Organisational investigations // For boards and chief executives

Find the Cause. Fix the System.

When a problem survives 2 restructures, the cause is usually in the system, not the people. We investigate it the way air accident investigators examine a crash. Your board receives a written finding: what is causing the problem, what it costs and what to change.

Book the first meeting Take the 4-minute check

12–16 weeks
Fixed scope and fee
No fee for the first meeting

Board pack // Item 6Page 14 of 62

Operations report, quarter 3

Overall statusGreen
Quarterly targetsMet, with minor variances
Regional restructureComplete
EscalationsNone raised this quarter
Risk registerNo new risks
Network investmentApproved unanimously

Prepared by management for the boardFor noting

Overall statusReported: green. Found: Green for 14 months. Nobody has tested it.
Quarterly targetsReported: met, with minor variances. Found: Missed for the third quarter running
Regional restructureReported: complete. Found: The third redesign of the same reporting line
EscalationsReported: none raised this quarter. Found: The route has no deadline, so issues go round it
Risk registerReported: no new risks. Found: 17 issues downgraded before they reached this page
Network investmentReported: approved unanimously. Found: Described 3 different ways by the people who voted

Move the lens across the board pack.Drag the lens across the board pack. Above the glass is what the board was told. Under it is what an investigation found. Specimen. A composite of disguised cases. It describes no client.

Exhibit 01 // When to call us

3 Signs the Problem Is in the System

If one of these describes your organisation, the cause is unlikely to be the people you have been changing.

It Keeps Coming Back

You have restructured the same area more than twice, and the problem returned each time. If the structure were the cause, the first change would have fixed it.

Agreed in the Room, Not Funded Outside It

The minutes say the decision was unanimous. What gets funded afterwards says something else.

The Reports Changed Nothing

You already hold good analysis. The recommendations were written before the diagnosis was finished, so the problem is still there.

Who Calls Us

Boards, and the executives who answer to them, in organisations where a small failure costs millions. At a mobile operator, network downtime does. At a mine, so does a power failure.

Board chairs and directorsReports say the plan is on track. Results say it is not.
Audit and risk committee chairsClean audits and a quiet risk register that feel too clean.
Chief executives and chief operating officersThe same problem back after the second restructure.
Chief risk officers and heads of internal auditFindings raised, closed and raised again.

Exhibit 02 // Where the method comes from

Fix the System. Not the Pilot.

Our method is adapted from air accident investigation. Under ICAO Annex 13, investigators may not ask whose fault it was, because blame does not prevent the next accident.

They ask what the crew believed when they acted, and what made it impossible to know otherwise. We ask the same question inside an organisation.

The organisation’s flight recorder // Specimen

Month 0Nothing on the record looks wrong yet.

Move along the record, month by month

Month 2An escalation route is set up with no deadline for a response.
Month 6A concern raised in public is answered badly. People learn to stay quiet.
Month 11The budget is set by a different team, on a different timetable.
Month 15The same reporting line is restructured for the third time. The approval limit does not move.
Month 22The trigger: the failure everyone sees.

Drag along the record. The failure is the last thing on it. The conditions were there long before. Illustrative: it describes no client.

An accident does not begin when something goes wrong. It begins when the system stops being able to see itself.

How the method works

Exhibit 03 // How it runs

Investigate First. Fix Second.

Every engagement runs in this order. A written finding comes first. Then the change that closes it, with an owner for every action.

Part 1 // The investigation

What happened, and why.

1Preserve the sceneCapture what happened before the story gets tidied.Before it starts: preliminary report
2Listen properlyConfidential interviews, from the leadership team to the front line.
3Trace the causeFollow each decision back to the conditions that made it likely.
4Score the defencesStrategy, communication, trust, accountability and escalation, each scored.Weeks 10–14: findings report

Part 2 // The FRANK intervention

What changes, and who owns it.

FFrame the realityThe facts in front of the team as one shared picture.
RReflect on alignmentWhat was done against what was agreed, without blame.
AArticulate the impactA number on it, so the board can act.
NName the adjustmentsWhat changes, who owns it and how anyone could tell.Weeks 12–16: alignment register
KKickstart realignment12 weeks of practice, tracked on the register.Then: embedding protocol

An investigation with no intervention is an expensive document. An intervention with no investigation is a guess.

See the 4 phases See the FRANK Method

Exhibit 04 // The defences

One Weak Defence Is Survivable

Every organisation leans on 5 defences: strategy, communication, trust, accountability and escalation. Most have one weak spot.

A failure gets through when the gaps in all 5 line up on the same path. Finding that path is the finding.

Line up the gaps

Stopped by strategy. The first defence still holds.

After the Swiss cheese model, James Reason, 1990.

Score your own defences in 4 minutes

Exhibit 05 // What it costs

A Fixed Scope and a Fixed Fee

The scope and the fee are agreed before the first interview. We do not quote a day rate. You are buying a finding, not a number of days.

The first meeting90 minutes with the executive who would sponsor the work.No fee
Investigation and correction12–16 weeks, ending in a written finding and the change that closes it.From US$35,000
Defences ReviewYour 5 defences scored before anything fails.From US$15,000
AI Decision ReviewPilotBefore approval, or after an automated decision goes wrong.From US$7,500
KeynoteFix the System, Not the Pilot, delivered by our founder.From US$5,000

Large groups and regulators: US$250,000 or more. Rand prices are on Engagements.

Answer 3 questions to find your route: start here.

If nothing changes, the problem continues. The first meeting costs nothing and ends with something written down.

Book the first meeting A decision in front of you, not a recurring problem? Frank Tsuro advises leaders one to one

The investigation // 12–16 weeks

The Same 4 Phases, Every Time

The 4 phases are the same in every engagement. That is what lets us tell you how your organisation compares with the others we have examined.

