Internal Communication KPIs: How to Decide Which Ones to Drop

By Nicolas Saliba · 8 min read

The monthly internal communication report in most large organisations runs to twenty-something numbers. Newsletter open rate, click rate, intranet page views, unique visitors, average time on page, Viva Engage community members, active members, posts, replies, reactions, top five articles, top three communities, a sentiment score, an adoption percentage, and one chart nobody can explain without a preamble. The CCO reads the first slide. Occasionally a director asks about a single number, and the honest answer is that it moved because of a public holiday.

The instinct at that moment is to measure better. It is usually the wrong instinct. Most internal communication scorecards do not have a measurement problem. They have a subtraction problem. Metrics get added and nothing is ever removed. This article is about how to decide what to cut.

Scorecards grow because nobody owns removal

Every number on that deck arrived for a defensible reason. A director asked about time on page once in 2022, so time on page went on and stayed on. The intranet relaunched, and the project brought its own adoption metrics. A vendor demo included a sentiment index that looked impressive in the sales meeting. A new Viva Engage community was created for the transformation programme, and it got its own row.

Adding a metric requires one person to ask a question. Removing one requires someone to decide that a question no longer matters, in public, in front of whoever asked it originally. That asymmetry is the entire mechanism. Left alone for three years, it produces an inventory rather than a scorecard.

Gallagher’s State of the Sector research has consistently found measurement to be the capability internal communicators name as their weakest^[1]. That finding is usually read as a call to measure more. Read it the other way round for a moment. A team that reports twenty-two numbers and still cannot answer whether last quarter’s safety campaign reached the plants does not lack data. It lacks a decision about which data is load-bearing.

Four tests a KPI has to pass

Run every metric currently on your deck through these four. A metric that fails any one of them is a candidate for removal from the executive view. A metric that fails two should be gone this month.

The decision test

If this number moved twenty per cent in either direction, what would you do differently next month? If the answer takes more than a sentence, or if it is “investigate”, the metric is context rather than a KPI. Context has a place. It does not have a place on the slide the executive committee reads.

Total intranet page views fails this test in almost every organisation. A twenty per cent rise could mean a successful campaign, a mandatory compliance page, a navigation change that added a click, or an IT outage that pushed people to look something up. None of those produce the same next action.

The ownership test

Is there a named person who can move this number, and do they know it is theirs? Newsletter open rate usually passes: someone writes the subject lines. Aggregate reaction volume across all Viva Engage communities usually fails, because it is the sum of forty community managers’ behaviour and no single person can act on the total.

The movement test

Has this number moved outside its normal noise band in the past twelve months? A newsletter open rate that has sat between 61 and 64 per cent for two years is telling you the same thing every month. Keep measuring it, because a genuine drop matters, but stop spending a line of executive attention on a number that reliably says nothing. Set an alert threshold instead and report it by exception.

The segmentation test

Can you break this number down by function, site, country, language or contract type? An aggregate you cannot decompose is an aggregate you cannot act on. When a CCO asks why engagement fell, “it fell four points” is not an answer. “It fell four points, entirely inside manufacturing, and the drop starts the week we changed the shift pattern” is an answer. Segmentation is what turns a number into a next step, which is why audience segmentation is worth structuring before you finalise the scorecard rather than after.

Metric	Usual justification	Test it fails	Better replacement
Total intranet page views	“Shows the intranet is used”	Decision, ownership	Reach against a defined population, per campaign
Viva Engage community count	“Shows the network is growing”	Decision, movement	Share of communities with activity in the last 30 days
Total reactions and likes	“Shows engagement”	Ownership, segmentation	Participation rate by function and site
Number of posts published	“Shows the team is productive”	Decision	Nothing. This measures the IC team, not the organisation
Sentiment score, unsegmented	“Shows how people feel”	Segmentation, decision	Sentiment on named campaigns, split by audience
Newsletter open rate as headline	“Everyone reports it”	Movement	Keep as an operating diagnostic, report by exception

What tends to survive

Teams that run this exercise honestly usually end with five to seven measures. The specific set depends on your channel mix and what your leadership actually asks for, and the canonical list of the measures that hold up under scrutiny is set out in our guide to internal communication KPIs for 2026. Use that as the menu. This article is only about the discipline of choosing from it.

What survives tends to share a shape. It is expressed as a rate against a defined population rather than a raw count. It is attached to a named audience rather than the whole tenant. And it can be traced to a decision someone will take. “Forty-one per cent of the commercial function saw the strategy update within seven days, against 9 per cent of manufacturing” survives an executive meeting. “148,000 page views” does not survive a follow-up question.

