Skip to content

Method

How we diagnose an entire company

Four weeks. Twenty dimensions. Nothing we can't back up with a document or a system query.
On this page
  1. 01Why don't you sell a standalone AI agent?
  2. 02How do you measure the state of a company?
  3. 03What is the evidence rule?
  4. 04What is the Misalignment Index?
  5. 05Do you only measure those four areas?
  6. 06Who do you interview?
  7. 07What information do you need from my company?
  8. 08How long does it take and how is it organized?
  9. 09What do I get at the end?
  10. 10Do you review the AI we already use?
  11. 11Do you use AI in your own work?
  12. 12What happens if a system you build makes a mistake?
  13. 13How do you handle legal and regulatory matters?
  14. 14What happens if the diagnostic concludes I shouldn't hire you?
  15. 15How much does it cost?

Why don't you sell a standalone AI agent?

Because every company runs on four connected areas (sales, marketing, administration and fulfillment) and a real change in one forces a change in the other three. An agent that optimizes one piece without touching the rest doesn't fix the business. It just moves the problem somewhere else.

  • 01

    If an agent takes over your first line of sales, marketing has to deliver better-qualified leads from the start, because the agent doesn't have the human judgment to rescue one that came in wrong.

  • 02

    If you speed up the sales cycle, operations needs the capacity to deliver at the same pace, or speed becomes the new bottleneck.

  • 03

    If administration changes the pricing policy, the sales pitch has to change the same day.

What breaks isn't the tool. It's the promise the tool helped make.

How do you measure the state of a company?

With twenty dimensions: four areas times five identical dimensions in each one (process, data, tools, people and results). Each is scored from zero to four on the same scale, which makes them comparable with each other and across companies.

Process
Is there a written sales process, with defined stages and clear criteria to move between them, and is it followed as written?
Data
Is every opportunity logged with its source, stage, value and date, and are those records reliable enough to decide on?
Tools
Does the CRM support the real sales process, or does the team fill it in out of obligation and actually work somewhere else?
People
Can a new salesperson reach the standard, or do results depend on one or two irreplaceable people?
Results
Is conversion by stage measured against a target, with deviations corrected on a set schedule?

The five questions are the same in all four areas, which is why the results are comparable. The evidence rule applies equally to all twenty: a 3 or a 4 requires a document or a system query. An interviewee's word caps at 2.

The scale, the same across all twenty dimensions:

  1. 0

    Nonexistent

    No process, data, tool or measurement. Things get solved by reacting.

  2. 1

    Heroic

    It works because one person holds it up. If that person is out, it falls apart.

  3. 2

    Documented

    It exists and it's written down, but nobody measures whether it's followed.

  4. 3

    Managed

    It's measured, reviewed on a set schedule and corrected when it drifts.

  5. 4

    Optimized

    It improves on the measurement, and much of it is reliably automated.

What is the evidence rule?

To score a 3 or a 4 we require a document or a system query. What an executive tells us in an interview, however senior, caps at 2. That rule is what makes two different consultants reach the same number for the same company.

Why it matters more than it seems. It's the difference between a measurable product and a well-written opinion. Without it, the result depends on who ran the diagnostic and how convincing the manager was in the meeting.

And it has an uncomfortable consequence we'd rather tell you now: your company will probably score lower than you expect. Not because it works badly, but because plenty of things that work aren't documented or measured. That finding alone often justifies the exercise.

A beam of light through a high window in an archive room
If we can't prove a statement, we don't write it. And if something couldn't be measured, we say so and explain why. A table with three honest "not measurable" entries is worth more than a full table with three made-up numbers.

What is the Misalignment Index?

It's the distance between your strongest area and your weakest. Once it goes above 1.5 points, investing in the strong area makes the business worse, not better. It's the finding that most often changes a client's decision, and nobody else measures it.

Four stacks of paper of uneven height against a wall
IndexReading
0.0 – 0.7Aligned. All four areas move at the same pace. You can optimize on any front.
0.8 – 1.4Tension. There's a gap that's already costing you money, even if it doesn't show on the P&L.
1.5 or moreCritical. The strong area is creating demand the weak area can't sustain.

We also deliver the propagation matrix: for every intervention we propose, its effect on the other three areas. It's what keeps fixing one thing from breaking two others.

Get your Index in eight questions

Eight questions, instant result. It's a self-reported estimate, not a measurement: here we only have your word, and in the real diagnostic we require evidence.

Do you only measure those four areas?

No. Four cross-cutting layers run through the four areas and are scored separately: finance, talent, technology and data, and legal and regulatory. The areas are the operating view; the layers are what holds them up or drags them down underneath.

They exist because, in front of a large industrial company or a bank, saying that administration and fulfillment cover half the business is a simplification an operations director will push back on in the first meeting, and rightly so. The layers let us take the same instrument from a 25-person company to a 2,000-person one.

Who do you interview?