Book the first meeting See the 4 phases

No blame
Nothing attributed
Recorded word for word

The 4 phases

What Happens, and in What Order

Weeks 1–2
Preserve the Scene
We collect what already exists before anything moves: board packs, minutes, exit interviews, old survey data, the official organisation chart and the one people actually use. We also note ordinary things: which meetings get cancelled, how fast people reply and who gets called before a decision reaches a room.
Weeks 2–6
Listen Properly
One-to-one sessions across your leadership team and the layer below it. Nothing is attributed to anyone. Every session uses the same questions, so the answers can be compared. Nothing counts as fact until something independent supports it: a second person, a document or a number.
Weeks 5–9
Trace the Cause
A dated timeline of decisions and changes. We follow each symptom back 4 levels and separate the immediate trigger from the conditions that were already there. A failure like this usually rests on 7 to 12 small factors. A chain that stops at one has stopped too early.
Weeks 8–12
Score the Defences
Strategy, communication, trust, accountability and escalation, each scored from what we found. One weak defence is survivable, and most organisations have one. The finding is the path where all 5 have a gap at the same point, because that path has nothing left to stop it.

Want to know whether this applies to you? Start with 90 minutes.

Book the first meeting

What we listen for

5 Things We Listen For

Each one is about the system, not the person. The same questions in a different organisation produce a different set.

Session 07 // OperationsSpecimen

QWhen was the network decision made?

AWe decided in March. Well, it was decided in March.

QAnd the budget for it?

AYou would have to ask finance. I would rather not get into that.

AIt was unanimous. A pilot, I think. Or the first phase. It depends who you ask.

QDo people raise problems in the leadership meeting?

ANot since the town hall in 2021. Nobody wants to be the next example.

AI raised the capacity issue twice. Then I stopped.

Illustrative. In real work, nothing is attributed by name.

Choose a signal to see where it shows in the transcript.

What you receive

4 Documents, and One You Keep Using

Week 1
Preliminary Report
A written position on 3 things: whether this looks like a system problem, what an investigation would examine and what it would cost. If an investigation is not warranted, the first paragraph says so. US$2,500, credited against the investigation.
Read a sample report (PDF)
Weeks 10–14
Findings Report
Numbered findings, one to a page. Each carries the pattern, the evidence and the consequence. How many people we heard it from is in the sentence.
Weeks 12–16
Alignment Register
Every change scored for severity, likelihood and how easily it would be spotted. Each has one named owner and a completion test anyone could check.
The next 12 weeks
Embedding Protocol
A published schedule for withdrawing our support, with the register re-scored at each step. Your board keeps the register when we leave.

An investigation with no intervention is an expensive document.

Find Out Whether This Applies to You

4 minutes on your own, or 90 minutes with us. Both end with something written down.

Book the first meeting Take the 4-minute check

The check reads one person on one day. An investigation reads your whole leadership team. That is the difference between an indication and a finding.

The FRANK Method

One Method, 2 Halves, in Order

Every engagement has an investigation and an intervention, in that order. The investigation finds out what is actually happening. The intervention turns that finding into changes people commit to and are held to.

The structure

How the 2 Halves Fit Together

We are often asked whether the investigation is part of the FRANK Method or separate from it. Here is the structure.

Part 1 // Investigation

4 Phases. What Happened, and Why.

Preserve the scene, listen, trace the cause, score the defences. It produces a written position on what is actually going on, whether or not anything is done about it afterwards.

Part 2 // FRANK intervention

5 Stages. What Changes, and Who Owns It.

Frame, reflect, articulate, name, kickstart. This is the half with the acronym. It turns what the investigation found into commitments a board can hold someone to.

An investigation with nothing built on it is a document. An intervention with nothing under it is a guess.

FRANK names the second half. The full engagement is both, run in that order, every time.

Where it comes from

Air Accident Investigation, and One Observation about Power

Neither source is a management theory. That is deliberate. Management theory comes from the same field that produces the failures we are called in to examine.

Source 1 // Aviation

ICAO Annex 13, Aircraft Accident and Incident Investigation

Investigators may not ask whose fault it was. This is not kindness. Blame does not work. Replace the pilot, and the next pilot sits in the same seat, in front of the same faulty instrument, trained on the same rushed schedule. The accident happens again.

So investigators ask a different question: what did this person believe was true when they acted, and what made it impossible for them to know otherwise? Asked in a boardroom instead of a cockpit, that question is the whole method.

Source 2 // Power

What Changes When Rank Leaves the Room

When someone senior removes their title and steps into their own organisation as nobody in particular, the truth they hear changes at once. People were not lying before. Telling your boss bad news carries real risk, and removing the boss removes the risk.

We do not use this as a technique on client work. It is the evidence behind the method’s starting point: hierarchy filters the truth before it reaches the top.

The discipline comes from a field that proved blame makes outcomes worse. The starting point comes from what changes when hierarchy is removed. Neither idea is ours. Applying both to one leadership team, in a bounded and independent way, is.

The 6 disciplines

6 Named Disciplines, Put to 2 Jobs

The Swiss cheese model and situational awareness both come from human factors: the study of how capable people make sense of complex systems under pressure. Situational awareness explains what went wrong inside one person’s head. The Swiss cheese model explains why the system let it through.

AviationICAO Annex 13How to trace a chain of causes, with blame excluded from the terms of reference.

Adopted by the International Civil Aviation Organization in 1951 as Aircraft Accident and Incident Investigation. It is the international standard for investigating a crash.

Used in the investigation, phase 1

Human factorsSwiss cheese modelHow to map weaknesses across several defences and find where the gaps line up.

Set out by James Reason in 1990. Every defence has gaps. A failure happens when several gaps line up on the same path.

Used in the investigation, phase 4

Human factorsSituational awarenessHow to separate what a team saw, what it understood and what it expected next.

Set out by Dr Mica Endsley in 1988, in 3 levels: perceiving the data, understanding what it means and predicting what happens next. Most executive failures happen at the second level, not the first.

Used in the investigation, phase 2

Safety scienceRestorative just cultureHow to keep people talking once uncomfortable findings are on the table.

