Scout Addison
Contact

FAQ

Straight answers about AI, with sources.

The questions leaders ask us most, answered directly. Every figure links to the research behind it, so you can check it yourself.

Reviewed September 29, 2026

Getting value from AI

Why hasn’t AI spend improved profit?

Usually because the value leaks somewhere between the tool and the P&L. In McKinsey’s 2026 survey, 60% of respondents expected their organizations to increase AI investment over the next year and 80% said AI had made them personally more productive, yet only 37% reported any positive effect on EBIT.

The leaks tend to be the same: workflows nobody redesigned around AI, time saved that goes into more of the same work, and results that no one owns or measures.

Sources: McKinsey, The state of AI in 2026 (Aug 2026) (opens in a new tab)

How we help: AI at Scale

How do you measure the ROI of an AI initiative?

Measure the work before AI touches it. Record a baseline for the workflow it should change, such as time per case, cost per case, error rate, and volume. Afterwards, track the same numbers alongside what the AI costs to run.

Few organizations do this well. Only 25% of AI initiatives have delivered the ROI expected of them, according to IBM’s 2025 CEO Study, and only 11% of CEOs in EY’s 2026 survey said AI’s impact is linked to financial reporting and reviewed regularly by senior management.

Sources: IBM Institute for Business Value, 2025 CEO Study (opens in a new tab); EY, CEO Outlook 2026 (opens in a new tab)

How we help: AI at Scale

From pilot to production

Why do AI pilots fail to reach production?

Most don’t fail on the model. They stall on what surrounds it: data that isn’t ready, integration with existing systems, security and legal review, and no business owner accountable for the result.

Deloitte’s 2026 research found only 25% of organizations had moved even 40% of their AI pilots into production. In MIT’s 2025 research, of organizations that evaluated task-specific AI tools, 20% reached a pilot and 5% reached production.

Sources: Deloitte, State of AI in the Enterprise, 2026 edition (opens in a new tab); MIT NANDA, The GenAI Divide: State of AI in Business 2025 (opens in a new tab)

How we help: AI at Scale

What does “AI-ready data” mean?

Data an AI system can use to give correct answers. It has to be accurate and accessible, and described well enough that the system knows what your terms mean, how your records relate, and which business rules apply.

Gartner predicts that through 2026, organizations will abandon 60% of AI projects that aren’t supported by AI-ready data. Confidence often runs ahead of reality: in a 2026 Drexel LeBow and Precisely study, 88% of data leaders said their data was AI-ready, while 43% named data readiness as their biggest obstacle to AI.

Sources: Gartner, Lack of AI-Ready Data Puts AI Projects at Risk (Feb 2025) (opens in a new tab); Drexel LeBow and Precisely, 2026 State of Data Integrity and AI Readiness (opens in a new tab)

How we help: AI at Scale

How long does it take to move an AI pilot into production?

It depends on how accessible the data is, how complex the integration is, and how much time your team can give it. MIT’s 2025 research gives a sense of the range: the top-performing mid-market companies reported about 90 days from pilot to full implementation, while large enterprises took nine months or longer.

The practical move is to set milestones around the specific blockers you find first, rather than a standard timeline.

Sources: MIT NANDA, The GenAI Divide: State of AI in Business 2025 (opens in a new tab)

How we help: AI at Scale

Should we build AI in-house or work with a partner?

Either way, your own team should end up owning it. The real question is how quickly you get there. In MIT’s 2025 research, external partnerships reached deployment about 67% of the time, compared with about 33% for tools built internally. Those figures are self-reported.

A partner who builds alongside your engineers can shorten the path without leaving you dependent on them.

Sources: MIT NANDA, The GenAI Divide: State of AI in Business 2025 (opens in a new tab)

How we help: AI at Scale

People and decisions

Why aren’t employees getting more value from the AI tools we’ve rolled out?

Access usually isn’t the issue. In BCG’s 2026 survey, 74% of frontline employees used AI, but only 36% of employees felt adequately upskilled, and more than half weren’t reinvesting the time they saved into more strategic work.

Microsoft’s 2026 Work Trend Index found organizational factors (culture, manager support, talent practices) account for more than twice the AI impact of individual factors, 67% versus 32%. The fix usually lies in how the work is designed and who owns it. More licences rarely help.

Sources: BCG, AI at Work 2026 (Jun 2026) (opens in a new tab); Microsoft, 2026 Work Trend Index (May 2026) (opens in a new tab)

How we help: Org Transformation

Who should own AI decisions in a company?

Each AI use needs one named business owner who is accountable for its result, plus agreed rules for who approves a new use and who signs off on its risk.

Many organizations haven’t settled this. In Grant Thornton’s 2026 survey, 78% of executives lacked full confidence they could pass an independent AI governance audit within 90 days, and only 52% of boards had set clear AI governance expectations.

Sources: Grant Thornton, 2026 AI Impact Survey (Apr 2026) (opens in a new tab)

How we help: Org Transformation

Governance and security

How should a company govern AI without slowing it down?

Split it into three parts and give each an owner. Technical controls such as access, data privacy, testing, and monitoring belong in the build. Decisions about who approves a new use and who signs off on risk belong in a short process people will actually follow. Customer-facing AI needs its own rules: how accurate it must be, what it must never do, and what happens when it gets something wrong.

The gap is real: only 21% of companies deploying AI agents report a mature model for governing them, according to Deloitte’s 2026 research.

Sources: Deloitte, State of AI in the Enterprise, 2026 edition (opens in a new tab)

How we help: Org Transformation

AI in your product

Why don’t customers use the AI features we ship?

Often because the launch stopped at the code. Product leaders in the 2026 CPO Insights Report named speed to market as their top internal challenge, and said products now get stuck at go-to-market execution, customer adoption, and organizational alignment rather than at the build.

Features also struggle when they’re chosen for how they demo instead of a customer problem. Set an adoption target before launch, plan pricing, sales enablement, and support alongside the build, and track use afterwards.

Sources: 2026 CPO Insights Report, Products That Count and Mighty Capital (May 2026) (opens in a new tab)

How we help: Product Strategy and Execution

How do you decide whether an AI feature is worth building?

Test it on real customer data before committing to the full build. Measure its accuracy against real cases and estimate its running cost per customer. If it misses the accuracy bar, or the numbers don’t work at your price, build a narrower version or skip it.

Only 5% of custom enterprise AI tools in MIT’s 2025 research reached production. Most failed on brittle workflows and poor fit with day-to-day work, which early testing on real data tends to expose.

Sources: MIT NANDA, The GenAI Divide: State of AI in Business 2025 (opens in a new tab)

How we help: Product Strategy and Execution

Tell us where it's stuck.

A few lines about what’s blocking it is enough to start. We’ll reply to set up a conversation.

Get in touch

Privacy controls

Cookie settings

Analytics is optional. Your choice is stored on this device and can be changed at any time.