Analysis Sponsor Readout
strataforge-cost-3-readout.pdf
The baseline, the savings ranked, the decisions the sponsor needs to make, and what the next four weeks will put in place.
Download PDF of Analysis Sponsor ReadoutSample case study · AWS cost analysis & optimization
Six weeks with StrataForge Robotics, a fictional robot maker whose AWS bill grew 49% in a year while revenue grew 9%. The Analysis found $23.5 million a year. StrataForge’s engineers put $11.1 million of it in place through their own pull requests and runbooks, and their Cost and Usage Report measured $9.4 million.At your size, the gate is three times the fee, $105,000 a year, whatever the bill. StrataForge shows the method, not a typical result.
StrataForge Robotics is a fictional company: every account, figure, and person on this page is simulated. The AWS prices are real list prices, retrieved October 13, 2026.

StrataForge Robotics, Inc. does not exist. We built it to show what this engagement produces when the client is big enough to need all of it: an AWS organization of 48 accounts, simulated resource by resource and priced at AWS list prices, with thirteen StrataForge people who make the decisions real owners make. Then we ran the six-week Optimization against it, week by week, in a real Pavilion workspace.
The eight reports are the real deliverables, written the way we write them for a client, and they agree with each other to the dollar because every figure comes from one deterministic simulation. Our clients’ bills stay confidential, so this is how we can show you every page.
In July 2026 StrataForge’s AWS bill was $4.6 million, 49.2% more than in August 2025, while revenue grew 9%. Left alone, the next twelve months would cost $70.6 million. A quarter of the spend carried no team tag, so nobody could say who owned it, and the board wanted to know whether this was growth or waste.
Priya Raman signed the six-week Optimization: the two-week Analysis first, then four weeks putting the savings in place through StrataForge’s own change process, and the same bill read again at the end. Tom Becker, who runs cloud financial management, led it day to day, and Ravi Menon, the vice president of finance, approved anything StrataForge would buy.
Each week ended with a report, and every report is below. The sessions, decisions, and tasks are counted from the engagement’s Pavilion workspace.
The read-only role went into all 48 accounts, and twelve months of the Cost and Usage Report came with it, tied to each month’s invoice before it was used. The baseline answered the board: the growth sat in the robot-learning platform, the fleet telemetry pipeline, and the five accounts that arrived with Kestrel Vision, not in one bad month.
It also showed the shape of the work: $1.1 million a month with no owning team, 1,428 unattached volumes, 63 of 148 databases and 15 of 16 Kubernetes clusters paying for extended support, and Savings Plans covering a third of the compute they could cover. Ten decisions were recorded, two of which shaped everything after: the security detection services are not waste, and nothing is bought or renewed until the readout.
In the room. Five sessions: the kickoff, a finance session with Ravi Menon, the architecture walkthrough, and two office hours.
Every saving was priced in a fixed order, so nothing was counted twice: $23.5 million a year across 13,020 opportunities on 73,592 resources, for a one-time cost of $128,812. All of it would take the bill from $4.6 million to $2.6 million a month.
The readout carried the finding finance had not expected. StrataForge’s private pricing agreement commits it to $42.0 million in its third year, so at flat usage only $13.7 million of the savings could be kept. The recommendation was to renegotiate the commitment, not to spend into it, and Ravi Menon agreed to open that negotiation.
Then each owner went through the roadmap line by line, to apply, defer, or decline. They approved 10,665 of the 13,020 changes.
In the room. Six sessions: a walkthrough of the robot-learning training schedule, a logging-scope check-in with the CISO, the savings review with ten owners, the sponsor readout, and two office hours.
9,302 changes in one week, estimated at $7.8 million a year: waste deleted after a seven-day Recycle Bin hold, non-production instances on nights-and-weekends schedules, and S3 lifecycle rules and Intelligent-Tiering on the storage. 1,300 went in as pull requests and 8,002 through scripted runbooks, across 652 change tickets.
Not everything went in. The SAP team declined sixteen quality-system schedules. The robot-learning team, which had approved ten lifecycle rules at the readout, withdrew them because a regression study needs a year of results. That was their call. The statement of work counts a change approved at the readout and declined later at its estimate, so the guarantee cannot be defeated by a change approved and then never made.
In the room. Seven sessions: two change reviews, two pairing sessions on the schedule stack and on S3 lifecycle, two office hours, and the first sponsor check-in.