Developed by Sidney Dekker. It asks what the system needs to learn, not who deserves blame. That is what keeps a team honest once a finding becomes uncomfortable.

Used in the intervention, Reflect

Flight deck trainingCrew Resource ManagementHow to measure whether junior people can safely challenge senior ones.

Developed after well-trained flight crews kept making avoidable, fatal mistakes because a junior officer would not challenge a captain.

Used in the investigation, phase 4

Reliability engineeringFailure Mode and Effects AnalysisHow to score what is left over, by severity, likelihood and how easily it is spotted.

A register kept after the intervention closes. It scores each remaining risk, so a gap that starts to reopen is caught early.

Used in the intervention, Kickstart

The intervention, in full

5 Stages, in an Order That Does Not Change

Find out what is really happening. Then fix what is causing it.

Frame the Reality
We put the facts and the evidence in front of the team as one shared picture, so the conversation afterwards stands on solid ground.
Reflect on Alignment
We check whether what the team did matches what it agreed to do. The question is what the system was missing, not who is at fault.
Articulate the Impact
We put a number on it: delay, rework, replacement costs, slow decisions and penalty exposure. A board can act on a finding it can price.
Name the Adjustments
We agree exactly what changes, who owns each change and how anyone outside the room could tell it was done.
Kickstart Realignment
12 weeks of putting the changes into practice, tracked against the register, with our support withdrawn on a schedule you can see in advance.

Why this orderNaming a fix before agreeing the facts produces compliance, not change.

Ask a team to commit to new behaviour before it has accepted what happened, and you get agreement in the room and nothing different by Friday.

Frame comes first because a shared set of facts is the only thing in the room nobody can argue with. Articulate comes before Name because a team that has not seen the cost in real numbers has no reason to change anything specific.

A specimen alignment register

Ref	Change	Owner	Sev.	Det.	Status
A-01	Decision rights restated for capital spend above threshold	Chief Financial Officer	8	7	Monitoring
A-02	Escalation route published, with a stated time to respond	Chief Operating Officer	7	9	Open
A-03	Minutes record dissent by function, not by name	Company Secretary	5	4	Monitoring
A-04	Incident reviews separated from performance reviews	Chief People Officer	6	8	Closed
A-05	Escalation on capital variance routed through the Chief Financial Officer’s own office	Chief Financial Officer	9	2	Open

Severity and detectability are each scored 1 to 10. Severity is how bad the impact would be if the gap stays open. Detectability is how easily anyone outside would notice that it exists. Specimen only: the figures are made up and describe no client.

The distinction that decides everything

This is the one call in the method that a technical framework cannot make alone.

Systemic blindness

The Leader Could Not See It

The system filtered, softened or never delivered the information that would have changed the decision. This is fixable, and quickly. A leader who was working in the dark is usually the most motivated person in the room once the unfiltered picture is in front of them.

Wilful blindness

The Leader Saw It and Chose Not to Act

Acting would have cost something: a bonus, an ally or a difficult conversation. No investigation fixes this. It is a governance matter for the board, not for us. Treating it as a lack of information gives a choice the cover of a blind spot.

Telling the 2 apart takes a different reading of the room: incentives, alliances, and who protects whom and why. We pair that reading with evidence from a blame-free investigation, so the conclusion rests on verified fact, not instinct.

A Bounded, Independent Investigation, Then a Structured Intervention

12–16 weeks, a fixed fee agreed before it starts, and gone once the register is handed over.

Book the first meeting Questions people ask first Listen to the 22-minute overview

Cases // 6 investigations, disguised

What We Found, and What It Cost

Names, countries, dates and exact figures are altered. The mechanism, the sequence and the order of magnitude are not. Open any file for the full brief.

The case files

6 Organisations Other Organisations Depend On

In every one, a capable leadership team was producing a result it had not chosen. In none of them was anybody hiding anything.

Case 01 // Telecoms

3 Strategies, One RoomA listed telecommunications group. 3 versions of the strategy running at once, and no argument anywhere on record.US$38m a monthOpen the fileClose the file

Why we were called

Quarterly targets missed 3 times despite heavy network investment. Functions blaming each other. A restructure under consideration that would have removed 2 executives.

What we looked at

Board packs and minutes against the organisation chart people actually use. One-to-one sessions across the executive team and the layer below it.

What we found

3 members of the executive team were working to 3 different versions of the strategy. Disagreement between them appeared nowhere in the record.

What it was costing

Duplicated investment and resource conflict, modelled at US$38m a month. That number moved the conversation from alignment as a value to alignment as a line in the accounts.

What changed the argument: 14 decisions across 22 months, traced and dated. Each one made complete sense if you believed the version of the strategy that person believed.

Case 02 // Telecoms

An Escalation Route with No DeadlineA regional operator losing senior people across 2 cycles, with a separate explanation for each.R310mOpen the fileClose the file

Why we were called

Senior departures were read one at a time, each with its own story. Exit interviews cited fit and pace, and nothing specific.

What we looked at

The exit record treated as evidence, not as reporting. Escalation routes traced upward from the operating level, and how long each step took to respond.

What we found

A formal escalation route existed, with no deadline for a response. So issues moved informally instead, carried by people whose own bonus depended on the function they were escalating past.

What it was costing

Replacement, lost delivery and deferred network decisions, modelled at R310m across 18 months. Most of it sat in the delay, not in the recruitment.

What changed the argument: the same 3 names appeared in 9 separate accounts of why a decision had stalled, and none of them was in the minutes.

Case 03 // Energy

4 Years of Clean AuditsA utility whose internal audits were perfect. The chief risk officer thought the silence was too clean to be true.US$90mOpen the fileClose the file

Why we were called

4 years of internal audits passed with perfect marks. The chief risk officer could not point to anything wrong and did not believe the record.

What we looked at

5 years of audit findings against what people described in session. Where a concern had been raised, and what happened to the person who raised it.