Keep two layers, publish one

Pruning the executive scorecard is not the same as deleting data, and conflating the two is how teams talk themselves out of the exercise. Keep an operating layer with as many diagnostics as your team finds useful: time on page, scroll depth, best posting hour, channel overlap, per-author performance. That layer exists so the team can work out why the headline numbers moved.

The executive layer is a different product with a different reader. It answers whether the function is doing its job and where the gaps are. Five to seven measures, each with a trend and a named owner. Our monthly reporting template shows one working structure, and the piece on executive reporting covers how to frame the narrative around it.

When the tooling quietly picks your KPIs for you

There is a failure mode worth naming: the metric you report becomes the metric your tools make easy to export. SharePoint site analytics gives you views and unique viewers for a site^[2]. Viva Engage community insights gives you activity inside a community^[3]. The Teams reporting reference gives you channel and message counts^[4]. All three are free, already switched on, and genuinely sufficient if you own one site or one community and want to know how it is doing.

The limit is that each panel reports on its own surface. None of them joins to your organisational structure, so none of them can answer a question phrased in terms of functions, sites or contract types, which is how executives phrase questions. The result is that scorecards fill up with what is easy to copy rather than what passes the four tests. That is the same underlying constraint described in our note on the limits of Viva Engage native analytics.

Some teams close the gap with a custom Power BI build on the Microsoft Graph reporting API. That route can work, and several organisations run it well, but it is more demanding than the first sprint suggests. The reporting API is not consistently reliable and produces days with missing data. Customers have observed KPI values in Power BI that do not reconcile with the native SharePoint reports for the same period, which is an uncomfortable thing to discover mid-presentation. A dependable build needs a robust ingestion process, scheduled reconciliation against the native figures, and a named owner for the pipeline. Budget for the pipeline, not the dashboard. Our honest assessment of SharePoint analytics in Power BI covers the specific failure modes.

This is the layer Tryane occupies. It reads SharePoint, Viva Engage, Teams and your newsletter platform together, joins that activity to your organisational structure through Entra ID or an HR file, and keeps unlimited history across product transitions. Practically: SOC 2 Type 2 certified, GDPR compliant by design, EU hosting by default with US data residency on request, single sign-on through Azure AD or Entra ID, deployed in a couple of hours, no agent to install.

A ninety-minute pruning session

Write every metric you currently report on a separate card. Include the ones buried in appendices, because they are still costing preparation time. Score each card against the four tests, out loud, with your team in the room. Force-rank the survivors and draw a line after seven.

Then do the part teams skip. For each dropped metric, write one sentence explaining why it left, and send that list to whoever originally asked for it. The sentence matters more than the decision. It converts “the team stopped reporting my number” into “the team made a call and told me why”, and it stops the metric quietly reappearing next quarter.

Frequently asked questions

How many internal communication KPIs should we report to leadership?

Five to seven for the executive view. Below five you tend to lose the ability to distinguish reach problems from engagement problems. Above seven, attention spreads thin enough that no single number carries weight. Keep a larger operating set for the team, but do not confuse it with what leadership reads.

Should we drop newsletter open rate entirely?

No. Drop it from the headline slide if it has been stable for a year, and report it by exception when it breaks its normal band. Open rate remains a useful operating diagnostic and a fast signal that something changed in subject lines, send timing or list hygiene. Our note on internal newsletter metrics beyond open rate sets out what to pair it with.

What if leadership specifically asks for a metric that fails the tests?

Report it, and report the better measure next to it. Asking a CCO to abandon a familiar number in the same meeting you introduce a new one rarely lands. Run both for two quarters. The weaker metric usually falls away on its own once the stronger one is answering the follow-up questions.

Can Viva Engage and SharePoint native analytics support a pruned scorecard?

For a single community or a single site, yes, comfortably. The constraint appears when your five to seven measures are expressed by function, country or contract type, or when they need to combine channels. Native panels report per surface and do not join to workforce attributes, so those questions need either a manual export and join or a dedicated analytics layer.

How often should we revisit the scorecard?

Once a year is enough for a full pruning session, ideally just before budget season so the surviving measures are the ones you will defend. Outside that, add a metric only by replacing one. The one-in-one-out rule is unsophisticated and it works.

Sources

• Gallagher, State of the Sector

• Microsoft Learn, SharePoint portal health and site analytics

• Microsoft Learn, Viva Engage overview

• Microsoft Learn, Teams reporting reference