1,232 changes worth $2.6 million a year, 58% of it from moves to Graviton. The first production move was one node group of three in the fleet ingest cluster; the other two waited for two clean weeks.
Ravi Menon approved two one-year Savings Plans, for databases and for SageMaker, and the platform team bought them. He deferred the largest single line, $2.1 million a year in three-year Compute Savings Plans, to a written answer by October 30, after the pricing renegotiation opens. Those plans would start in November, after the measurement window, so they were never going to count toward the gate.
In the room. Eight sessions: two change reviews, the commitment decision meeting, two pairing sessions on Graviton and on Aurora, two office hours, and a sponsor check-in.
131 changes worth $778,520 a year, at no one-time cost. The largest, $599,428 a year, narrowed the CloudTrail data events, and because that changed what gets logged, the CISO agreed to the new scope in writing first. Security operations re-ran its detections the next day, and every one fired.
Six guardrails went in so the savings stay: budgets, anomaly alerts, a tag policy, a retention check, a volume check, and showback by team, each with an owner Priya Raman named. $2.4 million a year of re-architecture, from telemetry delivery to Spot for simulation, kept rolling out through October 30.
One plan slipped. The Kubernetes 1.31 upgrades scheduled for weeks three to five did not run, so the workspace records the date AWS will force them, November 26, and the new target, November 20.
In the room. Nine sessions: two change reviews, the CISO’s detection review, two pairing sessions on guardrails and on Spot, the change board’s route-table plan, two office hours, and a sponsor check-in.
Every change was measured on the Cost and Usage Report the way Tom Becker agreed on Monday: net amortized cost, seven days before against seven days after, the change day left out. 10,504 changes counted for $9,377,111 a year, 95.8% of what was estimated for the same changes, and a whole-bill check agreed within 1.6%.
Twenty changes measured at or below zero, and every miss was published at its measured value. All of them traced to five incidents, from a schedule suspended for a release test to a new caller that doubled a function’s invocations.
The handover workshop gave every line of the remaining $11.8 million a year an owner and the event that releases it, and the monthly cost review was rehearsed on the August close. Priya Raman accepted the measurement on October 2, and the second invoice went out.
In the room. Seven sessions: a pairing session on the measurement, two office hours, a sponsor check-in, the two-hour handover workshop, the monthly-review rehearsal, and the closing readout.
Half the fee rode on one number: annualized savings in place at week six, measured on StrataForge’s own Cost and Usage Report, of at least three times the fee. Each step from estimate to measurement is in the reports.
a year across 13,020 opportunities on 73,592 resources, priced in week two.
a year at estimate, in 10,665 changes their owners approved and StrataForge’s engineers applied.
a year across the 10,504 changes measurable by week six: 95.8% of their estimate, every miss included.
Counted toward the gate: $9,652,024 a year, the measured $9,377,111 plus $274,912 for the ten changes approved at the readout and declined later. That is 275.8 times the $35,000 fee, against a gate of three times. Net of the $97,097 one-time cost of the changes, it is 273.0 times. The second invoice, $17,500, went out on October 2.
Still to come: $11.8 million a year of the roadmap, each line with an owner and a date, most of it waiting for robot firmware, a maintenance window, a build, or a governance review. $1.3 million a year of Intelligent-Tiering was already in place but could not show on the bill until objects had sat unread for thirty days. And the private pricing renegotiation decides how much of all of it finance keeps.
Want this gate on your own organization?
Book a call about the OptimizationThe engagement ran in a Pavilion workspace, created from the Optimization’s template with every week, session, deliverable, and obligation already in place, on StrataForge’s calendar and in its time zone. Fourteen people joined it: the sponsor, finance, the platform and team owners, the CISO as an observer, and Charles.
After each session the transcript came into the workspace, and an agent drafted the summary, the decisions taken, the action items, and a digest for the people who were not in the room, each claim citing the part of the transcript it came from. Charles reviewed every draft before StrataForge saw it. Six pairing sessions were screen shares that StrataForge chose on day one not to record, so each of those got consultant notes instead.
Every report was published there as an immutable, hashed version, and Priya Raman accepted each one against that version. The fifteen deliverables were checked off against that evidence, and the measurement panel kept the baseline, three weekly estimates, and the closing measurement side by side, so an estimate and a measurement were never mistaken for each other.

When Ravi Menon deferred the $2.1 million of three-year Compute Savings Plans, the decision went into the workspace with its condition and a task for his written answer, due October 30. Anyone reading the roadmap later finds the reason next to the line.