What we found

A concern raised publicly 5 years earlier had been answered publicly and badly. The lesson the organisation drew was that raising a concern costs you something.

What it was costing

17 issues downgraded before they reached the executive layer, 3 of them still live, with combined exposure modelled at US$90m.

What changed the argument: the audits were accurate. They recorded what had been reported, and what had been reported was shaped by one incident 5 years earlier.

Case 04 // Mining

The Structure Changed. The Authority Did Not.A group that restructured the same reporting line 3 times, each announced as the fix for the last.R1.2bnOpen the fileClose the file

Why we were called

3 restructures of one reporting line in 4 years. Structure had been the answer often enough that somebody on the board began to doubt it.

What we looked at

A dated timeline of all 3 restructures, followed back 4 levels past the symptom, separating the immediate triggers from the conditions that came before all of them.

What we found

Decision rights were redrawn every time. What people were actually allowed to approve stayed exactly where it was. Each new structure inherited the old bottleneck intact.

What it was costing

Deferred capital decisions and a stalled production ramp, modelled at R1.2bn across 2 years. The restructures themselves were the smaller number.

What changed the argument: one approval limit had not moved since before the first restructure, and every stalled decision had passed through it.

Case 05 // Logistics

The Curated BriefA European-headquartered operator losing senior people before a refinancing, with nothing specific in any exit interview.€14m a yearOpen the fileClose the file

Why we were called

Senior departures accelerating ahead of a refinancing. Exit interviews citing cultural friction, with the specifics missing from every one.

What we looked at

The chief executive’s calendar against the pattern of what reached them. Sessions with the layer below, and the way problems were phrased when they arrived.

What we found

Problems raised at the top were met with an answer, not a question. Disagreement was met with questioning that ran until the other person gave up. The layer below learned to bring only the version that ended the conversation fastest.

What it was costing

Talent loss and decisions taken on curated data, modelled at €14m a year. Direct replacement cost was a small part of it.

What changed the argument: the behaviour was not hidden. It had been described internally as decisiveness. The cost was not the behaviour. It was what the organisation stopped telling itself as a result.

Case 06 // Banking

Signed Off 3 Times, Funded OnceA group that approved the same programme 3 years running and funded part of it each time.US$12m–18m a monthOpen the fileClose the file

Why we were called

A new core banking system, approved 3 years running and late every year. The executive committee had settled on the view that it was a supplier problem.

What we looked at

What the committee approved, set against what the budget actually released, year by year. Sessions with the committee and with the finance team that set the budget.

What we found

The committee approved the programme in full. The budget was set months later by a different team, on a different timetable, treating it as one request among many. Neither side knew the other was working from a different number.

What it was costing

Delay and rework on the new system, modelled at US$12m–18m a month, before counting what the missing capability was worth.

What changed the argument: neither team owned the contradiction. Both were doing their job correctly. The money went into the gap between them.

How these are published

What Is Altered, and What Is Not

Clients agree to this work on the understanding that what we find stays between us. So the cases are disguised, deliberately and consistently.

Altered on purpose	Exactly as found
The client name, in every case	The sector, so you can judge whether it is your world
The country and the region	The mechanism, which is the part that transfers
Dates, and the year of the work	The sequence of what we looked at and what we found
The size of the organisation and the exact figures	The order of magnitude, so the number still means something
Any detail that would identify a person	The fact that nobody involved was hiding anything

Where a client clears a figure in writing, we publish it with their name attached. Until then, the shape is real and the identifiers are not.

In all 6, the leadership team was capable. The information it was working from was not.

If one of these reads like your organisation, the first meeting costs nothing and ends with something written down.

Book the first meeting Read the scenarios

Engagements and fees // 2027

A Fixed Scope, a Fixed Fee and a Written Finding

You know the cost and the end date before we start. Every engagement is scoped, and the fee is fixed before the first interview.

Book the first meeting Find your route in 3 questions

From US$35,000
No day rates
No fee for the first meeting

What you can buy

4 Ways to Work with Us

Start here // No fee

The First Meeting

90 minutes with the executive who would sponsor the work. If an investigation looks warranted, a preliminary report follows: US$2,500, credited against the investigation. If it does not, we say so and the matter ends there.

Book the first meeting

Correction

Investigation and Correction

For a problem that keeps coming back.

From US$35,000

From R600,000 // Large groups and regulators: US$250,000 or more

You know why it keeps happening, in writing.
The fix targets the cause, so it holds.
Your board can act, because the cost has a number on it.

What is in it

The first meeting at no fee, then a credited preliminary report
A 12–16 week investigation and a written finding
The FRANK intervention, with owners and measures
12 weeks of tracked realignment
Scope and fee fixed before the first interview

Prevention

Defences Review

For when nothing has failed yet. A merger, a new strategy or a system going live is coming, or someone is asking how you know.

From US$15,000

From R210,000 // Scoped to your organisation

You see where your defences are weak before a failure shows you.
Your board gets a scored, plain-language picture it can act on.
Fixes are ranked by how serious each gap is, and how hidden.

What is in it

Confidential interviews across levels
Your 5 defences scored: strategy, communication, trust, accountability and escalation
A register of gaps, scored by severity and detectability
A board briefing with priorities
Optional support to close the gaps

For boards applying King V: the code asks the board to conclude on whether its governance outcomes were achieved. A scored register gives that conclusion evidence.

AI // Pilot 2027

AI Decision Review

Before your board approves an AI system, or after an automated decision goes wrong.

From US$7,500

Board session from US$7,500 // Review after a failure from US$12,000 (from R170,000)

Your board approves AI-assisted decisions knowing what it is accepting.
When something fails, you learn why, instead of blaming the nearest person.
You can show regulators and shareholders how oversight works.

See the AI Decision Review

Keynote

Fix the System, Not the Pilot

For leadership conferences, executive teams and boards. Delivered by our founder, Frank Tsuro.

