Every internal-comms leader I have spoken with in the last 12 months has the same problem. Their CFO walks into the planning meeting, looks at the team's "engagement score is up 1.4 points YoY" slide, and asks the obvious follow-up: so what.
It is the right question. Engagement-score deltas measured to a tenth of a point look like noise dressed up as signal, because in most organisations that is exactly what they are. The metric is too aggregated to be diagnostic and too slow to be operational. By the time it has moved enough to be statistically real, the situation that caused it has changed.
Internal-comms leaders in 2026 are moving away from "the score went up" as their headline KPI and toward a small set of operational metrics that defend themselves in a budget review. Four of them are doing most of the work. Each has the same property: a CFO can ask "what does that mean for the business" and the answer is concrete.
1. Acknowledgment rate on critical announcements
This is the metric that pulls the most weight in 2026, partly because it has the cleanest causal story and partly because regulators have started caring.
What it measures. The share of intended recipients who confirmed they read a designated critical announcement, within a defined window (commonly 14 days), with the per-message timestamp preserved for audit.
Why it matters. Every time the answer to "did everyone know about X" used to be "we sent an email," there was a hidden cost: the rework when somebody hadn't seen it, the legal exposure when "X" turned out to be a compliance obligation, the leadership trust erosion when the same announcement had to be re-sent three times. Acknowledgment rate makes that hidden cost visible. It also fixes the core weakness of the metric most teams report instead: the email open. Internal email averages a ~67% open rate but only a ~6–7% click-through (ContactMonkey, 2025 Internal Email Benchmark Report). An open says the message rendered, not that anyone engaged with it. Acknowledgment is the deliberate action opens only pretend to measure.
What "good" looks like. In our experience, mid-market companies running structured-announcement workflows typically land in the 85-92% range on critical messages within 14 days. Below 80%, you usually have a publishing-discipline problem (probably too many "critical" announcements); above 95%, you have a reminder cadence that is annoying people. The healthy band is narrower than people expect.
Acknowledgment rate is also the fastest-growing line on comms-team scorecards: more teams report it every year as the tooling to capture it has become standard rather than bespoke.
2. Read time on high-priority content
The cheap version of this metric is open rate. The serious version is median read time among readers who opened, conditional on the content being designated high-priority.
What it measures. How long readers actually spent on the page, with a sane definition of "spent": scroll position and active-tab events, not just the time the message was visible.
Why it matters. Read time is the only metric that distinguishes a scan from a read. If your CEO sends a 600-word announcement about the year's strategy and the median read time is 4 seconds, you have a structural problem. Either nobody is reading the strategy or your CEO is not writing strategy people want to read. Both are useful findings; "open rate 71%" tells you neither.
What "good" looks like. Calibrated to the content length: roughly 220-260 words per minute is comfortable reading. A 600-word announcement should produce median read times of 90-150 seconds if people are actually reading. Anything below 30 seconds is a scan. Anything below 10 seconds is a glance.
The pattern to watch is not the absolute number but the trend by sender. If the head of engineering's announcements consistently produce 2-minute median read times and the head of marketing's produce 15-second median read times, you have learned something concrete about which voices the company actually pays attention to. That is useful information.
3. Sentiment delta on touchpoint messages
This is the operational version of the annual engagement survey, and it is the metric that has become viable in the last 18 months because cheap, fast LLM-based sentiment scoring is now available.
What it measures. The week-over-week change in average sentiment score on reactions, comments, and replies to messages published in a defined channel or by a defined sender. Run as a delta, not a level: the direction and pace of change is the signal.
Why it matters. Engagement scores measured once a year are too slow to act on. Sentiment delta measured weekly catches the early signs of trouble while there is still time to fix the cause. When engineering sentiment drops 0.4 points over two consecutive weeks and the comms team can flag that to leadership the same week, leadership can have the conversation that prevents the resignation.
What "good" looks like. Mature organisations watch for weekly deltas of greater than 0.3 sentiment-score-units (on a -1 to +1 axis) as their alerting threshold. Lower than that is normal noise; higher than that is something the leadership team should investigate.
The discipline is to treat sentiment delta as a leading indicator, not a scorecard. The point of catching a 0.4-point drop in engineering sentiment in week 2 is to act on it in week 3, not to write it up in the quarterly. Internal-comms teams who use sentiment delta as a quarterly slide get a quarterly slide. Teams who use it as a Monday-morning alert get to actually intervene.
4. Time-to-acknowledgment on policy-class messages
This is the underrated metric. It measures how long it takes a critical announcement to reach the threshold of organisational acknowledgment that you have defined as "covered", usually 90%.
What it measures. Calendar hours from publish time to the time the announcement crossed your acknowledgment threshold. Measured per-message and trended over months.
Why it matters. Speed of reach is the metric that distinguishes a healthy internal-comms function from one that is technically present but operationally absent. Most organisations think their critical announcements reach the company "within a day or two." When they actually measure, they often find the median time-to-90% is several days. That is a slow leak.
What "good" looks like. Functioning organisations hit 90% acknowledgment within 48 hours on policy-class messages. The slow tail is at 10-14 days, which usually means the message did not make it through to a significant portion of the deskless workforce until someone surfaced it manually.
Time-to-acknowledgment is the metric you put on the CFO slide. It is concrete, it has a clear unit (hours), it trends sensibly over time, and it ties to a business outcome (operational risk reduction) the CFO understands.
Putting the four together
A defensible internal-comms dashboard in 2026 looks like a single page with four panels:
- Acknowledgment rate on this quarter's critical announcements, with the bar set against the previous quarter.
- Median read time on high-priority content, segmented by sender, against a 90-day baseline.
- Weekly sentiment delta by department, with the 0.3-unit alert threshold rendered as a line.
- Time-to-90%-acknowledgment trended over the last 12 months, with the 48-hour target rendered as a line.
None of these are "engagement score." None of them are subject-line A/B-test results. None of them are aggregated to the point of being undebatable. They are concrete, they have business meaning, and they give a comms leader something to walk into the planning meeting with that survives the CFO follow-up. And they close a gap most organisations have not: Gallagher's State of the Sector 2025 found fewer than one in ten organisations use data to inform their channel strategy. Producing these four panels puts you ahead of 90% of the field.
What this requires structurally
Three things make this stack possible.
A structured announcement workflow that distinguishes critical messages from everyday updates. The metric is meaningless if every Tuesday update is "critical." Discipline is upstream of measurement.
An analytics layer that tracks read events at the content level, not just channel-level. You need to know whether this announcement was read for 4 seconds or 90, not the average across your feed.
Sentiment scoring that runs weekly, not annually. This used to require a full-time analytics function. In 2026 it requires a per-message LLM call, which costs cents.
This is the instrumentation that the Kayden Connect analytics layer is built to produce. The four-panel dashboard above is not aspirational. It is the default the system delivers, because it is what internal-comms leaders need to defend their function in a 2026 budget review.
The CFO is not going to fund internal-comms because employees said they feel engaged. The CFO is going to fund it because you produced the data showing the operational risk it prevents. Those are the numbers above.
See the dashboards that produce the four metrics →
