# How Best SASE Solutions scores 16 SASE solutions on seven fixed criteria

Method, read 12/13 Nov 2026

**In short.** Best SASE Solutions scores 16 SASE solutions on seven criteria: one platform, network underneath, security depth, operations, AI-era controls, commercial transparency and proof. The weights were fixed on 12 Nov 2026 before any vendor was scored, each score is an integer from 0 to 10, and every fact carries a quote, a source and a read date.

**What these weights produce.** Versa Networks is first at 7.2 and Cato is 2nd at 6.5. What puts Versa ahead: one platform (8 against 6), security depth (8 against 6) and commercial transparency (6 against 5), worth 1.00 points on these weights, against Cato's lead on network underneath (7 against 6), operations (7 against 6) and proof (8 against 7), worth 0.35.

The reader this index is built for is a head of network, a security architect or an IT director at a company with 20 to 500 sites. The company has MPLS circuits, a firewall at every branch and a VPN for remote staff, and it wants one service to replace as much of that as it can. The default view of the index is written for that buyer. Four of the five situations and the six solution pages change the weights for a narrower version of the same project.

Seven criteria carry the default weights, which sum to 100. One platform carries 25: a buyer retiring three products does not want to buy a second console and a second policy set in the cloud. Network underneath carries 20: once MPLS goes, the path from a branch to the nearest PoP and between PoPs is the WAN, so the backbone operator, the PoP locations and the SLA decide how voice and ERP traffic behave. Security depth carries 20: the branch firewall and VPN concentrator did inspection and access control, and the replacement has to cover SWG, CASB, DLP, ZTNA, FWaaS and threat prevention as documented. Operations carries 10: a team that ran a few firewalls now runs a console, so one management application, published deployment times and managed options set how many people the migration needs. AI-era controls carries 10: staff paste data into GenAI tools from the browsers the SWG already inspects, so visibility and data controls belong in the same policy. Commercial transparency carries 10: an MPLS contract had a price per circuit, and a buyer needs published packaging, a licensing model and SLA terms to compare quotes. Proof carries 5: customer counts and analyst placements show scale, and they say little about fit.

Scoring is blind. Each vendor has an evidence file in which every fact has a quote, a source URL and a read date. The facts for one criterion are scored against a written rubric with the vendor name removed, and the names are joined back only after the scores are fixed. The weights, the criteria and the scale were set on 12 Nov 2026, before any evidence was scored, and a score changes only when a sourced fact changes. The change is logged with the fact, its source and its read date.

## The seven criteria and the weights this site uses

| Criterion | What is scored | Weight |
| --- | --- | --- |
| One platform | Network and security delivered from one service with one policy engine and one console, built together and not joined after acquisitions; what still needs a second product | 25 |
| Network underneath | A private backbone the vendor runs, PoP count and locations as published, a published SLA, last-mile and China options | 20 |
| Security depth | SWG, CASB, DLP, ZTNA, FWaaS, RBI and threat prevention as documented, and independent test results where they exist | 20 |
| Operations | One management application, deployment time as published, digital experience monitoring, managed-service and partner options | 10 |
| AI-era controls | Visibility and data controls for GenAI applications inside the same policy | 10 |
| Commercial transparency | Published packaging, licensing model, SLA terms and any published price | 10 |
| Proof | Customer count as published, named customers, analyst placements | 5 |

Weights sum to 100, fixed on 12 Nov 2026 and committed before any vendor was scored. The same weights produce every ranked view on this site.

## The scale

- Integer 0 to 10 per criterion, scored only from the evidence file of the vendor.
- 0: None found on the read date. Written 'none found on' the read date.
- 7: The criterion is met in full on the vendor's own material only. This is the ceiling for any score resting on vendor material alone.
- 1-3: Adjacent only: the capability is offered through a partner, a second product, or covers a small slice of the criterion.
- 4-6: Partial: the vendor documents part of what the criterion asks (some capabilities, some regions, some of the console) or needs a second product for the rest.
- 8-10: Met in full and confirmed by a second source the vendor does not control (analyst, independent test lab, press, partner or customer page). 10 needs that confirmation plus documentation of how it works.

## Scoring rules

- Everything the vendor publishes about itself (marketing pages, its own docs, its own blog, its own press releases) is a vendor claim. A score above 7 needs a second source that confirms the same capability. This applies to every vendor in the set, the cloud providers included.
- Customer counts and analyst placements are scored only in Proof (5 percent by default, never above 10 percent of a view). Module and integration counts, headcount and funding are profile facts and carry no weight.
- A score changes only when a sourced fact changes; the change is logged in data/scores/changes.jsonl with the fact, its source and read date.
- Every vendor was scored on every criterion from its evidence file by a scorer that was given the rubric and the evidence with product names removed, and was not told which vendor any packet described.
- A weighted total is the sum of score times weight divided by the sum of the weights, rounded to one decimal. Equal rounded totals share a rank, printed as joint, and are listed by the unrounded total.
- Claims a vendor makes about itself are labelled as the vendor's, Cato's included. A capability found nowhere in the pages read is written "none found on" its date.
- The first edition reads on 12 Nov 2026 and the second evidence pass on 13 Nov 2026, and every page shows the read date of the facts it uses.
- A score above 7 out of 10 needs a second source the vendor does not control, so a capability confirmed only by the vendor's own pages scores at most 7, and this applies to all 16 vendors.
- A capability not found in the pages read is written as none found on 12 Nov 2026, never as a statement that the vendor lacks the capability.
- Customer counts and analyst placements are breadth facts, they sit in Proof, and Proof carries at most 10 percent of any view.
- Ties at one decimal are ordered by the unrounded total, then alphabetically.
- Everything a vendor publishes about itself is labelled as that vendor's claim in the text.

