce it for you until you are already a client. We would rather demonstrate the work than describe it.

There is a second reason, and it is the one that decided the shape of this document: nothing in it was requested. A search engine or an assistant answers the question you thought to ask. This report exists to raise the ones nobody inside your business has had a reason to ask, because your own contract end dates are scattered across 40 separate documents and nobody adds them up.