A franchise group runs a SharePoint intranet for its 900 corporate staff and a separate SharePoint extranet for the 300 franchised sites that carry its brand. Around 4,000 people work in those sites. None of them are on the payroll. When the group launched a brand refresh in the spring, the Head of Internal Communication was asked a fair question at the following board meeting: did it land?
For the corporate population she had an answer inside a day, broken down by country and function. For the 4,000 people who actually face the customer, she had a page view count and no way to say which sites had read the brief and which had not. That asymmetry is the intranet versus extranet problem as it really appears in an internal communication function. It is not a definitional question. It is an identity question, and it breaks your reporting at precisely the boundary where your most operationally important audience sits.
The definitions, briefly, because they are not the interesting part
An intranet is a private digital environment for people employed by the organisation. Authentication runs against the corporate directory, which in a Microsoft 365 estate means Entra ID. Content is company news, policy, HR procedure, project spaces. If you want the longer version, our note on what an intranet is covers the ground properly.
An extranet is a controlled space opened to people outside the legal entity: franchisees, dealers, resellers, suppliers, subcontractors, clinical partners. In Microsoft 365 this is usually one or more SharePoint site collections shared through Microsoft Entra External ID B2B collaboration, where the external person signs in with their own credentials and is represented inside your tenant as a guest object[1].
Most articles on this subject stop there. That is where the practitioner question begins.
The real dividing line is identity, not the network
Think about what your analytics layer sees when each of these two populations opens the same page.
An employee arrives with an Entra ID account that was created by a joiner process and is synchronised from the HR system. That account carries department, job family, office location, country of employment, manager, cost centre, contract type and preferred language. Every one of those fields is a segmentation axis. You can ask whether the message reached manufacturing in Poland, whether it reached people on fixed-term contracts, whether German-language readers behaved differently from French-language readers.
A franchisee arrives as a guest object. That object holds an email address, a display name, and an invitation date. Sometimes a company name, if whoever raised the invitation bothered to type one. There is no manager, no department, no country of employment, no contract type, because none of those concepts apply to someone your organisation does not employ. Guests are frequently created ad hoc by a site owner rather than by any governed process.
So for one half of your audience the analytics can answer “which countries read it”. For the other half it can answer “how many sessions”. Those are not the same class of answer, and no amount of dashboard work closes the gap, because the attributes were never captured in the first place.
Why this hurts most in the businesses that need it most
In a franchise network, a dealer network, an agent network or a clinical partner network, the extranet population is the population that touches revenue and represents the brand to the customer. The people you have the least direct authority over are the people whose behaviour matters most. Gallup’s workplace research has consistently found engagement to be weakest in frontline and deskless populations, which is exactly where this measurement blind spot sits[2]. The asymmetry runs in the wrong direction relative to business importance.
| Dimension | Intranet population | Extranet population |
|---|---|---|
| Identity source | Entra ID, synchronised from the HR system | Guest object created per invitation |
| Attributes available | Department, country, site, grade, manager, language, contract type | Email domain, display name, invitation date |
| Denominator | Headcount, known and actively maintained | Unknown, drifts with guest sprawl and shared logins |
| Lifecycle | Joiner, mover and leaver process | Invitations issued freely, rarely revoked |
| What you can report | Reach and participation rate by segment | Page views and unique visitors |
| What you cannot report | Very little, if the attributes are clean | Which sites, regions or roles actually engaged |
Four ways the numbers go wrong when you run both
The denominator problem. Reach is a ratio, and a ratio needs a defensible denominator. For the intranet you have one: headcount, owned by HR, reconciled monthly. For the extranet you have a guest count that simultaneously overstates and understates the audience. It overstates because stale guests from a supplier you stopped working with in 2023 are still in the directory. It understates because a franchised site with fourteen staff often operates a single shared login.
The shared-login distortion. That single shared login is worth dwelling on. Fourteen people read the brand refresh brief on the manager’s screen during a morning huddle. Your analytics records one visitor. Every engagement rate you calculate for that population is wrong by an unknown factor, and it is wrong in the direction that makes your work look ineffective.
The segmentation cliff. The moment a report is filtered by region, function or language, the extranet half of the audience silently disappears. Nobody notices, because the chart still renders. The IC team ends up presenting a segmented view of the corporate population and calling it a view of the network. This is the same failure pattern we describe in the piece on measuring frontline worker communications, where the hardest-to-reach population is also the least visible in the data.
History that does not survive the guest lifecycle. Guests get deleted during an access review and reinvited three weeks later with a new object identifier. Any cohort analysis that depends on following the same people over time quietly breaks. You cannot show a two-year trend for a population whose identifiers reset.
