Building an Internal Communications Strategy from Scratch

Most companies do not have an internal communications strategy. They have a Slack workspace, an email distribution list nobody trusts, and a monthly all-hands that runs twenty minutes over. Information moves, but nobody can tell you how it is supposed to move, who owns which message, or whether anything actually lands.

You notice the absence of a strategy in specific, repeated failures: a policy change announced in one channel that half the company never saw. An engineer who found out about a product launch from a customer. A manager who says "I told my team" when what they mean is "I forwarded an email." A CEO who writes a thoughtful quarterly letter and gets a 9% open rate.

This guide walks through building a real strategy from nothing: mapping your audiences, auditing what you have, designing a channel architecture, setting cadences, measuring what lands, and rolling the whole thing out in 90 days. It is written for the person who just got handed this job — a people ops lead, a chief of staff, a founder who realized the company got too big for hallway osmosis. No prior comms background assumed.

Why a strategy, not just more messages

The instinct when communication fails is to communicate more. Send another email. Post in another channel. Add a slide to the all-hands. This makes things worse, because the underlying problem is almost never volume. It is structure.

An internal communications strategy answers five questions, in writing, that most companies answer by accident:

1. Who needs to know what? Not everyone needs everything. A strategy defines audiences and what each one actually needs.
2. Where does each kind of message live? So people know where to look, instead of monitoring six channels defensively.
3. When and how often do messages go out? Cadence creates habit. Habit creates readership.
4. Who is allowed to send what? Ownership prevents both silence and spam.
5. How do we know it worked? Without measurement, comms becomes a faith-based activity.

Here is the test of whether you have a strategy: pick a realistic scenario — say, "we are changing the expense policy" — and ask three people in your company where that announcement would appear, who would send it, and how someone who missed it would find it later. If you get three different answers, you have channels, not a strategy.

Step 1: Map your audiences (days 1–10)

Internal comms fails when it treats "the company" as one audience. It is not. A 120-person company typically has at least these distinct audiences, each with different needs:

Audience	What they need	What they usually get
All employees	What is changing, why, what to do about it	Everything, undifferentiated
Managers	Context and talking points before their teams hear news	The same message as everyone, at the same time
A single team or department	Operational detail relevant only to them	Company-wide noise plus their own channels
New hires (first 90 days)	Orientation, norms, where things live	A firehose with no map
Leadership	Upward signal: sentiment, blockers, questions	Filtered good news
Deskless or field staff, if you have them	Mobile-reachable, short, timed to shifts	Email they never open

Do this concretely: open a spreadsheet and list your actual audiences. For each, write one sentence describing what a great communication experience looks like for them. For managers, that sentence is usually something like: "I hear significant news 24–48 hours before my team does, with a short brief on why and what questions to expect." That single sentence, taken seriously, will change how you sequence every major announcement for the rest of your tenure.

Two audiences deserve special attention.

Managers are a channel, not just an audience

Research on organizational communication consistently finds that employees trust their direct manager more than any other company source — more than the CEO, far more than HR broadcasts. When a message matters, the manager retelling is the message; your broadcast is just the manager's source material. Build a manager pre-brief into your process for anything significant: a short document sent a day early with the announcement, the reasoning, three likely questions, and suggested answers. It takes thirty minutes to write and multiplies the reach of everything you send.

New hires reveal your comms debt

Every new hire runs a natural experiment: can a person with no tribal knowledge figure out where information lives? Ask people in their third week: "Where would you look for the PTO policy? Where do product launches get announced?" Their confusion is a precise map of your gaps. Schedule these ten-minute interviews permanently — three per quarter is enough.

Step 2: Audit what you already have (days 5–15)

Before designing anything, inventory the current state. For every channel — email lists, chat channels, all-hands, newsletters, wikis, notice boards, the plumbing between them — record four things:

• Nominal purpose — what it is supposedly for
• Actual use — what actually gets posted there (read the last 30 items; do not rely on memory)
• Reach — who sees it, and roughly what fraction engage
• Owner — who decides what goes there; often the answer is "nobody"

The audit usually surfaces three patterns:

Duplication. The same announcement goes to email, two chat channels, and the all-hands, with slightly different wording. This trains people that they can ignore any single channel, which destroys all of them.

Orphans. Channels with a real purpose and no owner. The "#announcements" channel that has become a place to sell concert tickets. The intranet page last updated fourteen months ago that still ranks first in search.

Shadow channels. The private group DM where the actual information moves. You cannot eliminate these, and you should not try — but their existence tells you which official channels have failed. If managers maintain a private thread to find out "what's really going on," your leadership comms have a trust problem, not a channel problem.

Write the audit up in two pages. You will use it to justify every change you make in the next 75 days, because every channel you kill has one defender who loves it.

Step 3: Design the channel architecture (days 15–30)

Now design the target state. The core principle: every message type has exactly one home. People should never have to monitor multiple places for the same category of information, and they should always know where to look something up after the fact.

