Viva Engage Topics That Last: Judge Them on the Second Week, Not the First Day

By Nicolas Saliba · 8 min read

Your community manager brings the monthly numbers. The recognition post about the Toulouse maintenance team reached four thousand people and collected a hundred and eighty reactions, the best single post of the year. Six weeks later the community is quieter than it was before. Meanwhile a shift-handover thread that nobody has ever screenshotted for a slide has been running continuously since last September, at maybe twelve replies a week. One of those is a topic. The other was an event. Native insights present them identically, and if your editorial plan is built from last month’s top posts, you will keep choosing the second one.

A spike and a live topic look the same on day one

The difference is in the shape, and the shape does not appear until the following week. A spike has a single sharp peak on the day of publication, almost all of it reactions rather than replies, and a participant list where nearly everyone appears once and never again. A live topic has a lower peak, a long uneven tail, replies that arrive on days three, nine and twenty-two, and a participant list where a core of people return alongside a steady trickle of newcomers.

Both are legitimate. Recognition posts do something real: they build the goodwill that makes people answer a stranger’s question three weeks later. The mistake is not publishing them, it is reading their numbers as evidence that the community is healthy. The Yammer years taught a generation of internal communicators to optimise for the first-day peak, because that was the number the old dashboards surfaced, and Viva Engage inherited the habit along with the network^[1]. The consequence is a familiar pattern: a community with an impressive monthly highlight and a falling number of people who post anything at all. We set out the wider version of that measurement break in our piece on what Yammer changed and what Viva Engage did not inherit.

Four signals that separate a topic from an event

All four are computable from data a Microsoft 365 estate already holds, and none of them is a view count.

Second-week share. What proportion of a thread’s replies arrived after day seven. Under about a tenth, you published an event. Above a third, something is alive.
First-time participation, and what happened next. How many people posted in this community for the first time on that thread, and how many of them posted again within thirty days. A thread that recruits fifty first-timers and converts none of them has entertained people rather than connected them.
Unprompted recurrence. Did somebody other than the community manager raise the same subject again without being asked. This is the strongest single signal and the hardest to fake, because it means the topic has an owner other than you.
Breadth. How many distinct sites, functions or countries took part. Forty replies from one department is a team conversation happening in public, which is fine, but it is not an organisational topic and it should not be reported as one.

Read together these four tell you something a reach figure cannot: whether the community is accumulating participants or cycling through them. Gallagher’s State of the Sector research keeps finding measurement at the top of the capability gaps internal communicators name for themselves^[2], and this is a good example of why: the data exists, the question is which of it you look at.

What sustains, and what spikes

After a few hundred community reviews the categories are fairly stable. The point of the table is not to rank them. It is to stop treating them as interchangeable in an editorial calendar.

Content type	Typical shape	What it is genuinely good for	How to programme it
Recognition and celebration	Sharp spike, reactions not replies	Goodwill, and the reciprocity that makes people answer later	Regularly, but rationed. It is seasoning, not the meal
Leadership announcement with questions open	Spike with a tail, if the questions get answered	Line of sight, and surfacing what people are actually worried about	Value is in the tail. Commit to answering before you post
Practitioner question and answer	Low peak, long tail, repeat participants	The core asset: solving a problem once instead of forty times	Needs a named expert who answers within a working day
Operational rhythm, such as shift handover or release notes	Flat and continuous	Low ceiling, very high floor. Keeps a community alive between events	Let the team own it. Do not editorialise it
Campaign or contest	Spike by design, then zero	Short-term reach on a defined message	Fine, as long as nobody reports it as community health
Peer teaching, how we did it	Slow build, strong recurrence	Turning one team’s method into an organisational one	Works only with a visible owner and a predictable slot

The row that surprises people is the operational rhythm. It is unglamorous, it will never win an internal communications award, and it is the single most reliable predictor of a community that is still alive in eighteen months. Communities die between events, not during them. Our note on building a recurring editorial programme covers how to structure the calendar around that.

What a sustaining topic needs that a spike does not

Three conditions, and the first one accounts for most of the variance.

