Book a working session
A CLOSED-LOOP OPERATING METHOD

Find where your operations and products lose results, and fix the system behind them.

SPADE OS is a diagnostic and operating method that treats a business the way an engineer treats a control system, tested over 20+ years in enterprise technology, logistics and aviation. It traces each problem across strategy, data, processes and people, inside teams and in the handoffs between them, redesigns what is broken and runs a closed loop until the change holds.

Book a working sessionSee the cases

A 30-minute working session on your operation, with a first map of where results leak that is yours to keep.

Results from past engagements
$100M+
efficiency gains
one platform used by 400K+ people, enterprise IT
90%
complex projects delivered on time
up from 30%, $5M+ projects, enterprise IT
70×
lift over a control group
one personalized campaign, aviation
2 weeks → immediate
data request turnaround
self-service data platform, logistics

Figures from engagements in enterprise IT, logistics and aviation.

Sound familiar?

The problems that cost the most rarely belong to a single team.

  1. 1Two teams report different numbers for the same thing, and meetings start with whose number is right.
  2. 2Analysts spend most of their week reconciling data, so decisions wait.
  3. 3The roadmap fills up with requests, and no one can say which shipped features moved a metric.
  4. 4Sales promises something that product and delivery have never heard of.

Each of these sits between teams, in a handoff no one owns, and it only becomes visible when a business or a product is read as a system of systems, which is how SPADE OS reads it.

In one recent diagnostic, about 70% of the improvement ideas required two or more teams to change together.

MARKET NEEDSMarketingProductFinanceSalesOperationsLEADSWHAT TO DELIVERBOOKINGSCOSTSMETRICS IN CONFLICTOUTPUT NO ONE USESINPUT GUESSEDFEEDBACK LOOP: MEASURE, COMPARE, CORRECTWORKINGBROKENCOMPANY LOOPDEPARTMENT LOOP

Swipe to see the full diagram.

Each team turns inputs into outputs for its neighbors, and results leak where those handoffs break.

How it differs

One team, seven disciplines, and the whole operation in view.

AspectTypical approachSPADE OS
Where it looksInside one department, where the metrics already liveAcross the whole system, including the handoffs between teams that no one owns
Starting pointThe solution named in the requestThe problem behind the request, diagnosed before any solution
MetricsOne KPI per teamEach metric paired with a counter-metric, designed from the behavior you want
ExpertiseSeparate specialists for strategy, data, process and changeSeven disciplines handled as one system
After deliverySet once and left to driftA closed loop that measures, compares and corrects every cycle
Where it applies

A fulfillment operation and a product team both take inputs, turn them into results and hand work to other teams, so both can run as a closed loop with the same method.

  • Operations

    Departments, processes and the handoffs between them.

  • Products

    The roadmap, product metrics and the product operations that connect product, engineering, sales and customers.

  • Data and analytics

    The numbers each team decides on, from one trusted source.

  • Transformations

    Taking a company, team or process from its current state to a target state, with the targets, incentives and routines that make the change hold.

SPADE is the method and OS is the ground it runs on.

SPADE, THE METHOD
StrategyClear goals, real targets, and the feedback loops that keep them honest.
ProductProduct management and product operations, built for the outcome and grounded in the real problem behind the request.
AnalyticsData that drives decisions and traces back to the question it answers.
Behavioral DesignThe conditions that make a change get adopted and hold.
ExecutionThe processes and incentives that turn decisions into consistent behavior.
OS, THE GROUND IT RUNS ON
OperationsThe day-to-day work the whole system has to move.
Systemic thinkingReading the organization as one connected system, where every part touches the others.
How it works

Diagnose first, change in small tested steps, and keep the loop closed.

Phase 0

Diagnostic

Before any solution is proposed, the business is mapped team by team, whether the problem sits in an operation or a product, and the request is traced back to the problem underneath it.

  1. 1Context and hypothesis: a systemic read of the business before going deep on any team.
  2. 2Team mapping: each unit's goal, the questions it must answer, the data it trusts, and what it hands to and takes from its neighbors.
  3. 3Gap and conflict analysis: where metrics pull against each other, where information never reaches the team that needs it, and where loops are broken.
  4. 4Recommendation and roadmap: three horizons run in parallel, 30 to 60 days, 3 to 6 months and 6 to 18 months.
Phase 1

Intervention

Change happens in A/B tested baby steps, each aimed at the behavior that, once changed, moves the metric that matters.