A workable architecture for a company of 20–500 people looks like this:

Message type	Home	Cadence	Owner
Company announcements (policy, org, strategy)	Company feed, pinned	As needed; expect 2–6/month	Comms lead / CoS
Company narrative & context	All-hands + written recap	Monthly	CEO
Cross-team news roundup	Internal newsletter	Weekly or biweekly	Comms lead
Team status & progress	Async status updates per team	Weekly	Each team lead
Urgent operational issues	Chat, dedicated incidents channel	Rare, by definition	On-call / ops
Social, wins, kudos, life events	Company feed	Continuous	Everyone
Reference material (policies, how-tos)	Wiki / handbook	Updated on change	Doc owners
Questions with lasting answers	Q&A forum	Continuous	Everyone

Three design rules make this work:

Separate the durable from the ephemeral. Chat is where information goes to die — it is a stream, not a shelf. Announcements, decisions, and policies need a home where they persist, can be pinned, and can be found by search three months later. Use chat to point at the durable home ("New expense policy posted → link"), never as the primary record. This is the single most common architectural mistake, and fixing it alone justifies the whole project.

Distinguish push from pull. Push channels interrupt (chat, email, notifications); pull channels wait to be visited (feed, wiki, dashboards). Reserve push for things that are genuinely time-sensitive for the recipient. Everything else goes to pull channels with a predictable rhythm so people build a checking habit. Teams that get this backwards — pushing everything — train employees to ignore notifications entirely, and then the one urgent message drowns.

Consolidate where you can. Every additional destination is a tax on attention. If your feed, newsletter archive, wiki, and Q&A live in four different tools with four different logins and four separate searches, findability collapses. This is a place where platform choice genuinely matters: in Openbook, for example, a company space can hold a Feed room for announcements and social posts, a Wiki room for the handbook, and a Q&A room for lasting questions, all under one search — which means "where would I look for this?" has one answer instead of four. Whatever tooling you use, aim for that property. (See how to choose collaboration software for the broader selection process.)

Document the architecture on a single page titled "Where things go." This page becomes part of onboarding, gets linked from every channel description, and is the artifact you point to when someone posts a policy change in a chat thread at 6 p.m. on a Friday.

Step 4: Set the cadence calendar (days 20–30)

Cadence is underrated. A newsletter that arrives every Friday at 10 a.m. gets read out of habit; a newsletter that arrives "when there's enough content" gets read out of luck. Predictability is the mechanism by which pull channels work at all.

Build a simple annual comms calendar:

• Weekly: newsletter or roundup (pick a day and never move it); team status updates due (pick a day — Friday afternoon for the week ending, or Monday morning for the week ahead).
• Monthly: all-hands with a written recap posted within 24 hours; leadership Q&A (collect questions async beforehand — you will get better questions in writing than raised hands ever produce).
• Quarterly: strategy/goals update from the CEO; engagement pulse survey; comms metrics review (see below).
• Annually: planning kickoff comms; policy refresh cycle; audience interviews refresh.

Two cadence rules from hard experience:

Never skip silently. If the newsletter has a thin week, send a short one. Skipping teaches people the channel is unreliable, and unreliable channels get unsubscribed from mentally long before anyone clicks a button.

Batch the non-urgent. Individual announcements throughout the week each cost an interruption. A Friday roundup that packages five minor updates costs one. Reserve standalone announcements for things that would be diminished by batching — anything with a "what you need to do" component, or anything people will hear about through side channels within hours anyway.

Step 5: Write like a person (ongoing)

Strategy determines whether a message arrives. Writing determines whether it lands. Set explicit standards — a half-page style guide beats a hundred edits:

• Lede first. The first sentence carries the news. "Starting March 1, we're moving to a 4-day RTO schedule" — not three paragraphs about our journey as a company. If a reader stops after one sentence, they should still know what happened.
• Always include "what this means for you." Every announcement gets an explicit section: what changes, for whom, starting when, and what action (if any) is required. If the answer is "nothing is required of you," say so — it is the most relaxing sentence in corporate life.
• State the why, briefly and honestly. One honest paragraph of reasoning builds more trust than five paragraphs of framing. If a decision involved tradeoffs, name them. People can tell when they are being managed.
• Ban the corporate voice. "We are excited to share a strategic evolution of our workplace flexibility framework" is a sentence that makes readers trust you less. Write the way the CEO would explain it to one employee over coffee.
• Cap length by type. Announcements: 300 words plus a link to detail. Newsletter items: 50–80 words each. All-hands recap: one page. Long context belongs in a linked doc for the minority who want it.

For deeper treatment of announcement mechanics — formats, repetition, and handling comments — see Company Announcements That Do Not Get Ignored.

Step 6: Measure honestly (days 30 onward)

Internal comms measurement has a reputation problem because people measure activity ("we sent 14 announcements") instead of outcomes. Use three layers:

Layer 1: Reach

Did the message arrive? Track open/view rates on major announcements and the newsletter. Absolute numbers matter less than trends and comparisons — if the benefits update got 80% views and the strategy update got 30%, that difference is information. Most tools give you this for free; if yours gives you nothing, that is a real argument for changing tools.

