How much time do employees lose searching for policy and SOP answers?

Employees spend roughly 1.8 hours every working day — about 19% of the workweek, or 9.3 hours per week — searching for and gathering internal information, according to McKinsey Global Institute research. For a 50-person company that's the equivalent of nine or ten full-time people doing nothing but looking for answers that already exist somewhere in the company's own documents. Survey after survey compiled by Cottrill Research lands in the same range or worse — the numbers have been embarrassing for a decade, and they haven't moved much, because the problem isn't effort. It's that documents are stored to be filed, not to be asked.

Why policy questions are slower than they look

A policy or SOP question feels like it should take thirty seconds. It rarely does, because the answer is almost never one document:

• The answer is split. The expense policy says one thing; the exceptions memo from January says another; the per-diem table lives in a spreadsheet nobody linked.
• The version is the question. "What's the overtime rule?" really means "what does the current revision say, and did the update that went around in March change it?" Answering from the wrong revision is worse than not answering.
• Variants multiply. Jurisdiction, union agreement, customer contract, site — the same question has different correct answers depending on which document governs the asker.
• The expert is the bottleneck. When search fails, the question goes to the one person who knows — and their day becomes a queue of interruptions. When that person is away or leaves, the answers leave with them.

What the lost time actually costs

Illustrative math applying McKinsey's 9.3 hrs/week figure; your mix of roles and rates will vary. Even at half the benchmark, the number justifies fixing the problem.

And the hours are only the visible cost. The quieter one is the answers given from memory: the leave entitlement quoted from an old revision, the safety procedure recalled "close enough," the expense exception granted because that's how it was done last time. In regulated or unionized environments those become findings and grievances. An answer tied to the exact clause of the current document is defensible; a recollection is not.

Making policies and SOPs answerable

1. Establish one source of truth per policy. One current revision, clearly dated, with superseded versions marked as such — most wrong answers are right answers to an old document.
2. Write rules as rules. Numbered clauses, defined terms, explicit effective dates and scope ("applies to: hourly staff, BC sites"). Prose essays can't be cited; clauses can.
3. Keep exceptions with the rule. An exceptions memo filed separately from its policy is a wrong answer waiting to happen. Cross-reference or consolidate.
4. Put an answer layer on top. The step that actually moves the needle: a system that takes the question in plain language, finds the governing clause in the current revision, and returns the answer with the citation — so the asker doesn't search and the expert isn't interrupted.

A 30-day path to an answerable policy library

This doesn't need a documentation rewrite project. A workable sequence most teams can run in a month:

1. Week 1 — inventory the askables. Pull the last month of "quick questions" from inboxes and chat: what got asked, who answered, how long it took. This list — usually 30–60 recurring questions — is your real requirements document.
2. Week 2 — gather the governing documents. For each recurring question, identify the document and revision that actually answers it. Expect surprises: questions with two conflicting answers, policies nobody can find, and rules that exist only as folklore. Log every conflict you find rather than quietly picking a winner.
3. Week 3 — fix the worst five. Don't boil the ocean. Consolidate the five policies generating the most questions or the most conflicting answers: one current revision each, dated, exceptions merged in, superseded versions marked.
4. Week 4 — put the answer layer in front of people. Route the questions to the system instead of to the expert, and watch the citations. Every answer that comes back citing an outdated or ambiguous clause is free QA on your documentation — fix the document, and the next answer fixes itself.

The measure of success is simple: the expert's interruption count drops, and answers start arriving with section numbers attached. From there, the library improves continuously, because every question that can't be answered cleanly tells you exactly which document needs work next.

Where IntelMS fits

That answer layer is exactly what IntelMS is. Load your policies, SOPs, handbooks, and contracts; your team emails questions the way they'd ask a colleague — "What's the home-office equipment allowance?" "Which SOP covers lockout/tagout on the packaging line?" — and gets back the answer with a citation to the exact document, section, and revision. If two documents conflict or a revision has been superseded, it flags it instead of papering over it; questions that touch compliance judgment or approvals get routed to a human rather than answered automatically. The expert stops being a search engine, the answers stop drifting from the documents, and every reply leaves a cited trail — the same discipline that makes document intelligence different from a generic chatbot, applied to the documents your whole company runs on. See how it works for corporate teams.

Frequently asked