Innovation & Capability7 min read·By , Independent AI Advisor & Fractional AI Lead·

What an innovation strategy firm should leave behind

A consultant's best workshop is still a poor investment if the organization needs the consultant to run the next one. Here's how to judge the actual transfer of capability.

The first workshop usually goes well. There are opportunity maps, thoughtful questions, a few surprising connections and a wall covered in possible futures. People come out of the room with the welcome feeling that an organization stuck in its routines has started thinking again.

Six months later, a less glamorous question becomes unavoidable. Can the team make the next important innovation decision without booking the same consultants?

This is the test I would apply to innovation strategy firms. Their work can be genuinely valuable, especially when a company is entering an unfamiliar market, questioning an established business model or trying to understand how an emerging technology might change its economics. But an engagement should be judged by what the client can decide and do afterward, not by how convincingly the future was described during the engagement.

The point is not to make external expertise unnecessary. It is to know when expertise has become a capability rather than a dependency.

What does an innovation strategy firm actually do?

An innovation strategy firm helps an organization choose where to create new value and how to test whether that value is real. The work can include research into unmet customer needs, technology assessment, portfolio choices, business model design, prototypes and the operating changes needed to bring an idea into use.

Not every consultancy approaches the problem in the same way. A research-led firm may excel at identifying emerging behavior; a design practice may be particularly strong at reframing the customer problem; a technically oriented team may uncover what is feasible inside a difficult legacy environment. Large consulting organizations may coordinate complex changes across many functions. Small specialist partners may be more effective when the question needs direct involvement from people who can both frame and build.

None of those descriptions is a ranking. They are reasons to define the problem before choosing the partner.

A firm that produces excellent foresight may not be the right team to implement a product. An engineering partner who can build quickly may not be the right partner to challenge whether the product should exist. The most expensive mismatch is hiring a company because its portfolio looks impressive, without specifying the decision you need its work to improve.

The first deliverable is a better decision

Imagine an established service business concerned that AI will compress the value of its current offering. It could ask consultants for a generative AI opportunity roadmap. The likely result would be a list of plausible use cases, platform options and implementation phases.

A more useful starting brief is narrower: Which part of the customer relationship is becoming easier to substitute, which part still commands a premium, and what evidence would justify changing the service model?

That brief cannot be answered with an inventory of tools. It requires understanding the customer, the unit economics, what people are actually willing to pay for and which assumptions the current business has never tested.

A good engagement should make the central uncertainties visible and turn them into choices. What are we willing to stop doing? Which customer segment will we prioritize? How much money are we prepared to put at risk before a test has earned further investment? Who has authority to change the offer if the evidence disagrees with the original strategy?

Those decisions remain the company's responsibility. A partner can structure them and make the evidence stronger, but cannot outsource the consequences.

This is also why a presentation full of exciting possibilities can be less useful than a short list of rejected ones. Each rejection should record the assumption or constraint behind it, so the company knows when new information would justify reopening the decision.

What to look for when comparing firms

Rather than asking which firm has the most impressive methodology, I would compare four concrete capabilities.

Problem framing. Ask how the firm distinguishes a real customer or business problem from a fashionable technology opportunity. The team should be able to challenge the brief without turning every assignment into an undefined exploration.

Quality of evidence. Ask what information would change the recommendation. Good research identifies the people and situations it covers, the limitations of the sample and the difference between an observed behavior and an executive's preferred story. A confident slide without a traceable evidence trail should count against the proposal.

Ability to run a meaningful test. A prototype is not automatically an experiment. A useful test specifies an assumption, a credible comparison, a measure that reflects the real outcome and a decision threshold agreed before the results arrive. Sometimes the right test is a service offered manually to a small group, not a software product.

Transfer of capability. Ask who on the client's side will own the methods, data, decisions and next iteration. The partner should be willing to train a , document what did not work and make the client less dependent on its own presence.

A beautiful deck might be evidence of communication skill. It is not evidence of these four capabilities.

A practical way to structure an engagement

Consider splitting a proposed assignment into three decisions, rather than committing up front to one long transformation program.

The first decision is whether the problem is worth solving. Use a short discovery period to establish current costs, customer evidence, constraints and the credible alternatives. Some opportunities should end here. That is a successful outcome if the company avoided an expensive distraction.

The second decision is whether a proposed intervention produces enough value to deserve a real test. At this stage, specify the behavior you expect to change and compare at least one low-complexity option with the more ambitious proposal. For an AI-related project, this might mean comparing a redesigned form and simpler routing rule with a model-enabled workflow. Use the same outcome measure for both.

The third decision is whether the organization can operate what it has learned. Who will maintain it? Can another team repeat it? What will the new process displace? What does the ongoing supervision cost? If the company needs permanent external rescuers to keep a alive, the business case is incomplete.

These gates do not need to become a new bureaucracy. Their purpose is to make uncertainty cheaper to resolve, not to create ceremonies around it.

I have written elsewhere about why the second team is the real innovation test. A promising pilot inside one exceptionally motivated team does not prove that the organization has acquired a repeatable capability.

The danger of innovation theatre

Some engagements reward the appearance of movement: idea counts, workshop attendance, demo days and prototypes. Those can be useful inputs, but they should not be mistaken for business outcomes. An organization can produce more experiments while becoming no better at deciding which ones to stop.

The same applies to a portfolio of emerging technologies. Artificial intelligence, spatial computing and automation may belong in the conversation when they solve a concrete problem. Their presence in a presentation is not itself a strategic advantage.

Where AI systems introduce significant risk, governance should be part of the operating design, rather than a paragraph added after the product roadmap. ISO/IEC 42001 specifies requirements for establishing and improving an AI management system. It is a useful reference for organizational responsibilities, but simply citing the standard is not proof that a consultancy's proposed system is safe or compliant.

The strongest signal of a good partner is often the willingness to recommend a smaller intervention, or none at all. That may be less exciting than a new innovation lab. It is more credible.

Five questions worth putting in the brief

Before signing, ask the shortlisted firms to answer these questions in writing:

  1. What exact decision should be better at the end of this assignment?
  2. Which assumption would you test first, and what result would make you change direction?
  3. What would your team advise us not to build or buy?
  4. Which assets, methods and will belong to our people after handover?
  5. What would count as a successful early termination of the engagement?

Their answers will tell you more than another capability deck. They reveal whether the firm sees innovation as an ongoing consulting activity or as a capacity the client should develop.

If your organization needs an external partner, the right one can accelerate learning and make unfamiliar choices more disciplined. A specialist innovation and AI advisory engagement can be valuable on precisely those terms. The commercial relationship becomes stronger when the client is free to disagree and when the work can survive the consultant's departure.

The measure of a good innovation strategy firm is not how many options it can put on the wall. It is how clearly the organization can explain the few choices it made, what the evidence showed and what it is now capable of doing without assistance.

Get new essays as they publish.

One email per essay. No noise between.

Berk Bayri

Independent AI Advisor & Fractional AI Lead

I help leadership teams decide where AI belongs, test it before it scales and build the teams that run it, drawing on 25+ years of building digital products, experiences and capabilities.