You're already paying for AI
you don't use. I'll tell you exactly how much.

Because the discovery call is a sales filter, not work — it exists to qualify you and anchor a price you haven't seen yet. Every part of this is published before you pay: the exact deliverables, the exact scope limits, the exact data I need, the typical plan, and the payment terms: the full fee by card at checkout, or by invoice with 50% to start and the balance at delivery or over two monthly payments. You are buying a fixed scope on a written plan, the same way our active client engagements run.

You are not getting a partner markup, an analyst pyramid, a six-week timeline, or a 60-slide deck. Most of what larger assessments charge for is organizational overhead. This is one senior person, a short written schedule, and a scope that does not move. What you are also not getting: any assessment of AI model quality, any custom benchmarking research, or anything outside the published scope.

The roadmap names the work, and yes — my other company builds that kind of work. I am telling you that here rather than hiding it. But every item on the Unclaimed Wins list ships with a build-it-internally estimate in days, so your own team can quote it, and the 90-day roadmap is written to be executable by any vendor or by you. There is no implementation contract attached to either tier, no retainer, and no auto-renew. If the audit's honest answer is "cancel three things and build nothing this year," that is what the report says — and roughly half the value of this product is telling companies to stop buying.

Because I am not learning your business — I am reading finite artifacts: your contracts, your license usage exports, your card and expense lines, your SSO and identity logs, your renewal dates. That set is bounded and it does not require domain expertise to read. The six interviews cover intent — what people think they bought and why. The data covers reality. The gap between those two is the entire product. Where the data cannot settle a question, the item is marked "Needs Verification" with the specific document required, and it does not get counted in the recoverable-dollars total.

They could. They haven't — and the reason is structural, not a skill gap. The person who would run this audit is the same person who chose, championed, and now owns the tools the audit would recommend killing. That is not a report anyone writes about themselves. Separately, do the arithmetic on doing it in-house: $2,500 for a typical week of work is $62.50 an hour across a 40-hour week. No VP of IT's loaded rate is that low, and that week comes out of the roadmap work they are already behind on.

It is the best possible timing. This audit is deflationary — its primary output is a list of things to stop paying for, footed to a dollar number, with the cancellation deadline for each one. It is not a proposal to spend more. If your budget was frozen because nobody could show a return, the one-page CFO memo is the artifact that ends that argument in either direction. Buy it out of the software budget it is going to shrink.

Read-only exports only. No production system access, no admin credentials, no PII. Specifically: your software contracts or a vendor spend export, Microsoft 365 / Google Workspace license assignment and activity reports, SSO or identity provider application logs, 12 months of expense-report lines filtered to software, and your renewal calendar if one exists. Confidential by default. Everything is deleted 30 days after delivery with written confirmation. If your security team will not release those exports, do not buy this — it will not work.

Daniel Graham does all of it, personally. There is no team, no junior analyst, no offshore partner — which is exactly why the capacity is four per month and why the calendar cuts off when it is full. I use AI tooling to accelerate the ledger reconciliation the same way an accountant uses a spreadsheet; the findings, the confidence ratings, and every recommendation are mine, and I sign the report.

It is built for 150–2,000, and 250 is squarely inside that. Under 150 employees there usually is not enough software spend to make the ledger worth a week of your own time, so don't buy it. Over 2,000 employees, one legal entity and three business units stops being a realistic boundary and you need something bigger than this. Those limits are published so you can disqualify yourself before checkout instead of after.

You still get the full deliverable: the report, the ledger, the scorecard and the roadmap, every line sourced to your own exports. The qualification gate above exists so an engagement only starts where there is enough software spend to find. Terms: the full fee by card, or by invoice with 50% to start and the balance at delivery or over two monthly payments — the same written process I run for active client builds.