From US$5,000

From R70,000 in South Africa // Travel is extra

The room learns to separate the trigger from the conditions.
It sees why a failure rarely has one big cause.
It leaves with a question to ask on Monday.

See the keynote

Also available

Realignment

The fix on its own, where a sound investigation already exists and the organisation has not been able to act on it. We check the existing findings first. If the diagnosis is wrong, the fix will be wrong too. By scope.

Register Oversight

Quarterly re-scoring of the alignment register, reported straight to the board, not through the executives it measures. Only after an investigation has produced the register. Retained.

Rand prices are for South African clients and exclude VAT where it applies. Travel and third-party costs are extra.

Fees by situation

What a Typical Situation Costs

Indicative fees for common situations. The first meeting confirms the scope, and the fee is fixed before any work starts.

Your situation	What we would suggest	Indicative fee
A problem has survived 2 restructures	Investigation and correction, 12–16 weeks	From US$35,000
The same problem across several countries or business units	Investigation and correction, wider scope	US$35,000–250,000
A multinational group or a regulator	Investigation and correction, large	US$250,000 or more
A merger, a new strategy or a system go-live is coming	Defences Review	From US$15,000
Your board is about to approve an AI system	AI Decision Review: board session	From US$7,500
An automated decision has gone wrong	AI Decision Review: 4–6 week review	Pilot from US$12,000
Good findings exist, but nothing changed	Realignment	By scope
A conference or board needs to stop looking for who to blame	Keynote	From US$5,000

The wider-scope fee is set by how many defences are in scope, and fixed before the first interview.

Find your route in 3 questions

Is this us

One Question Decides Whether We Are the Right Firm

Does somebody else’s ability to operate depend on you? If it does, the cost of a misaligned leadership team is carried by people who are not even your customers.

Where we have worked

Telecommunications
Energy
Mining
Logistics
Infrastructure
Construction
Banking
State-owned enterprises

Where the method transfers

Ports and terminals
Rail and freight
Water and sanitation
Payments and mobile money
Data centres and cloud
Downstream fuel
Healthcare

The second list is where the method works, not where we have worked. The test is the same in all of them.

Sector Is the Easy Part. The Condition Decides.

Worth investigating	We would decline
The same remedy has been tried more than twice	One function underperforming against a clear plan
Agreement in the room, and different funding outside it	2 people who need a mediator, not a finding
Good analysis already exists, and the problem is still there	A board wanting a document to support a decision already taken
A sponsor with the authority to act on whatever we find	A team that has already decided who is to blame
A failure here would be public within hours	A programme that needs delivery staff, not a diagnosis

4 things we agree first

Agreed Before a Fee Is Quoted

We do not quote day rates. A day is not the thing being bought.

Somebody senior enough to act on the finding is sponsoring the work.
Blame sits outside the terms of reference, in writing.
No restructuring during the first phase, because it changes the evidence.
You receive the whole findings report, including the parts that were uncomfortable to write.

The First Meeting Costs Nothing

90 minutes, questions rather than a presentation, and a written position on whether an investigation is warranted.

Book the first meeting See what we have found before

AI Decision Review // Pilot places for 2027

When an Automated Decision Goes Wrong

Boards are approving systems that make or shape decisions. When one fails, the board will ask what the business believed at the time. Most governance rules cannot answer that. This review can.

Ask for a pilot place See the 3 ways in

5 of 163. In a 2025 University of Cambridge study of the AI Incident Database, only 5 of the 163 reports tagged as responses met the definition of an official response from the company involved. Even those lacked the detail needed to learn from them.

Richards, Benn and Zilka, From Incidents to Insights, University of Cambridge, 2025. Read the paper

Blame tends to land on the nearest person, even when they had little control. Researchers call this the moral crumple zone.

3 ways in

Choose Where You Are

Settle the right question before approval. Learn properly after a failure.

Before approvalPrecisely Wrong: A Board SessionFor a board or risk committee about to approve an AI system.From US$7,500

A session for directors before they approve an AI-assisted decision or system. It settles the question to answer first: what the system will decide, from which records, and who answers when it is wrong.

The board leaves with a written standard for what it needs to see before it approves, and before it relies on the output.

After a failureA 4–6 Week ReviewFor an automated decision that has already gone wrong.Pilot from US$12,000

The discipline of our investigations, applied to one automated decision. We establish what the business believed at the time, how the decision was actually made and where the defences had gaps. The finding names conditions, not a scapegoat.

You receive a written finding, the changes with named owners and a record the board can show regulators and shareholders.

OngoingOversight, Reported to the BoardFor boards that rely on AI-assisted decisions every day.By scope

A regular review of how AI-assisted decisions are approved, monitored and corrected, reported straight to the board, not through the executives it measures.

Pilot places for 2027 are open to a small number of boards. Rand prices are on Engagements.

Governance context

Where King V Comes In

King V asks boards to oversee decisions that AI systems influence or make. We tie the review to the code only where it helps a board.

What the code asks, and what the review gives the board3 points, in brief.

King V asks

The board governs data, information and technology, AI included, with clear accountability for the decisions, actions, outputs and outcomes of AI systems (Principle 10).

The review gives

A dated record of who decided what, on what information.

King V asks

AI is used with ethics, human centricity, accountability, transparency, explainability, security, privacy, fairness and trustworthiness.

The review gives

A test of the decision against the values that matter for it.

King V asks

The board explains any recommended practice it has not adopted, and concludes on whether its governance outcomes were achieved.

The review gives

Evidence for that explanation and that conclusion.

King V was published by the Institute of Directors in South Africa in 2025. This is a summary for context, not legal advice.

Ask for a Pilot Place

Tell us about the decision you are about to approve, or the one that went wrong. The first meeting is 90 minutes, with no fee.

Book the first meeting Questions about AI and King V

The keynote // Delivered by Frank Tsuro

Fix the System, Not the Pilot

What aviation learned by refusing to accept human error as an explanation. A keynote from FranklySpeaking, delivered by our founder, Frank Tsuro.