Somebody credible answers, quickly. The strongest predictor of whether a practitioner topic survives is the latency on the first substantive reply. Within a working day, the topic compounds. After a week, people stop asking, and they stop quietly rather than complaining, which is why nobody notices until the community is already dead. If you can only fix one thing, fix answer latency.

A reason to return that is not social. Recognition brings people once. An answer they needed brings them back. Gallup’s workplace research has consistently found that the largest engagement gaps sit in frontline and deskless populations^[3], and those are exactly the people for whom a community has to be useful before it can be pleasant.

A boundary. A topic that belongs to everybody belongs to nobody. The communities that sustain are the ones a specific group recognises as theirs, which is also why the general all-company community is almost always the least alive space in the estate. The portfolio logic behind that is in our piece on launching, sustaining and closing communities.

Reading this in the data you actually have

Viva Engage community insights are free, already switched on, and genuinely sufficient for a community manager running one community^[4]. If you want to know whether your posts land inside your own community, the native panel answers that and you do not need anything else. The tenant-level activity report adds a usable view of active posters and readers across the organisation^[5].

Three of the four signals above sit outside what native reporting will give you. It does not classify first-time against returning participants over a long window, so the conversion question stays unanswered. It does not join to organisational structure, so breadth by site or function is unavailable. And its retention window is finite, which matters here more than elsewhere, because judging a topic means comparing this quarter with the same quarter last year, and for many organisations that comparison crosses the Yammer transition. Those limits are set out in detail in our review of what native Viva Engage analytics cannot show.

Tryane sits above that layer. It reads Viva Engage alongside SharePoint, Teams and your newsletter platform, joins the activity to organisational structure through Entra ID or an HR file, and keeps history across product transitions so a topic can be judged over years rather than over a retention window. It is SOC 2 Type 2 certified and GDPR compliant by design, hosts in the EU by default with US data residency on request, and deploys in a couple of hours over single sign-on with no agent to install.

How to tell a sustaining topic is dying

Live topics rarely collapse. They erode, and the erosion is visible about a quarter before anyone raises it in a meeting. Three things move first. Answer latency creeps from hours to days. The participant list narrows until the same three or four people are carrying every thread. And first-time participation goes to zero, which means the topic has stopped recruiting and is now spending down a fixed population. Any one of those on its own is noise. All three together is a topic with roughly two quarters left, and it is far easier to recruit a second expert now than to relaunch the community later. What to put on the dashboard so you see this coming is covered in our piece on what a community dashboard should contain.

Frequently asked questions

What are the best topics for a Viva Engage community?

The ones a specific group of people would keep discussing whether or not internal communications was involved. In practice that means practitioner problem-solving, operational rhythms owned by the team that runs them, and peer teaching. Recognition, announcements and campaigns all have a place, but they are events, and a calendar made only of events produces a community that is busy and not alive.

How long should I give a new topic before deciding?

Six to eight weeks, and judge it on the second week of each thread rather than the first day. A topic that has produced no unprompted recurrence and no returning participants in two months is not going to start. That is not a failure worth agonising over: closing it frees attention for the one that will work.

Can native Viva Engage insights tell me whether a topic is sustaining?

Partly. It will show you activity over time within a community, which gets you the shape of the curve. It will not tell you whether the people replying in week three are the same people who replied in week one, and it will not tell you which sites or functions they belong to. Those two questions need activity joined to your organisational structure.

Our best post ever did not help the community. Why?

Because reach and participation are different things. A post can be seen by four thousand people and recruit none of them into the conversation, which is a broadcast success and a community non-event. Look at how many people posted for the first time on that thread and how many posted again within the month. If both numbers are small, the post did its job as an announcement and nothing more.

Does Tryane replace Viva Engage analytics?

No. Viva Engage carries on recording what it records and you keep your Microsoft 365 investment. Tryane sits above it, reading Viva Engage alongside SharePoint, Teams and newsletters and joining all of it to your organisational structure, so you can answer the returning-participant and breadth questions that no single native panel holds.