The Network Is Up. The Business Is Down

By Stu Long, CTO, Orro

Most people who have run technology in a large organisation have heard this sentence, or said it themselves: “I know your business isn’t working, but the network is up.” It is usually accurate. It is also the least useful thing a technology partner can say. And it is the clearest sign that the business now expects experience assurance, not device uptime.

Picture the person hearing it. The contact centre queue is climbing. Agents are watching a spinning wheel inside the application they use to take calls. Store teams can’t process returns. Someone senior wants an update every fifteen minutes. On the other side of the call, an engineer is looking at a dashboard where every switch, router and access point is green. The engineer is not wrong. The network is up. The business is down.

Customers are done with that conversation, and they are right to be. A green light on a device dashboard tells you the infrastructure is available. It does not tell you whether the service is working for the person trying to use it. Uptime has become a hygiene factor. Experience is the measure.

Up is not the same as working

Orro has written before about why network uptime is the wrong thing to measure. This piece is about what follows. Once uptime stops being the measure, the question the business asks of its IT team and its partners changes too. It used to be “is it up?” Now it is “is it working, and is it working well for the people using it?”

The shift has happened because of where work now runs. Core business processes sit in SaaS platforms and in Azure, AWS and Google Cloud. Users are in offices, stores, homes, depots and vehicles. Between a person and the application they rely on sit Wi-Fi, a local network, a carrier, a security edge, a cloud region and a provider’s own backend. Most of those segments have different owners.

Every one of them can be healthy on its own terms while the service fails. Cisco ThousandEyes saw exactly this during a global SaaS disruption in March 2026. Network paths to the service appeared to stay healthy throughout, while its backend failed and the service became unavailable (ThousandEyes, 2026). Measured at the network, nothing was wrong. Measured at the user, very little was right. That gap is exactly what experience assurance exists to close.

The hours spent proving it wasn’t us

Consider a national retailer that runs much of its business through a single SaaS platform. Its contact centre and unified communications sit inside it. Its question used to be whether the network was up. Now it asks whether traffic is reaching the platform, and whether the experience lets an agent serve a customer. It is an experience assurance question, and device monitoring cannot answer it.

Too often, my team is called into problems that aren’t network problems at all, but a slow response from a cloud or SaaS service. Without visibility of that service, the first hours of an incident go on proving it isn’t the network. They should go on showing where the problem actually is. Orro’s network operations team can quickly show the switches are healthy. That evidence is necessary, but it doesn’t restore a single customer call.

ThousandEyes saw the same instinct in a separate 2026 incident, where the failing layer sat behind the affected organisations’ own IP addresses, so they had every reason to look inward first (ThousandEyes, 2026).

Meanwhile, the business usually knows before IT does. In Splunk’s 2026 downtime research, 47 per cent of technology executives at Global 2000 companies said customers often or very often detect service degradation or outages first (Splunk, 2026). The same study found that only 38 per cent consistently identify the root cause of downtime (Splunk, 2026). When the customer is your monitoring system, every minute spent proving innocence is a minute the customer spends waiting.

This is also changing how customers value support. The conversations I have now are less about a cheaper price per ticket. They are about a shorter time to resolution, and about the tickets that never need to be raised. That is experience assurance expressed in commercial terms.

Consolidation can quietly narrow the view

The major platforms see the same problem and are responding to it. Cisco and Microsoft are each building out a data fabric and bringing more of their tooling together around it. Consolidation is a sensible answer to tool sprawl, and it is well under way. Splunk’s 2026 observability research included Australian and New Zealand organisations. In it, 67 per cent of respondents said they use a single observability platform (Splunk, 2026).

Consolidation does reduce silos. What it can also do, quietly, is replace tool silos with vendor-shaped visibility. Each fabric is strongest across its own ecosystem. That is a fair consequence of how good platforms are built, not a criticism of anyone. In practice, organisations that standardise on a cloud monitoring tool for their applications often find it doesn’t cover the network well. The reverse is just as common. Different teams add another layer, because IT operations, cloud and DevOps teams often want to run these platforms differently, for sound reasons of their own.