Check a date Download the keynote sheet (PDF)

Keynote, 45–60 minutes
Keynote and workshop
Board session
Panel

The promise

The Audience Stops Looking for Who to Blame

They start looking for what made the mistake likely. That one change makes the next incident less probable.

The room leaves with

The Trigger and the Conditions

The difference between the immediate trigger and the conditions already in place, and what each costs to fix.

The room leaves with

Never One Big Cause

Why a serious failure usually rests on several small factors, not one big one.

The room leaves with

When Blame Stops Coming First

What changes in an organisation the day blame stops being the first question.

The talk

The Talk in 3 Movements

Pilot Error
For most of the 20th century, aircraft accident reports ended with pilot error. The next crew sat in the same seat, in front of the same instruments.
The Decision That Changed Aviation
The industry decided that human error would start an investigation, not end it. If a capable crew did something that killed them, the question became what made that action reasonable at the time.
Inside Your Organisation
Frank Tsuro applies the same discipline to business failure. He separates the trigger from the conditions, shows what each costs to fix and what a leadership team must do to hold the change in place.

A capable person doing something dangerous is a fact about the conditions, not about the person.

Formats

4 Ways to Bring It to Your Organisation

45–60 minutes

Keynote

For leadership conferences, executive teams and boards. The talk ends with a question your people can ask on Monday.

Half day

Keynote and Workshop

Applied to one of your own incidents, closed and safe to discuss. Every participant receives an editable workbook.

90 minutes

Board or Risk Committee Session

For directors who receive incident reviews. It ends with a written standard for the reviews the board will accept.

45–60 minutes

Panel or Moderated Conversation

As panellist or chair. 6 questions that open a real conversation, not a speech.

For operating businesses where failure is visible and expensive: energy, telecoms, mining, ports, logistics, construction, financial services and regulators.

How Booking Works

BeforeA 15-minute call on who is in the room and what they need to leave with. Then 2 or 3 background conversations with your people, 20 minutes each, in confidence.
On the daySlides 48 hours ahead, as PDF and PowerPoint. Frank works from the stage, not from a lectern.
AfterA one-page delegate follow-up with the 3 questions from the talk, yours to circulate.

Dates are held for 10 working days while you get approval.

Fee

From US$5,000

From R70,000 in South Africa. Travel is extra. Workshops and board sessions are scoped.

The Keynote Comes from Our Investigation Work

It gives a room the core of the method in under an hour.

Check a date Download the keynote sheet (PDF)

The 4-minute check

10 Questions. 4 Minutes.

Every organisation leans on 5 defences. This scores each one from your answers, then draws where your gaps are and whether they line up.

Start the check

4 minutes
Nothing stored
No sign-up

Where this sits

3 Steps. You Are on the First One.

You do not have to do all 3. If you already know you have a problem, start at step 2.

Step 1 // 4 minutes // You are here
The 4-Minute Check
10 questions, on your own, on screen. It ends with a drawing of where your gaps line up.
Step 2 // 90 minutes
The First Meeting
About your organisation, with whoever would sponsor the work. It ends with a written position.
Book the first meeting
Step 3 // 12–16 weeks
The Investigation
The full inquiry and the intervention that closes it. From US$35,000.
How it runs

The check reads one person on one day, so it is an indication, not a finding. A finding needs evidence from more than one source. That is what steps 2 and 3 are for.

The questions

Score Each One against the Last 6 Months

1 means this is not us. 5 means this is us. The next question appears on its own.

Your result

Your Drawing Appears Here Once All 10 Are Answered

One weak defence on its own is survivable. What matters is whether the gaps line up, because that path has nothing left to stop it.

Go to the questions

The first meeting // No fee

90 Minutes, and a Written Answer

Questions rather than a presentation. It ends with a written position on whether an investigation is warranted. If it is not, that is what we tell you.

Book the first meeting

No fee
90 minutes
Written outcome

How it runs

One Session with Whoever Would Sponsor the Work

Bring whoever has the authority to act on what we find. Without that person, we can describe the problem but we cannot tell you whether it can be fixed.

Before
Nothing to prepare. Board papers and minutes come later, under the engagement.
During
A consistent set of questions about what your team sees, what it understands and what it expects next.
After
A written position: whether this looks like a system problem, what an investigation would examine and what it would cost.

Pick a Time

The calendar opens in a new tab and takes about a minute. You can add a note about what you are facing.

Open the calendar

How it can end

3 Outcomes. 2 of Them End the Conversation.

Outcome
This Is Not Our Problem to Solve
We say an investigation is not warranted and show what we looked at to reach that view. We will not follow up.
Outcome
It Belongs to Somebody Else
The difficulty is real, and it is a different discipline. We tell you which, and what a good version of that hire looks like.
Outcome
An Investigation Is Warranted
A preliminary report sets out what we would examine, which defences are in scope, how long it takes and what it costs. US$2,500, credited against the investigation. You decide after reading it.

We need a sponsor with the authority to act on whatever the investigation finds.

4 Minutes, If You Want a Look First

The 4-minute check scores the same 5 defences from 10 questions. It is an indication, not a finding, and it is free.

Book the first meeting Take the 4-minute check

Ready to read

The Documents, as Delivered

The brochure, the paper a board receives, a preliminary report exactly as it was sent and an audio overview of the method. Read them before the first meeting, or instead of it.