  1. 1Small: each step is sized so nothing important is put at risk.
  2. 2Tested: each step is measured against a control before it counts as a result.
  3. 3Growing: once a step proves itself, the next one is larger, so the pace grows with the evidence.
  4. 4Owned: the people involved help find the problem, so they own the change when it arrives.

Acceleratingly large, A/B tested baby steps is one of AE Studio's core values.

Ongoing

Closed loop

Targets, measurement and response routines are built into the operation, so results are compared to the plan every cycle and corrected before they drift.

Phase 1: metrics start from the behavior

Three forces shape behavior in every organization:

  1. 1People do what they are measured on.
  2. 2People come to want what they see every day.
  3. 3Every measure people see often enough becomes a target, and faster when their incentives depend on it.

SPADE OS puts these forces to work intentionally. It defines the behavior first, measures it with a metric rich enough to track it, pairs it with a counter-metric, and then keeps both in front of people every day, so the target people chase is the one the business needs.

A note on Goodhart's Law

Goodhart's Law says that when a measure becomes a target, it ceases to be a good measure, and it is often read as a reason to measure less. Since every measure people see often enough becomes a target, measuring less only leaves that target undesigned. SPADE OS takes the opposite lesson from the law and designs the target with care: it starts from the behavior, makes the measure rich enough to resist simple optimization, and adds a counter-metric. Don't let perfect be the enemy of measured.

Example: why every NBA draft lottery reform creates a new way to tank →

Speed with quality

Behavior

Ship working software faster.

Metric

Delivery speed.

Counter-metric

Code quality, so going faster cannot come from cutting corners.

Revenue with margin

A logistics unicorn

Behavior

Sell loads that make money.

Metric

Revenue, set each quarter during OKR planning.

Counter-metric

Margin, set at the same time and carried into every department.

Health scored by the weakest part

A product and AI consultancy

Behavior

Catch a failing project before the client does.

Metric

A health score for product, client and team.

Counter-metric

The project rolls up on its weakest area, so a strong client relationship cannot cover a failing delivery.

Cases

Three operations, three industries, one method.

Each case shows what was asked, what the diagnosis found and the result, with the full trace one click away.

Logistics2019 to 2021

A logistics unicorn: from a reconciliation queue to self-service data

2 weeks → immediate
data request turnaround
2.3 → 4.9
data integrity rating, out of 5
2.1 → 4.7
internal satisfaction
2 → 35
team size

The request

Lead the BI team.

What we found

The data platform was a single table with no business rules. Numbers did not match across departments, the team spent about 80% of its time reconciling them, and a data request took two weeks.

Key changes

  1. 1

    Mapped the operation and how each department consumed data.

  2. 2

    Projected the company's growth and data demand two years ahead, and designed the team, operating model and service architecture for that scale.

  3. 3

    Separated the scopes of data engineering and analytics, secured approval to hire data engineers, and kept executive reporting running while the team built.

See the full trace
  1. 1

    Mapped the operation and how each department consumed data.

    Systemic thinking
  2. 2

    Projected the company's growth and data demand two years ahead, and designed the team, operating model and service architecture for that scale.

    StrategyOperations
  3. 3

    Separated the scopes of data engineering and analytics, secured approval to hire data engineers, and kept executive reporting running while the team built.

    OperationsExecution
  4. 4

    Built a self-service platform and embedded analysts in each department as partners, replacing the ticket queue.

    AnalyticsProductBehavioral design
  5. 5

    Set shared business definitions across sales, finance, pricing, operations, product and engineering.

    Analytics
  6. 6

    Led the OKR process and carried the leadership practice of setting revenue and margin together into every department, so each target had a counter-metric.

    StrategyBehavioral design
  7. 7

    With accounting, found that slow surcharge invoicing was eroding margin within three months, then automated invoicing with the product team.

    OperationsSystemic thinking
  8. 8

    Found that sourcing lead time had the strongest positive correlation with margin, and operations and pricing changed their processes to gain lead time.