The head of each area and, without exception, at least one person who does the work and doesn't manage it. The gap between the process the manager describes and the one the person doing it every day describes is one of the most valuable findings of the method, and we report it explicitly.

Between six and twelve interviews, depending on size. Always recorded with each person's consent, always transcribed and always coded against the twenty dimensions. And when you allow it, two or three short conversations with your own customers: twenty minutes that often change how we read an entire area.

The manager describes the process they designed. The person doing it describes the process that exists. They're rarely the same, and the difference is money.

What information do you need from my company?

Twelve to twenty-four months of sales history, read-only access to the systems you already use, two years of financial statements, your cost structure, your operation's cycle times and defect rate, and an inventory of systems with their monthly cost. We ask once, with five business days to deliver.

And if you don't have some of it, that works too. Whatever doesn't arrive is marked as not available, and that gap is itself a scored finding. A company that can't pull its defect rate in five days has already told us something important about the "data" dimension of its fulfillment.

Everything comes in under a mutual non-disclosure agreement and a data processing agreement, both signed before we touch a single file.

How long does it take and how is it organized?

Four weeks from signature, in five phases. On your side, we need one person with time set aside as the operational counterpart, and each area head available for one interview.

A long corridor with doors at regular intervals
  1. Before we start

    Qualification

    Twenty minutes with the decision-maker to define where to start

  2. Week 0 – 1

    Kickoff

    Contract, confidentiality, information request and system access

  3. Week 1 – 2

    Fieldwork

    Interviews, live observation of two processes, document review, digital audit and AI inventory

  4. Week 2 – 3

    Analysis

    Scoring of the twenty dimensions, the Index, the opportunity matrix and the vulnerability study

  5. Week 4

    Delivery

    A two-hour session with the decision-maker. Findings, roadmap and a joint decision on the path forward

What do I get at the end?

A 25 to 40 page document: the map of your four areas with the twenty scores and their evidence, the Misalignment Index, the gap between what's reported and what's real, the prioritized opportunity matrix with its risk analysis, and a roadmap of no more than twelve initiatives.

Each initiative comes with an objective, a success metric with a measured baseline, a named owner on your side, an investment range, time to first result and residual risk.

A spiral-bound report, closed, on a wooden desk

Two rules of the deliverable worth knowing before you buy it:

  • No initiative goes in without a measured baseline. If the starting point couldn't be measured, the first initiative on the roadmap is to instrument that measurement. We don't promise a thirty percent improvement on a number nobody knows.
  • Every opportunity comes with at least one alternative that doesn't involve us. A tool you can buy directly, a capability you can build in-house, or another provider. It costs us sales, and it spares us the awkward conversation about whether the diagnostic was written to sell you something.

Do you review the AI we already use?

Yes, and it's included. We inventory the AI systems your company runs, including the ones someone switched on without telling anyone: a chat widget on the website, a voice agent on the phone line, a text or image generator in marketing. Then we check what each one requires under the rules that apply to you.

It's the part of the diagnostic that surprises clients most often, because almost no company has that list. In the US, an AI-generated voice on outbound calls falls under the TCPA, several states require bots to disclose that they're bots, and state privacy laws reach the data those systems collect.

Do you use AI in your own work?

Yes, heavily, and it's why a diagnostic that takes a traditional consultancy eight weeks takes us four. A team of specialized agents handles document review, the digital audit, data analysis and first drafts.

And this is the line we don't cross: judgment, final scores and the recommendation are always human. The agents produce candidates with their evidence. A person validates, decides and signs. Every diagnostic records which version of the system produced it, the same discipline we require of anything we put into production for you.

The most honest way to prove what we sell is to use it on ourselves first.

What happens if a system you build makes a mistake?

Everything any agent we build does goes into an audit log you can query and export. There's never a black box. And every agent we put into production has rules for handing off to a person, designed in from the start, not bolted on later.

Before it goes live, it goes through adversarial testing: we try to break it ourselves (prompt injection, edge cases, harmful answers) before one of your customers does.

What happens if the diagnostic concludes I shouldn't hire you?

We refund half of what you paid and you keep the full deliverable and the roadmap, with no restrictions on use. We put it in writing here because a diagnostic that can only end in "yes, hire us" isn't a diagnostic. It's a sales proposal you paid for in advance.

There's also step 01, Prepare, for when the problem is process, data or governance and technology can't solve it yet. It's not a consolation prize. It's what makes the AI conversation worth having afterward.

How much does it cost?

The diagnostic is paid upfront and fully credited toward whatever you hire us for next, if you do it within sixty days of delivery. We give you the figure in the twenty-minute conversation, together with the exact scope for your company.

We don't publish rates because a number without a scope means nothing, and the scope of a diagnostic for a 25-person company and a 300-person company isn't the same. What we do publish is the condition: if we tell you no, we refund half.

If you've read this far, you already know how we work.

What's left is finding out whether it makes sense to work together. That takes a five-minute application and a twenty-minute conversation.

Nine questions. Instant answer. No commitment and no sales call.

Start your diagnostic