Written to be tabled

Open, Read, Forward

Brochure // PDF
The Firm, in Brief
What we investigate, how the phases run, what you receive and what it costs.
Open the brochure
Board pack // PDF, 17 pages
The Paper You Put in Front of Your Board
What an investigation is, what it costs, what the board receives and a resolution it can adopt as written.
Open the board pack
Sample report // PDF, 13 pages
A Preliminary Report, as Delivered
A complete preliminary report, disguised, as it was sent after one 90-minute session.
Open the sample report
Audio overview // MP3, 22 minutes
The Method, in 22 Minutes
Generated from the FranklySpeaking documents. It describes the method. It is not a recording of a client engagement.
Play the audio overview
Keynote sheet // PDF, 5 pages
Fix the System, Not the Pilot
Abstracts, takeaways, formats and what your production team needs.
Open the keynote sheet
Board programmes // PDF, 12 pages
Programmes for Boards, from Our Founder
6 programmes from Frank Tsuro for boards, chairs and non-executive directors.
Open the board programmes

The First Meeting Costs Nothing and Ends in Writing

Book the first meeting Take the 4-minute check

Questions // Plain answers

What People Ask Before They Call

Short answers to what boards, executives and AI assistants ask about FranklySpeaking. If yours is not here, ask it.

Book the first meeting

Questions and answers

About FranklySpeaking

What does FranklySpeaking do?

FranklySpeaking investigates problems that keep coming back in high-stakes organisations, using a method adapted from aircraft accident investigation. Boards and chief executives receive a written finding: what is causing the problem, what it costs and what to change.

We offer investigations, correction and prevention, and a keynote that gives a room the core of the method.

Who is it for?

Boards and the executives who answer to them: board chairs and directors, audit and risk committee chairs, chief executives and chief operating officers, chief risk officers and heads of internal audit.

Most of our clients are organisations other people depend on, where a small failure costs millions.

How is this different from an audit, a consultant or a coach?

An audit checks what was reported against a standard. A consultant usually recommends a solution. A coach works with one person.

We investigate why a problem keeps coming back, and we write the finding before anyone proposes a fix. The scope is bounded, the work is independent and we leave once the register is handed over.

Is this the same as Frank Tsuro’s advisory work?

No. Frank Tsuro advises a leader, personally, on a decision. FranklySpeaking works on the organisation, with a team of associates and partners, when a problem keeps coming back.

If the question is a decision in front of you, start with Frank Tsuro’s advisory.

Who leads FranklySpeaking?

FranklySpeaking was founded and is led by Frank Tsuro. Investigations are delivered by FranklySpeaking’s team of associates and partners, to one method.

Where are you based?

Harare and Johannesburg. We work internationally.

Working with us

How do we know if this is worth investigating?

A rough test: ask 5 people on your leadership team, separately, what the strategy is right now. If you would get 5 different answers, there is something here.

The other test is cost. If you can put a real number against missed targets, people who have left or board time spent on questions nobody could answer, the first meeting is worth having. It exists to answer this question before either of us commits to more.

Why can’t our own chief risk officer or head of transformation do this?

Because they report into the structure that is hiding the problem.

A chief risk officer measures compliance against written policy, not the gap between policy and what people believe.
A transformation director owns delivery of the strategy, and is too close to tell the board the strategy itself is the problem.
A chief people officer reads sentiment well, but rarely frames a cultural finding as a business case.

This is not a criticism of those roles. You cannot ask someone to audit the hierarchy that sets their own bonus.

How do you get people to tell you the truth?

Confidentiality that is built into the method, not promised. Nothing is attributed by name, and nothing becomes a finding until a second, independent source confirms it.

We say at the start of every session that we are not there to evaluate anyone, and the structure makes that true.

Will people notice you and manage what they show you?

They would if we announced ourselves as investigators. So we do not. The work is described internally as understanding the current operating landscape. That is accurate, and it does not put anyone on the defensive before the first conversation.

What if we disagree with a finding?

Say so, and point to what you think we got wrong. Every finding is built from verbatim evidence and confirmed by more than one source.

Disagreement usually means one of 2 things: context we did not have, which we will revisit, or a fact that is uncomfortable but not wrong. The Frame stage exists for exactly that conversation.

Do you put our information into AI tools?

No confidential client information goes into any AI tool unless you have agreed and the tool’s terms allow it.

Fees and timing

What does it cost?

The first meeting: no fee. A preliminary report follows if an investigation looks warranted: US$2,500, credited against the investigation.
Investigation and correction: from US$35,000 (from R600,000). Large groups and regulators: US$250,000 or more.
Defences Review: from US$15,000 (from R210,000).
AI Decision Review: board session from US$7,500. Review after a failure from US$12,000 (from R170,000).
Keynote: from US$5,000 (from R70,000 in South Africa).

Rand prices exclude VAT where it applies. Travel and third-party costs are extra. See Engagements.

How long does it take?

An investigation and its intervention take 12–16 weeks, followed by 12 weeks of tracked realignment. An AI Decision Review after a failure takes 4–6 weeks. A Defences Review is scoped to your organisation.

Why don’t you charge a day rate?

You are buying a finding, not a number of days. The scope and the fee are fixed before the first interview, so you know the cost and the end date before we start.

What do we receive?

4 documents: a preliminary report in week 1, a findings report in weeks 10–14, an alignment register in weeks 12–16 and an embedding protocol for the 12 weeks after. Your board keeps the register when we leave.

What happens in the first meeting?

90 minutes with the executive who would sponsor the work. Questions, not a presentation. It ends with a written position on whether an investigation is warranted. If it is, a preliminary report sets out the scope and the fee.

AI and King V

What is the AI Decision Review?

A pilot service for 2027, for a small number of boards.

Before approval: a board session that settles what an AI system will decide, from which records, and who answers when it is wrong.
After a failure: a 4–6 week review of what the business believed at the time and where the defences had gaps, without blaming the nearest person.
Ongoing: oversight reported straight to the board.

See the AI Decision Review.

Does your work help with King V?

Where it genuinely helps a board. King V asks boards to govern technology and AI with clear accountability, and to conclude on whether their governance outcomes were achieved. The AI Decision Review and the Defences Review give those conclusions evidence.

We do not sell King V compliance work, and this is not legal advice.

Getting started

Where do we start?

With the first meeting: 90 minutes, no fee, and a written position either way, including the answer that there is nothing here. If you want a look first, take the 4-minute check.

