Managed IT
The MPLS invoice is the easy number to compare, so it is usually the only one in the spreadsheet. The bandwidth premium is real, but a private WAN also costs you in headquarters internet capacity, carrier ticket queues and the day a single circuit fails. This is what belongs in the comparison, and a checklist to find out where your own network stands.
Ask for an MPLS quote and an SD-WAN quote and you get two prices for two different things. One covers a circuit. The other covers a managed network. Setting them side by side and picking the smaller number answers a narrower question than the one you have.
The MPLS bill is not the MPLS cost
Gartner puts MPLS at 3-5x more expensive per Mbps than broadband, and that premium is the part everyone finds. It is on the invoice, it repeats every month, and it is easy to compare. IDC puts the typical WAN cost reduction at 40-60% when organizations replace legacy MPLS with managed SD-WAN.
Treat both as starting points. What a specific network saves depends on how many sites it has, what each one can get for broadband or fiber, and how far into its MPLS terms it is.
Why comparisons stop at the circuit
Circuits are the cost that arrives with a name and a due date. The others are real, but they land in different places. The headquarters internet bill sits in one budget, outage hours in nobody's, and a project that waits weeks on a carrier shows up as a delay rather than a charge. No single line holds the total, so nobody adds it up.
That is how MPLS ends up looking like the predictable option in a spreadsheet that leaves out several of the things that make it costly to live with.
Three costs the comparison leaves out
What Gets Missed
Three costs are easy to miss because none of them appears on an MPLS invoice. Each one comes from how a traditional WAN is built, not from what the carrier charges.
Easiest to measure
Backhaul
Sized for every site combined · Latency in both directions · Grows with each cloud app · Not on the carrier invoice
On a traditional WAN, a branch reaches the internet by going through headquarters or a data center first. IDC finds that 80% of enterprise network traffic now goes to cloud and SaaS, so on a typical network the private circuit spends most of its time carrying traffic on a detour. Every Teams call and Salesforce lookup adds latency twice, on the way in and on the way out.
The bill for that detour arrives in two places. Headquarters needs internet capacity sized for every site's cloud traffic combined, and that number grows each time another application moves to the cloud. And users at the branches pay in slower calls and slower pages, which is a cost even though no invoice carries it. Direct cloud breakout from each site removes both.
Least budgeted
Single-circuit sites
One carrier per site · Ticket queue on failure · Cost varies by site · Rarely in the comparison
Most MPLS sites run on one circuit, because a second MPLS circuit is a second MPLS bill. So the site accepts the risk. When that circuit fails, the site is offline until the carrier's ticket queue reaches it, and the length of that wait is not yours to control.
The cost is simple to estimate and almost never estimated: people affected, times hours down, times what an hour of their time costs. Put that figure next to the price of a second link. Broadband costs a fraction of MPLS per Mbps, so the redundant pair on the SD-WAN side is often cheaper than the single circuit it replaces, and the site fails over automatically when one link drops.
Slowest to notice
Change requests
Carrier provisioning · Truck roll per site · Multi-year terms · Manual per-site configuration
Adding a site, moving one or raising its bandwidth on MPLS goes through the carrier. Provisioning and a truck roll turn a simple expansion into a project measured in weeks. The circuit fee is only part of that cost. The rest is the site that opened without a network, or the acquisition that waited to be connected.
Terms are the other half. A multi-year carrier contract locks in pricing and conditions long after the business has changed shape. On the SD-WAN side, a new site is software-defined and comes up in days, and terms scale with the business instead of against it.
What the three have in common
Each one is a cost of the network's design, not of its price. That is why negotiating the circuit rate harder does not remove them. A cheaper MPLS circuit is still a single circuit that sends cloud traffic through headquarters and waits on a carrier when something needs to change.
Comparing like for like
A fair comparison puts the same line items on both sides, and puts a number on the ones that never arrive as an invoice.
What goes on each side
Circuits. Every MPLS circuit against the broadband, fiber or LTE links that would replace it, including the second link at sites that would get one.
Equipment and management. SD-WAN is typically priced per site per month, covering hardware, software licensing, management and monitoring. On the MPLS side, that means router maintenance and whoever administers it, whether that is your team or the carrier's managed service.
Headquarters internet capacity. The extra bandwidth that carries every site's cloud traffic under MPLS, and what it drops to when sites break out locally.
Time. Hours spent on change requests, carrier tickets and new-site turn-ups.
Outages. People affected, times hours down, times loaded hourly cost, for each site on a single circuit.
Mind the contract dates
MPLS is usually sold on multi-year terms, and leaving one early can trigger termination charges. Terms and notice windows differ by carrier and agreement, so read yours before planning anything. The cleaner path is to run both networks in parallel and retire each MPLS circuit when its term ends, ordering the migration by renewal date instead of by geography.
Parallel running is also how a migration avoids downtime. The legacy WAN stays live until the new one is validated at each site, so the overlap is a cost of the changeover that belongs in the comparison, not a surprise later.
When MPLS still makes sense
For some sites it does. A location with no viable broadband or LTE, or an application whose contract ties a latency guarantee to the private circuit, may keep MPLS. SD-WAN does not require removing it. It can run over existing connections, add circuits where redundancy is needed, or replace MPLS entirely. The decision is made site by site, so the question becomes which sites move first rather than whether the whole network does.