    AnalyticsExecution

This work ran during the growth that took the company to unicorn status. The VP of Product, who had doubted the proposal, asked for the same model for all his teams once he saw it delivered.

Aviation2025 to 2026, delivered with AE Studio

A major airline: four AI projects connected into one revenue system

70×
lift over a control group, one personalized campaign
~100×
sales relative to campaign cost
~60×
sales relative to cost, second campaign
1 week → 1 click
time to run a campaign

The request

Each department asked for its own AI or analytics project: pricing, competitive pricing, route profitability and CRM.

What we found

Each project delivered value on its own, and the largest opportunity sat between them. CRM promoted flights that revenue management was protecting, with neither team able to see the other, and route profitability was confirmed only after accounting closed, months later.

Key changes

  1. 1

    Interviewed each team and mapped how each project's output could feed another.

  2. 2

    Reframed the request to evaluate competitor P&L into two decisions: which markets to enter, and whether weak markets reflect internal efficiency or a market limit.

  3. 3

    Restructured the four projects into shared components for market data, pricing, forecasting and recommendations.

See the full trace
  1. 1

    Interviewed each team and mapped how each project's output could feed another.

    Systemic thinking
  2. 2

    Reframed the request to evaluate competitor P&L into two decisions: which markets to enter, and whether weak markets reflect internal efficiency or a market limit.

    StrategyProduct
  3. 3

    Restructured the four projects into shared components for market data, pricing, forecasting and recommendations.

    ProductOperations
  4. 4

    Connected the chain so predicted empty flights from revenue management and competitor price moves feed CRM targeting.

    AnalyticsSystemic thinking
  5. 5

    Built a model that estimates the profitability of flights already flown, so decisions no longer wait for the accounting close.

    Analytics
  6. 6

    Designed campaigns around behavioral segments, such as customers who searched and did not buy or repeat travelers on a route, with several message variants each and a control group in every test.

    Behavioral designAnalytics
  7. 7

    Set a weekly operating rhythm with the client: demo, joint sprint planning and formal acceptance.

    ExecutionOperations
Enterprise IT2011

A global technology company: tool adoption solved as a behavior problem

30% → 70%+
projects using the tool, in 10 months

The request

Increase use of the requirements management tool across all projects.

What we found

Usage was about 30% of projects. A survey of every business analyst, run before any action, showed four causes: people did not know the tool's benefits, did not know how to request it, gave up when the deployment team did not respond and they had no one to escalate to, and dropped it when they could not use it fully.

Key changes

  1. 1

    Surveyed every business analyst before acting.

  2. 2

    Formed a group of four analysts with tool experience to act as multipliers.

  3. 3

    Built training around each of the four drop-off points.

See the full trace
  1. 1

    Surveyed every business analyst before acting.

    AnalyticsSystemic thinking
  2. 2

    Formed a group of four analysts with tool experience to act as multipliers.

    OperationsBehavioral design
  3. 3

    Built training around each of the four drop-off points.

    ExecutionBehavioral design
  4. 4

    Contacted project teams proactively to offer help.

    Behavioral design
  5. 5

    Replaced the request process with one 30-minute intake per need and a named person assigned, giving orientation so people learned by doing.

    OperationsBehavioral design
  6. 6

    Ran the same survey after 10 months.

    Analytics

Also at the company

  1. Complex ($5M+) projects delivered on time rose from 30% to 90%, starting from unhappy clients, poor delivery and budgets running out.
  2. $100M+ in efficiency gains from a platform used by 400K+ people. The request was a technology update, and the work added design practices built around how people actually update their profiles.

Eduardo has consistently led our team and our Customers to success on complex engagements.

Christopher Wolters, manager and client at the company
Who you'll work with

The AE Studio team

SPADE OS engagements are delivered with AE Studio's engineers, data scientists and AI specialists, and the team that runs the diagnostic also builds what it calls for.

Eduardo Mylonas da Silva

20+ years in product, operations and analytics at IBM, Loadsmart and AE Studio, trained in control and automation engineering.

LinkedIn

AE Studio is an applied AI studio that builds production AI for its clients.

The working session

A working session on your operation.

Bring one problem your operation keeps running into. In 30 minutes we trace it across the departments it touches, and you leave with a first map of where results leak and what to look at first, yours to keep whether or not we work together.

Book a working session