The objections I would raise if I were you.

I would rather you did not take my word for it. Some of the hard parts are open source and you can read them today: orhof, chachaml, the book. The Rainbow repository is not published yet, and the page that describes it says so. A worked sample of the audit findings is published on a synthetic company, so you can judge the thinking before you pay for it. And the first step is a fixed-price audit that can end in "do not build", which I have watched happen and which nobody enjoys, least of all the person who wanted to build. We do not publish client margin numbers because they are the client's to disclose, and most clients will not. I could estimate them. But the first CFO to check one estimate against the client would be the last CFO to read this site, and that trade is not worth a paragraph.

Machinery, because I do not trust intent to survive a year of edits, including my own. Every quantitative claim on the site resolves through one facts file that records where the number came from and who owns it, and anything marked unconfirmed does not render at all. A build check compares figures that appear in more than one place against that file and fails naming the file and the line. A second check holds a list of claims that were wrong once and must not come back, twenty of them as I write this, each carrying its reason. The counts on this page and on the validation page are computed from the collections when the site builds, so nobody can leave one stale by forgetting to update it. None of this guarantees good faith. But it makes drift expensive and visible, and you can check visible.

An agency sells you build capacity. A big consultancy sells you a deck and a junior team, and I have been on the receiving end of both. We sell a sequence that reduces uncertainty: find the decision, prove a better method on your data, and only then build, with the same senior people from the first meeting to the last. So the money follows the evidence, and if the evidence says the software is not worth building, the audit ends there. You have still bought something. The certainty not to spend is usually the cheapest thing we sell.

In two to three weeks: a decision map, an economic baseline, a cost-to-complete forecast on your own projects, a ranked shortlist of interventions, and an explicit go, reshape or stop recommendation. The sample audit page shows the full shape of it, and I would read that before reading anything else on this site. The fee is creditable against a larger engagement if you proceed.

Then you are who we work best with. We bring operations-research, formal-methods and product-engineering depth, and we turn analysis into an operated system. Your team brings domain knowledge and continuity, which we cannot supply and would not pretend to. The test is simple: does the combined team reduce the important uncertainty faster than yours alone? If it does not, the audit says so and the engagement ends there, and your team keeps the baseline we built.