WAN Readiness Checklist
The Checklist
Fifteen questions across circuit cost, performance, redundancy and cloud readiness. Tick the ones you can answer without going and asking someone, and the score updates as you go. Use it yourself or share it with your team.
Tick what you can answer without checking. The blanks are the numbers your comparison is missing.
Circuit Cost & Contracts
What does each site's WAN circuit cost per Mbps?
Divide the monthly charge by the bandwidth. The gap to a broadband quote in the same town is where the comparison starts.
When does each circuit's contract end?
A date per circuit, not per carrier. Renewal dates decide the order sites can move without termination charges.
What are you paying beyond the circuit itself?
Port and access charges, router maintenance and managed-service fees often sit on separate lines.
Does anyone own the total cost of the WAN?
Circuits, headquarters internet, support time and outages usually sit in different budgets, so nobody sees the sum.
Performance & Backhaul
How much of your traffic goes to the internet rather than to another site?
Pull a week of data. Whatever share is bound for Microsoft 365, Teams and other SaaS is riding your private circuits to headquarters first.
Is headquarters internet capacity sized for every site's cloud traffic?
It has to carry the combined total, and it grows each time another application moves to the cloud.
Do Teams calls degrade at branch sites but not at headquarters?
That pattern points at the detour, not at the application.
Can you see performance by site and by application in one place?
A separate carrier portal per circuit rarely adds up to a single view.
Redundancy & Failover
Which sites run on a single circuit?
Each one is a carrier fault away from being offline.
What happens when a circuit fails at 4 p.m. on a Friday?
A ticket, a queue and a wait. Know who you call and what the carrier commits to in writing.
Have you ever costed an outage at one of your sites?
People affected, times hours down, times loaded hourly cost. It rarely appears in a comparison at all.
Are backup links on a different carrier from the primary?
Two circuits from one carrier often share a route, a building entry or an upstream network.
Cloud Readiness
Which cloud apps does each site depend on every day?
Microsoft 365, Teams, telephony, line-of-business SaaS. That list decides what should break out locally.
Is there a firewall at each site that could inspect direct internet traffic?
Local breakout moves the perimeter to every location, and security has to move with it.
How long does it take today to bring a new site online?
Carrier provisioning and a truck roll take weeks. If a new site or an acquisition has ever waited on that, it belongs in the comparison.
Frequently asked questions
Not always, and a fair comparison can come out differently for different sites. Broadband-based transport costs less per Mbps, and IDC puts the typical WAN cost reduction at 40-60% when legacy MPLS is replaced with managed SD-WAN. A site with limited broadband options, or one that recently renewed a multi-year MPLS term, can narrow the gap. The comparison that matters is total cost across every site, including the costs that never appear on the circuit invoice.
Three. Headquarters internet capacity sized to carry every site's cloud traffic, because branch traffic is backhauled through it. The cost of outages at sites that run on a single circuit, which is people affected times hours down times loaded hourly cost. And the time spent on carrier change requests and new-site provisioning, which take weeks on MPLS. None of them appears on an MPLS invoice, so they rarely make it into the spreadsheet.
No. SD-WAN can run over your existing MPLS circuits alongside broadband, add circuits where redundancy is needed, or replace MPLS entirely. Sites that need a private circuit, such as a location with limited broadband or an application with a contractual latency guarantee, can keep one while the rest move. The pace is often set by contract end dates, so the question becomes which sites move first.
Sequence the migration by renewal date. Run SD-WAN in parallel with the legacy WAN and retire each MPLS circuit when its term ends instead of cancelling it early. Check every contract for an auto-renewal clause and its notice window, because a missed window can renew the contract for another full term. Terms differ by carrier and agreement, so read yours before deciding the order.
A single broadband link is less predictable than a dedicated circuit, which is why SD-WAN does not rely on one. It monitors every link and reroutes traffic to the next best connection within seconds when one degrades or fails. Reliability comes from having two links, ideally on different carriers, instead of one perfect one. For a site that ran on a single MPLS circuit, two links is more resilient than one.
Add five things for each site. The circuit and any port, access, equipment or management charges. The share of headquarters internet capacity that exists to carry that site's cloud traffic. Staff hours spent on change requests and carrier tickets. The cost of an outage, worked out as people affected times hours down times loaded hourly cost. And how much of the contract term remains. The checklist above walks through the questions that produce each figure.
Aureon customers typically see net savings within the first year, though the timing depends on your contract dates. While an MPLS circuit is still under term you pay for both networks, and the length of that overlap affects when the savings show up. Ordering the migration by renewal date keeps the overlap short.
Itemized pricing for both sides, so circuits, ports, equipment, licensing and management appear as separate lines. The termination date and notice window for every existing circuit. What each proposal commits to in writing on repair times and failover. And a list of which sites would keep a private circuit and why. Quotes built from the same line items can be compared. Quotes that bundle different things cannot.
What does your WAN actually cost?
Aureon maps your circuits, sites and cloud dependencies and builds a cost baseline, so you can compare total cost against managed SD-WAN with real numbers.
