95% of Your Consulting Firm's Best Slide Decks Are Locked Away. Here's Why.

Last updated: March 2026

I spent six years at Kearney. In that time, I watched hundreds of brilliant slide decks get created — sharp analyses, original frameworks, sector deep-dives that took weeks to build. Every one of them ended up in a project folder that almost nobody else in the firm could access.

Not because the work wasn't valuable. Because it was confidential.

This is the single biggest structural problem in consulting knowledge management, and almost nobody frames it correctly. It's not a search problem. It's not a taxonomy problem. It's the fact that 98–100% of usefully reusable material in consulting is confidential — and the tools we have for managing confidentiality weren't built for that ratio.

The best work your firm has ever done is sitting in a project folder somewhere

Every consulting firm has the same experience. A partner mentions a brilliant deck from a project two years ago — a market sizing model for a healthcare client, or a due diligence framework that cracked a tricky valuation. The team that built it has moved on. The deck sits in a project archive with restricted access.

The knowledge isn't lost. Everyone knows it exists. But it's locked behind confidentiality, and there's no practical mechanism to extract the insight while removing the sensitivity.

Think about what's actually in a consulting slide deck. Client names and logos on every page. Revenue figures tied to specific business units. Strategic recommendations that would move share prices if they leaked. Competitive benchmarking that names the client's rivals and their performance. Interview quotes from the client's leadership team.

The insight — the analytical framework, the structure of the argument, the methodology — is tangled up with the confidential detail at every level. You can't just pull out the good bits. The good bits are the confidential bits.

One procurement advisory firm I've spoken with has a backlog of 3,000 projects sitting untouched. The knowledge is there. Nobody can get to it.

Why document-level permissions don't work when every deck is confidential

Most firms manage sensitivity at the document level. Tag the document, classify it, apply permissions. Restrict access to the engagement team and maybe a few senior partners. This works perfectly well when a small proportion of your content is sensitive — you fence off the sensitive stuff and let people access everything else.

But in consulting, the ratio is inverted. It's not 5% sensitive and 95% open. It's 98–100% sensitive and essentially 0% freely accessible. When everything is confidential, document-level permissions don't manage sensitivity — they prevent knowledge from circulating at all.

The result is that firms end up with two categories of content: restricted project deliverables that contain the real knowledge, and sanitised thought leadership that's been scrubbed of anything proprietary. The thought leadership is accessible but generic. The project work is specific but locked.

Only 5–15% of institutional knowledge is actually accessible for AI or reuse in most consulting firms. That's not a guess — it's a pattern I hear in every conversation with knowledge management leaders. The number varies by firm, but the shape of the problem is always the same.

The real cost: recreated slides, underperforming AI, missed insights

This has direct, measurable consequences.

Knowledge workers spend roughly 2 hours per week recreating work that already exists somewhere in their organisation, according to APQC research — a pattern we explore in detail in the hidden cost of manual document sanitisation. In consulting, where deliverables are complex and time-intensive, that figure is probably conservative. A team in London builds a market entry framework from scratch while an identical one sits in a Singapore project folder they can't access.

Then there's the AI problem. Every major firm is investing in AI-powered knowledge tools — RAG pipelines, internal copilots, search systems built on top of their document repositories. But as we discuss in building AI-ready knowledge bases, those systems are only as good as the content they can access. If your AI copilot runs on 5–15% of your firm's actual knowledge, it's operating on scraps.

I've spoken with firms that have spent millions on AI infrastructure, only to find that the system keeps returning the same thin layer of sanitised case studies and published thought leadership. The real knowledge — the detailed analyses, the proprietary frameworks, the sector-specific models captured in PowerPoint decks — never makes it into the AI's reach.

The irony is painful: the firms investing most heavily in AI are the ones most exposed to this problem, because they're the ones that notice the gap between what their AI could do and what it actually does.

This isn't just consulting

The same structural problem appears wherever high-value knowledge is created under confidentiality constraints.

Legal services. Every matter file is confidential. The brilliant brief a partner wrote for a pharmaceutical patent dispute three years ago? Locked in a matter folder. The precedent research is reusable; the client details are not.

Financial services. Deal documents, investment memos, due diligence reports — all contain the kind of detailed analysis that would be enormously valuable across the firm. All confidential to the specific engagement.

Government and defence. Classification constraints mean that analytical work products — assessments, briefings, strategy documents — can't be reused or even referenced outside their classification level. The knowledge exists but the clearance model prevents circulation.

Different flavour, same structure. High-value knowledge, created under confidentiality, with no practical mechanism to extract the insight from the sensitivity.

What this means for firms trying to deploy AI on their knowledge base

Here's the uncomfortable truth: you can't build an effective AI assistant on a knowledge base that's 5% populated.

The consulting firms I talk to all describe the same trajectory. They invest in an AI-powered knowledge platform. They feed it their "safe" content — published reports, anonymised case studies, internal templates. The system works, technically. But it's thin. It returns the same handful of resources for every query. Consultants try it, find it underwhelming, and go back to asking colleagues or searching SharePoint manually.

The problem isn't the AI. The problem is the knowledge base. And the knowledge base is empty because everything worth putting in it is confidential.

This is why the standard approaches to this problem all break down. You can't just ingest everything and hope guardrails catch the sensitive content. You can't exclude everything confidential without gutting the system. And you can't manually redact thousands of PowerPoint slide decks — the maths doesn't work.

The answer is treating the content itself — removing confidential detail at the element level while preserving analytical value. That's a different kind of problem, and it requires a different kind of tool. One that understands what's sensitive in a consulting slide deck, not just what contains a keyword match. Our complete guide to consulting redaction covers exactly how this works in practice.

Knovari's sensitivity framework exists because we mapped exactly this: the full landscape of what makes consulting content confidential, from direct client identifiers to indirect inference risk to non-public information. Understanding the problem at that level of specificity is what makes it possible to solve it at scale.