Layer 2: Engagement

Did anyone respond? Reactions, comments, and questions on announcements; poll participation; Q&A activity. Watch the comment-to-view ratio on significant news. Near-zero comments on a controversial change does not mean acceptance — it usually means people are having the conversation somewhere you cannot see. A healthy comms culture produces visible questions.

Layer 3: Comprehension and trust

Did it work? This layer needs deliberate measurement, quarterly:

• Message recall: in your pulse survey, ask 2–3 factual questions about recent major announcements. ("What is the new PTO carryover limit?") Sub-50% correct on something you announced twice is a wake-up call.
• Channel confidence: "I know where to find the information I need to do my job" (agree/disagree, tracked over time). This is arguably the single best internal comms KPI in existence.
• Trust: "Leadership communicates openly and honestly," tracked over time and — importantly — read alongside events. A dip after a hard announcement is normal; a dip after a routine quarter is a signal.

Review all three layers quarterly in a one-page report to leadership. Include one change you are making based on the data. This habit — visible iteration — is what separates a comms function from a comms mailbox.

The 90-day rollout plan

Here is the whole build, sequenced. Adjust dates to taste; keep the order.

Days 1–30: Learn and design

• Run the audience mapping exercise; interview 8–10 employees across functions and tenure, including 2–3 recent hires.
• Complete the channel audit; write the two-page findings memo.
• Draft the channel architecture ("Where things go") and cadence calendar.
• Review both with leadership and 3–4 managers. Incorporate feedback, but hold the line on the one-home-per-message-type principle — it is the piece everyone tries to negotiate away.
• Quick win: pick the single most visible broken thing (usually the announcements channel or a dead newsletter) and fix just that. Early proof buys patience for the rest.

Days 31–60: Build and pilot

• Stand up or restructure the core channels per the architecture. Archive orphan channels — with a two-week notice and a pointer to the replacement, not a silent deletion.
• Launch the newsletter (or relaunch with the new format and fixed schedule). Write the first three issues before launching issue one, so a busy week cannot kill the cadence in month one.
• Pilot the manager pre-brief on one real announcement. Debrief with the managers afterward.
• Publish "Where things go" and add it to onboarding.
• Start collecting reach metrics as your baseline — you want before/after data.

Days 61–90: Institutionalize

• Run the first full-dress major announcement through the new machinery: pre-brief, single-home post, chat pointer, newsletter inclusion, all-hands mention, metrics collected.
• Run the first quarterly pulse with recall and channel-confidence questions.
• Deliver the first metrics report with one iteration decision.
• Write the comms playbook: architecture, cadence, style standards, the announcement checklist, and the pre-brief template — roughly six pages. This document is what makes the strategy survive your vacation, and eventually your successor.

By day 90 you will not have perfect comms. You will have a system: known homes, known rhythms, known owners, and a feedback loop. That is the whole difference.

Failure modes to expect

A few things will go wrong. Plan for them:

The executive workaround. A leader will bypass the architecture because "this one is special." Do not fight it in the moment — route the message properly in parallel, then follow up privately. Two or three demonstrations that the system produces better results than the workaround usually ends the workarounds. Usually.

The volume creep. Once the machinery works, everyone wants to use it, and the newsletter swells to 2,000 words. Your job becomes editorial: protect the reader. A good rule: anything without a clear "why an average employee cares" answer gets cut or demoted to a link.

The metrics mirage. View rates will jump in month two simply because the channels are novel. The honest baseline is month four. Report it that way.

The silence trap. In a quiet quarter you will be tempted to conclude comms is "done." Comms is a maintenance function, like security. The moment it seems finished is the moment the decay starts — stale wiki pages, drifting cadences, a new shadow channel. The quarterly review exists precisely to catch this.

Where to start this week

If the 90-day plan feels heavy, here is the minimum viable version, doable in one week:

1. Interview five employees: "Where do you look for X?" for five common information needs. Write down where the answers disagree.
2. Write the one-page "Where things go" doc — even if it currently describes an imperfect reality.
3. Pick one cadence and commit to it publicly: a weekly roundup, every Friday, starting this Friday.
4. Add a "what this means for you" section to the very next announcement anyone sends.
5. Put a recurring 30-minute "comms review" on your calendar for the first Monday of each quarter.

Everything else in this guide is an elaboration of those five moves. A communication charter can formalize the norms with your team — here is how to write one — and if your channels are scattered across too many tools to ever have one search box, consolidating them is the structural fix that makes everything above easier.

If you are assembling that consolidated home, Openbook gives you the pieces — a company feed with pinned announcements and polls, a wiki for the handbook, Q&A for lasting questions, and async check-ins for upward signal — in one space with one search. Start free at openbook.work and stand up the "Where things go" architecture in an afternoon.