The work itself is already crossing those lines. Observability teams troubleshooting alongside security teams rose to 77 per cent in Splunk’s 2026 research, up from 62 per cent (Splunk, 2026). The user’s experience doesn’t respect team or vendor boundaries, so experience assurance can’t either.

Our position at Orro is to work with the data fabric a customer already has, or whichever one makes most sense for their environment. Forcing a single platform rarely serves the customer. The value lies in designing, integrating and operating the view across them. Who owns that view is a question in its own right, explored in The Blind Spots Live Between the Platforms.

Digital experience assurance starts at the browser

“End to end” has become a phrase that means everything and nothing, so it is worth being precise. It means seeing the service the way the user experiences it. That starts at the browser, where the user actually sees whether a page loads and a transaction completes. It runs through the network path. It ends at the cloud or SaaS application, and whether it responds quickly enough. The switch is one vantage point for experience assurance. It is no longer the most important one.

Done well, experience assurance lets a team act before anyone complains. Take a Wi-Fi network that notices a user failing to connect again and again. It can help them before they ever raise a ticket. Apply that thinking to the journeys a business depends on, such as signing in, completing a sale or connecting a call. The operating model then shifts from responding to tickets to preventing them.

It also has to reach beyond what an organisation owns. ThousandEyes has found that several recent major outages came from systems interacting in unexpected ways, not from any single component failing (ThousandEyes, 2026). Its conclusion is that understanding them takes visibility across the full service delivery chain, including dependencies outside an organisation’s direct control.

The same logic is about to extend to a new kind of user. As AI agents begin acting on the business’s behalf, they become both consumers of services and dependencies in their own right. That is the subject of Who’s Watching the Agents? in this edition.

Experience assurance in 2027 means outcomes, not devices

If experience is the measure, assurance has to be rebuilt around it. For most of the industry’s history, a service level has been a promise about devices. Network availability, response times for a fault and a monthly uptime percentage still matter. They just no longer describe what the business is buying.

In 2027, I expect the most useful service levels to be written around outcomes. How quickly a degraded experience is detected, ideally before a user reports it. How long it takes to establish where the fault actually sits, whoever owns it. How long until the service works again for the people who rely on it. How many tickets never needed to exist. The research points the same way. Among the organisations with the lowest downtime costs, 98 per cent rate end-to-end visibility as very or extremely important (Splunk, 2026).

This asks something of technology leaders as well as their partners. It means deciding which user journeys matter most and agreeing what good looks like for each. Above all, it means making someone accountable for the whole service, not just one domain within it. That is experience assurance as an operating commitment rather than a dashboard.

Visibility is the first step

None of this means the green light was a mistake. Network availability remains the foundation everything else stands on, and it deserves the same engineering discipline as ever. It simply isn’t the finish line any more.

The organisations that get ahead of this won’t start by buying another tool. Instead, they will decide that the user’s experience is the thing they measure. Then they will build a view that can actually see it. Visibility is the first step. You cannot assure a service you can’t see from end to end. Nor can you hold anyone accountable for an outcome nobody is measuring.

If your organisation is starting to rethink what experience assurance should mean, that is a conversation worth having. You can talk it through with Orro.

---

Sources & Further Reading

1. Orro. (2026). The Blind Spots Live Between the Platforms.
2. Orro. (2026). The Black Box Problem: Why Network Uptime Is the Wrong Thing to Measure.
3. Splunk, a Cisco company. (2026). Splunk State of Observability 2026: Avoiding the Hidden Costs of AI Agents (press release, research by Oxford Economics).
4. Splunk, a Cisco company. (2026). The $600 Billion Wake-up Call: New Splunk Research Reveals Downtime Is a Systemic Business Crisis (press release, research by Oxford Economics).
5. Cisco ThousandEyes. (2026). When the Cloud Goes Down: Unpacking Recent SaaS Disruptions.
6. Cisco ThousandEyes. (2026). Looking Ahead: 2026’s Biggest Outage Risks.

---