Here's the thing about working in internal comms: we're meant to be the ones keeping everyone informed, but half the time we're flying completely blind ourselves. Internal communications is often a team of one, trying to be the thinker and the doer, which can leave very little space for proper analysis.
We know our All-Hands happened. We know we sent that Slack update. We know the newsletter went out on time. But do we actually know if any of it worked? If people read it, engaged with it, or changed their behaviour because of it? Often, no.
It's not our fault. Comms is inherently difficult to measure. We don't have clean data sources like Sales (revenue, pipeline, conversion rates) or Product (feature adoption, user retention). Our metrics are fuzzy at best, and even when we can measure something, it's often impossible to tie it back to the KPIs that actually matter to the business. Leaders (understandably) aren’t going to be interested in an individual post’s click rate, unless you can tie that to a metric they do care about.
So we end up reactive. Putting out fires. Responding to requests from leadership. Trying to fix things that feel broken without really understanding why they broke in the first place.
An internal comms audit changes that. It shifts you from reactive to proactive. It helps you understand where you're keeping people well informed and where you're missing the mark entirely. And crucially, it gives you the data to build a strategy for the year ahead that isn't just based on gut feeling or whatever the loudest voice in the room wants.
If you're heading into planning season for 2026 and you don't have a clear picture of what's actually working in your comms, this is for you!
The Nine Stages of an IC Audit
Stage 1: Start by defining the questions
You cannot answer questions you haven't asked. I know that sounds obvious, but this is where most people go wrong.
The temptation is to pull all the data you can get your hands on first, and then try to find patterns in it afterwards. Don't do this. If you work backwards like that, you'll inevitably manipulate the data to confirm what you already think, rather than discovering what's actually true.
Instead, start by writing down the questions you need answered. What are the things keeping you up at night? What does leadership keep asking you about? What patterns have you noticed anecdotally that you want to validate (or disprove)?
Some examples of good audit questions:
- What's the best way of making strategic announcements that land?
- What's the right balance of synchronous vs asynchronous comms for our company right now?
- Are our core comms rituals actually helping people do their jobs week to week?
- Why do certain teams or geographies seem less engaged than others?
- Are managers equipped to cascade information effectively?
Write these down. Pin them somewhere visible. You'll need to keep coming back to them.
Stage 2: Map your questions to metrics
Now work backwards from each question to figure out what data would help you answer it.
This will be a mix of quantitative (attendance numbers, view rates, engagement metrics, click-through rates) and qualitative (surveys, interviews, focus groups). Don't dismiss either – you need both to get the full picture.
Let me give you an example from our recent audit at Oyster. One of our key questions was: "What's the right balance of sync and async comms for the Oyster of today and tomorrow?"
To answer that, we needed:
- All-Hands live attendance vs Loom playback numbers
- AMA live attendance vs Loom playback numbers
- View rates and engagement on our core Slack channels
- Qualitative feedback from focus groups about how people prefer to consume information
Stage 3: Stress-test feasibility
Not all the data you want will actually be available to you. This is the bit that's incredibly frustrating, but you need to know about it before you waste weeks chasing something that doesn't exist.
There are three main barriers: privacy constraints, tool limitations, and political barriers.
In our audit, I wanted to understand the average hours per week individuals spent in meetings, what percentage of these were recurring vs. ad hoc, and how that varied across departments and time zones. This would have been brilliant data for understanding sync vs. async preferences.
IT shut it down immediately. It’s not possible to pull with our privacy configurations, and they weren't willing to change that just to satisfy my curiosity, which is completely understandable.
So before you start actually pulling data, make a list of everything you'd ideally want. Then go and ask the relevant teams — IT, People Ops, whoever owns the tools — whether it's actually feasible. Cross off what isn't. Adjust your approach accordingly.
Stage 4: Choose your qualitative methods wisely
Feasibility is key. However, just because something is feasible doesn't mean you should do it.
The biggest trap here is survey fatigue. If your company has just run an engagement survey, a pulse survey, or any other kind of survey in the last month, do not run another survey. People will ignore it, or worse, they'll resent you for it.
Focus groups and 1:1 interviews take more time to coordinate, but they often give you richer insights anyway. You can dig into the 'why' behind the numbers in a way that a survey never will.
Think strategically about what will actually give you useful signals, rather than just noise you'll have to wade through later.
Stage 5: Define your deliverables upfront
Before you start pulling any data, decide what you're actually going to produce at the end of this.
What does the business need from this audit? What do you need as a comms leader? A slide deck for leadership? Updated guidelines? A new strategy document?
The reason you define this now — not later — is that you can start dropping insights into your deliverables as you go, whilst the content is still fresh in your mind. If you wait until the end, you'll have forgotten half of what you learned, and you'll have to go back through everything again.
For our most recent audit, I defined five deliverables:
Definite:
- Deck covering audit results and recommendations for 2026
- Updated Async Comms guidance
- New guidance on what channels to use for different purposes
Potential (depending on what the data showed):
- Strategic announcements strategy
- Set of Oyster audience segments/personas for use in comms
Having this structure meant that when I spotted something interesting in the data, I could immediately slot it into the right deliverable rather than keeping it in my head or scribbling it on a post-it note that I'd lose by the next day.
Stage 6: Map your channels (all of them)
This stage sounds basic, but you will be surprised — possibly horrified — by how many channels you actually have when you sit down and list them all.
Create a spreadsheet (I use Google Sheets). Break your channels down into three categories:
IC-managed channels: The ones you own. For us, that's All-Hands, AMAs, our core Slack channels, and our Oyster News hub on Notion.
Non-IC-managed channels: The ones other teams own but that still play a role in keeping people informed. For us, that includes our manager collaboration channel, department announcement channels, social media, and department All-Hands.