Do you speak at conferences?

Yes. Our founder, Frank Tsuro, gives the keynote Fix the System, Not the Pilot, for leadership conferences, executive teams and boards. Keynotes start from US$5,000, or from R70,000 in South Africa. See the keynote.

Scenarios // The CEO blog

What Was Really Happening

One disguised situation a month, written for chief executives and boards: what it looked like, what was really happening and what fixes it.

By Frank Tsuro, founder

Scenarios

No. 01 // MiningThe Restructure That Changed Nothing3 restructures of one reporting line in 4 years. Each was announced as the fix for the last.

What it looked like

A mining group restructured the same reporting line 3 times in 4 years. Each new structure was announced as the fix for the last one. Production decisions kept stalling, and somebody on the board began to doubt the answer they kept being given.

What was really happening

Decision rights were redrawn every time. What people were actually allowed to approve did not move. One approval limit had not changed since before the first restructure, and every stalled decision had passed through it.

The deferred capital decisions and the slow production ramp were modelled at R1.2bn over 2 years. The restructures themselves cost less than that.

What fixes it

Not a fourth structure. The work is to restate who may approve what, at what level, and to give that change one owner and a test anyone could check. Then the board re-scores it until it holds.

The question for your board: which approval limit has not moved since your last restructure?

Disguised. Names, places, dates and exact figures are altered. The mechanism and the order of magnitude are not. Read the case file

No. 02 // BankingAgreed in the Room, Not Funded Outside ItA new core banking system, approved 3 years running and late every year.

What it looked like

A bank approved a new core banking system 3 years running. It was late every year. The executive committee had settled on the view that the supplier was the problem.

What was really happening

The committee approved the programme in full. The budget was set months later by a different team, on a different timetable, treating it as one request among many. Neither side knew the other was working from a different number.

Delay and rework were modelled at US$12m–18m a month. Neither team owned the contradiction, and both were doing their jobs correctly. The money went into the gap between them.

What fixes it

One owner for the gap between approval and funding. Both the committee and the budget team work from a single number, and the board checks the 2 against each other until they match.

The question for your board: what did we approve last year, and how much of it was actually funded?

Disguised. Names, places, dates and exact figures are altered. The mechanism and the order of magnitude are not. Read the case file

No. 03 // EnergyWhen Internal Audit Is Treated as the Police4 years of perfect audits, and a chief risk officer who did not believe them.

What it looked like

A utility passed 4 years of internal audits with perfect marks. The chief risk officer could not point to anything wrong, and did not believe the record.

What was really happening

5 years earlier, a concern raised in public had been answered in public, and badly. The organisation learned that raising a concern costs you something. After that, teams kept problems away from audit instead of solving them with it.

17 issues were downgraded before they reached the executive layer, 3 of them still live, with combined exposure modelled at US$90m. The audits were accurate. They recorded what had been reported, and what was reported had been shaped by one incident.

What fixes it

Separating learning from punishment. Incident reviews are kept apart from performance reviews, and the route for raising a concern gets a stated time to respond.

The question for your board: when did someone last raise a concern early here, and what happened to them?

Disguised. Names, places, dates and exact figures are altered. The mechanism and the order of magnitude are not. Read the case file

Coming in 2027

No. 04The Green Dashboard
No. 05The AI Decision Nobody Owned
No. 06The Succession Everyone Saw Coming

Recognise One of These?

The first meeting costs nothing and ends with something written down.

Book the first meeting Read the 6 cases

Find your route // 3 questions

3 Questions, Then the Right Next Step

Tell us what you are facing. You will see what we would suggest, what it usually costs and how to start. Your answers go into the booking, so you do not repeat yourself.

The 3 questions

This needs JavaScript. You can still book the first meeting.

Nothing is stored on this site. Your answers are added to the booking form only when you open it.

Page not found

This Page Is Not Here

The address may have changed. Everything is one click away.

Go to the home page The Investigation Cases Questions

Find the Cause. Fix the System.

It Keeps Coming Back

Agreed in the Room, Not Funded Outside It

The Reports Changed Nothing

Who Calls Us

The Same 4 Phases, Every Time

Preserve the Scene

Listen Properly

Trace the Cause

Score the Defences

Preliminary Report

Findings Report

Alignment Register

Embedding Protocol

Find Out Whether This Applies to You

One Method, 2 Halves, in Order

4 Phases. What Happened, and Why.

5 Stages. What Changes, and Who Owns It.

ICAO Annex 13, Aircraft Accident and Incident Investigation

What Changes When Rank Leaves the Room

Frame the Reality

Reflect on Alignment

Articulate the Impact

Name the Adjustments

Kickstart Realignment

The Leader Could Not See It

The Leader Saw It and Chose Not to Act

A Bounded, Independent Investigation, Then a Structured Intervention

What We Found, and What It Cost

Why we were called

What we looked at

What we found

What it was costing

Why we were called

What we looked at

What we found

What it was costing

Why we were called

What we looked at

What we found

What it was costing

Why we were called

What we looked at

What we found

What it was costing

Why we were called

What we looked at

What we found

What it was costing

Why we were called

What we looked at

What we found

What it was costing

A Fixed Scope, a Fixed Fee and a Written Finding

The First Meeting

Investigation and Correction

Defences Review

AI Decision Review

Fix the System, Not the Pilot

Realignment

Register Oversight

Sector Is the Easy Part. The Condition Decides.

The First Meeting Costs Nothing

When an Automated Decision Goes Wrong

Ask for a Pilot Place

Fix the System, Not the Pilot

The Trigger and the Conditions

Never One Big Cause

When Blame Stops Coming First

Pilot Error

The Decision That Changed Aviation

Inside Your Organisation

Keynote

Keynote and Workshop

Board or Risk Committee Session

Panel or Moderated Conversation

How Booking Works

The Keynote Comes from Our Investigation Work

10 Questions. 4 Minutes.

The 4-Minute Check