The robot-learning team’s ten withdrawn lifecycle rules were settled in one thread: Hiro Tanaka explained the regression study, Charles answered that it was his call, and the ten were recorded as approved and then declined. When the sponsor accepted the measurement, the reason was a click away.
Marcus Oyelaran agreed to the CloudTrail scope in a thread, in his own name, including the reads of the data lakes that would no longer be logged. Security operations tested its detections against the new scope and posted the result to the same thread before the change went in.
StrataForge’s obligations sat on the same page as Vantalect’s: sponsor time, engineer hours, access, a named owner for every account, and the 110-hour cap on Charles. Over six weeks one was missed, week three’s pairing time, 3.25 hours of the 4, and it is on the record.
The engagement’s last action item was Tom Becker’s: one archive of the workspace, with every report, query, runbook, decision, and transcript, in StrataForge’s shared finance folder. Then the workspace closed.
Nine PDFs and 314 pages. Start with the readout; the reading guide shows how the rest fit together. Then read the weeks in order and you have the engagement the way StrataForge’s sponsor did. Each one is tagged for screen readers, and each says on its cover that StrataForge is fictional.
Written for the sponsor, and the one to read if you read nothing else: the baseline, the savings ranked, the decisions the sponsor needs to make, and what the next four weeks will put in place.
strataforge-cost-3-readout.pdf
The baseline, the savings ranked, the decisions the sponsor needs to make, and what the next four weeks will put in place.
Download PDF of Analysis Sponsor ReadoutThe full analysis behind the readout: the waste register, the rightsizing and commitment plans, the architecture review, and the ranked savings roadmap, with the prescriptive change for each.
strataforge-cost-2-analysis.pdf
The waste register, rightsizing plan, commitment plan, architecture review, and the ranked savings roadmap, with the prescriptive change for each.
Download PDF of Cost Analysis and Savings RoadmapThe reading guide first, then what each week delivered, in order.
strataforge-0-guide.pdf
Who StrataForge Robotics is, what each of the sixteen sample reports (and the audit’s code appendix) delivers, what the two engagements found and fixed, and what is real and what is simulated.
Download PDF of Sample Engagement Reportsstrataforge-cost-1-baseline.pdf
Twelve months of StrataForge’s AWS spend by account, service, environment, and team, the full resource inventory, and the commitments in place.
Download PDF of Spend Baseline and Inventorystrataforge-cost-4-quick-wins.pdf
Waste removed, non-production schedules, storage lifecycle, and log retention, change by change, with what was declined and why.
Download PDF of Quick Wins Appliedstrataforge-cost-5-rightsizing-commitments.pdf
Compute, databases, caches, and search resized to measured use, the Graviton moves, and the commitment decision finance approved.
Download PDF of Rightsizing and Commitmentsstrataforge-cost-6-architecture-guardrails.pdf
Data transfer, NAT, logging, metrics, and Spot changes, and the budgets, anomaly alerts, tag policies, and showback that keep the bill down.
Download PDF of Architecture Changes and Guardrailsstrataforge-cost-7-measurement.pdf
Every change measured on StrataForge’s Cost and Usage Report, annualized, and the guarantee’s gate.
Download PDF of Week-Six Savings Measurementstrataforge-cost-8-handover.pdf
The remaining roadmap with owners and dates, the commitment calendar, the monthly cost review, and the KPIs StrataForge runs from here.
Download PDF of Handover Playbook and Monthly Cost ReviewThe same six weeks and the same gate, on your own Cost and Usage Report.

The Analysis and the Optimization: the baseline, the waste register, and the roadmap in two weeks, then the savings put in place and measured, with half the fee riding on the result.
Read the engagement
The same client, July 13 to August 21: 48 accounts scanned, 54 findings verified, and every critical closed or accepted in writing before the cost work began.
Read the case study
A fictional ticket exchange in the middle of a migration, with a bill of about $50,000 a month: $148,967 a year measured, and the gate cleared 4.35 times.
Read the case studyTell us roughly what the bill is and what prompted the question. You will hear back from Charles within one business day, and the first call is fifteen minutes.
AWS, Amazon Web Services, and the AWS service names are trademarks of Amazon.com, Inc. or its affiliates. Vantalect is not affiliated with or endorsed by Amazon. StrataForge Robotics, Kestrel Vision, and every person named on this page are fictional.