Comms elements: The supporting bits that make comms work but aren't channels themselves. This includes our async comms guidelines, company newsletter, Crisis Comms Playbook, and comms training course.
Once you've got your list, add columns for each channel:
- Current owner
- Back-up owner (if applicable)
- Intended purpose
- Actual used purpose (this is often different – write down what's really happening)
- Primary audience
- Secondary audience
- Content types typically shared in this channel
- Can we track performance? How?
When you can see it all laid out in front of you like this, patterns start to emerge immediately. You'll spot overlaps, gaps, channels that have drifted from their original purpose, and channels that probably shouldn't exist at all.
Then decide which channels are in scope for this audit. I typically include all IC-managed channels and most comms elements, but I'm selective about non-IC managed channels. Some have a complicated ownership status that would be more effort to unpick than valuable. Others are just impossible to get proper visibility on.
Stage 7: Pull the data (brace yourself)
Right. This is where it gets laborious.
Data pulling takes an enormous amount of time, and it's often more complicated than you think it's going to be. I’ll walk you through a real example so you know what you're in for.
Example: Analysing All-Hands Attendance at Oyster
Step 1: Download your employee org list (from your HRIS, probably). You need this as your source of truth for all of the characteristics. Basically, you need to pull any characteristic that you may want to then analyse data against - we focused on their department, country, manager status, and number of direct reports. We couldn’t pull tenure directly, just the start date, so then needed to write some formulas to convert the start date into tenure, rounded to the nearest year.
Step 2: Download your last All-Hands attendee list from Zoom (or whatever you use for your live events).
Step 3: Clean the data. This means:
- Removing any duplicates or external attendees who shouldn't be in your analysis
- Manually inputting the correct email addresses where they're missing or formatted weirdly
- Setting up formulae to pull each person's department, tenure, and geography from your org list (usually matching on email address)
Step 4: Realize that one month isn't a representative sample. Pull three more months of data. Repeat steps two and three for each month.
Step 5: Now add the people who watched the recording. Download viewer data from Loom (or your video platform). Clean it the same way. Add a column that tags each person as either "Live" or "Async".
Step 6: Create summary tables that calculate averages. For example:
- Total population of each department (from your org list)
- Average live attendance by department
- Average async attendance by department
- Average total attendance (live + async, accounting for people who did both)
- Percentage of each department who haven't attended any All-Hands in the last four months (this one's important – it shows you complete drop-off)
Step 7: Do the same analysis by geography, by tenure band, by whatever other cuts matter for your business.
Step 8: Build charts that make the patterns visible. I typically create:
- Average live attendance by department/geography
- Average async attendance by department/geography
- Average total attendance by department/geography
- Percentage who haven't attended at all
And that's just for one channel.
Now repeat variations of this process for your AMAs, your Slack channels (using Slack analytics for view rates and engagement), your Notion pages (using page view analytics), your newsletter (open rates, click rates), and everything else in scope.
There may be AI options in your tools to help with this — and if so, use them! But make sure you check that output. I tried to use Gemini (in Google Sheets) to help with the cleaning stage, and things got messy. However, it was a real help when I was analyzing Slack patterns and it could rapidly tell me which days and times had the best and worst engagement. So play around with AI, but make sure you double-check what it’s giving you!
Stage 8: Check in with yourself
Data pulling is a rabbit hole.
You’ll get sucked into chasing tiny data points that don't actually matter. You'll spend two hours trying to figure out why one tiny team’s attendance record looks weird, or trying to get one more decimal place of precision on a percentage that's already clear enough.
Stop. Regularly.
Pause and ask yourself: "Will this actually help me answer my audit questions, or am I just chasing data for data's sake?"
Keep your list of audit questions somewhere visible. Glance at it every time you're about to start a new analysis. If what you're doing doesn't clearly connect to one of those questions, don't do it.
Another tip: schedule your qualitative work — focus groups, interviews — after you've pulled most of your quantitative data. That way, you'll have a sense of what the numbers are showing you, and you can ask much better questions to understand the 'why' behind the patterns.
For example, when I pulled our All-Hands data, I discovered that three departments had significantly lower attendance than others (both live and async). So I ran a focus group specifically with people from those teams to understand what was going on.
Stage 9: Answer your questions and build your deliverables
Right. You've done all the hard work. Now bring it together.
Go back to your original audit questions. What did you learn? Can you actually answer them now, or do you need to pull more data?
For each question, write out:
- What the data showed
- What it means
- What you recommend doing about it
Then build the deliverables you defined back in Stage 5. Make them actionable. You haven't done all this work just to produce a nice-looking slide deck that gets filed away and forgotten. You've done it to drive real change in how your company communicates.
A few final thoughts
- This process takes time. It's not a week-long project. Our most recent audit took me about six weeks of solid work (alongside everything else I was doing). If you're doing this for the first time, budget even longer.
- You’ll feel like you're drowning in spreadsheets. That's completely normal. Keep going.
- The insights you get will be worth it. They'll make your 2026 comms strategy infinitely stronger, because it'll be grounded in evidence rather than guesswork. You'll be able to walk into planning meetings and say "Here's what's working, here's what isn't, here's why, and here's what we're going to do about it" with actual data to back you up. I can confirm that that feels amazing!
Time to audit your comms
If you're heading into planning season without a clear picture of what's working in your internal comms, now's the time to fix that.
You don't have to do everything I've outlined here. If you're overstretched (and let's be honest, you probably are), start small. Run an audit with just one key question, and only focus on the data that’ll answer it. That’ll still give you more insight than you have right now.
Your future self will thank you — as will everyone who relies on your comms to do their jobs well.
.webp)