What to do about it
Build a partner directory and treat it as the HR file for the extranet. This is the highest-value action and it is not a technical project. The minimum viable version is a spreadsheet with one row per site: partner identifier, site name, region, country, language and approximate headcount. Someone in franchise operations or channel management already maintains most of it for commercial reasons. Ask for it. Once that file exists, extranet activity joins to organisational structure the same way employee activity does, and the segmentation cliff disappears.
Agree the denominator in writing before the first report. Decide whether you are reporting against sites or against individuals, write the definition down, and put it on the slide. A number whose definition is contested in the room is worth less than a smaller number everyone accepts. Our guide to audience segmentation for internal communications sets out how to structure the attribute model before you start collecting anything.
Report at site level where person level is not honest. If shared logins make individual reach unmeasurable, stop pretending otherwise. “212 of 300 franchised sites opened the brief within five working days” is a defensible sentence, it is operationally actionable, and a regional manager can do something with it on Monday morning. It is a better measure than a person-level percentage you cannot stand behind.
Report the two populations separately, then combine deliberately. Blending employee and partner numbers into a single reach figure produces something that means nothing. Show them side by side with the denominator stated for each, and combine only when the question genuinely spans both. The approach in our note on cross-channel measurement applies here almost unchanged.
The honest position on Microsoft native analytics
SharePoint site analytics is free, already switched on, and adequate for a single site owner who wants to know whether a page is being read. If you run one extranet site for one supplier, open the usage panel and you are done.
The limits show up at scale and at the boundary. Site analytics is scoped per site, so an extranet built as 300 site collections gives you 300 separate usage views and no rollup. Guests appear as unique visitors with no organisational attributes attached, so no segmentation is possible from the native data alone. And the intranet numbers, the extranet numbers, the Viva Engage numbers and the newsletter numbers all live in different places, which is the structural reason native SharePoint analytics runs out of road for anyone reporting at group level. Gallagher’s State of the Sector research has repeatedly found measurement to be the capability gap internal communicators name first, and this is one of the concrete reasons why[3].
Teams often try to close the gap with a Power BI build against the Microsoft Graph reporting API. That route can work and some 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 in front of an executive audience. A dependable build needs a proper ingestion process, scheduled reconciliation against the native figures and a named owner. Budget for the pipeline rather than the dashboard. Our honest guide to SharePoint analytics in Power BI goes through 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 accepts a second attribute file for the partner population so the extranet stops being a blind spot. History is kept across tenant changes and product transitions. Tryane 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 through Azure AD or Entra ID single sign-on with no agent to install. Before joining any partner data, read our note on security and compliance for internal communications analytics.
Frequently asked questions
Is an extranet just an intranet with external users added?
Technically it can be built that way, and in Microsoft 365 many are. Operationally it behaves differently, because the people arriving through it have no HR record, no manager in your hierarchy and no governed lifecycle. The technical similarity is what misleads teams into assuming the reporting will be similar. It will not be.
Can SharePoint site analytics tell me which franchise sites read a page?
Not on its own. Native usage data records visits and unique visitors, and guest accounts carry no site, region or role attribute. To answer a site-level question you need to join the activity to a partner directory that holds those attributes, either by exporting and joining the data yourself or by using an analytics layer that performs the join.
Should the extranet be part of the internal communication remit at all?
It depends on whether that audience represents your brand to a customer. In franchise, dealer, agent and clinical partner models it does, which means those people need the same message discipline and the same measurement as employees even though they sit outside the payroll. For a pure supplier portal holding procurement documents, the answer is usually no.
How do we handle GDPR when the extranet population is not our employees?
The lawful basis and the information notice differ, because your relationship with a franchisee is contractual rather than employment based. Most groups handle this through the partner agreement and report at site level rather than individual level, which reduces the personal data footprint considerably. Involve your data protection officer before the first join, not after.
Do we need two separate measurement systems?
No, and running two is how organisations end up with numbers that contradict each other. You need one measurement layer that accepts two attribute sources: the HR file for employees and the partner directory for the extranet population. The activity data already comes from the same Microsoft 365 estate.
Sources
• Microsoft Learn, B2B collaboration in Microsoft Entra External ID
• Gallagher, State of the Sector
• Gallup, State of the Global Workplace
Further reading
• Audience segmentation for internal communications
• Measuring frontline worker communications
• Measuring cross-channel internal communications
• Security and compliance for internal communications analytics
• Internal communication KPIs for 2026
Tryane runs a 15-minute working session with Heads of Internal Communication, mapping which parts of your audience carry organisational attributes and which do not, on your own tenant. Book a slot with Jérémy to schedule yours.