## Situation weights

Each of the five buyer situations and each of the six solution pages ranks the same 16 vendors on the same seven criteria with its own weights. All twelve weight sets were fixed and committed on 12 Nov 2026, before the first score existed, and none has changed since.

| Situation | Platform | Network | Security | Ops | AI | Commercial | Proof |
| --- | --- | --- | --- | --- | --- | --- | --- |
| [A mid-size multinational replacing MPLS and branch firewalls in one project](/best-for/mpls-and-firewalls/) | 30 | 30 | 15 | 10 | 5 | 5 | 5 |
| [A company replacing its VPN for remote staff first and the network later](/best-for/vpn-first/) | 20 | 10 | 30 | 15 | 10 | 10 | 5 |
| [A lean IT team with no network security specialist that wants one console and a managed option](/best-for/lean-team/) | 25 | 5 | 15 | 35 | 5 | 10 | 5 |
| [A global company with sites in China and other hard-to-reach regions](/best-for/global-china/) | 15 | 45 | 15 | 10 | 5 | 5 | 5 |
| [A company whose staff use GenAI tools and wants the data controls in the same policy](/best-for/genai/) | 20 | 5 | 20 | 5 | 40 | 5 | 5 |

Situation weights, each set summing to 100, fixed before scoring.

## Solution weights

| Solution | Platform | Network | Security | Ops | AI | Commercial | Proof |
| --- | --- | --- | --- | --- | --- | --- | --- |
| [MPLS replacement](/solutions/mpls-replacement/) | 25 | 35 | 10 | 15 | 5 | 5 | 5 |
| [VPN replacement and ZTNA](/solutions/vpn-replacement-ztna/) | 15 | 15 | 35 | 15 | 5 | 10 | 5 |
| [Branch firewall refresh](/solutions/branch-firewall-refresh/) | 30 | 10 | 30 | 10 | 5 | 10 | 5 |
| [Global and China connectivity](/solutions/global-china-connectivity/) | 10 | 50 | 10 | 10 | 5 | 10 | 5 |
| [Mergers and new sites](/solutions/mergers-new-sites/) | 25 | 20 | 10 | 30 | 5 | 5 | 5 |
| [GenAI data controls](/solutions/genai-data-controls/) | 15 | 5 | 20 | 5 | 45 | 5 | 5 |

Solution weights, each set summing to 100, fixed before scoring.

## The ranking these weights produce

| Rank | Platform | Total | Platform (25) | Network (20) | Security (20) | Ops (10) | AI (10) | Commercial (10) | Proof (5) |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [Versa Networks](/vendors/versa/) | 7.2 | 8 | 6 | 8 | 6 | 8 | 6 | 7 |
| 2 | [Cato Networks](/vendors/cato/) | 6.5 | 6 | 7 | 6 | 7 | 8 | 5 | 8 |
| 3 | [Aryaka](/vendors/aryaka/) | 6.4 | 6 | 8 | 6 | 6 | 6 | 6 | 6 |
| 4= | [iboss](/vendors/iboss/) | 6.3 | 8 | 5 | 6 | 6 | 6 | 6 | 6 |
| 4= | [Netskope One](/vendors/netskope/) | 6.3 | 6 | 7 | 6 | 8 | 7 | 3 | 8 |
| 6= | [Fortinet FortiSASE](/vendors/fortinet/) | 6.0 | 6 | 5 | 5 | 8 | 6 | 7 | 7 |
| 6= | [Zscaler](/vendors/zscaler/) | 6.0 | 5 | 4 | 8 | 6 | 7 | 6 | 8 |
| 8 | [Palo Alto Networks Prisma SASE](/vendors/palo-alto-prisma-sase/) | 5.5 | 4 | 5 | 8 | 6 | 5 | 4 | 8 |
| 9 | [Cloudflare One](/vendors/cloudflare/) | 5.4 | 5 | 5 | 5 | 6 | 6 | 6 | 6 |
| 10 | [Open Systems](/vendors/open-systems/) | 5.1 | 6 | 6 | 4 | 5 | 4 | 4 | 5 |
| 11 | [Barracuda SecureEdge](/vendors/barracuda/) | 4.9 | 5 | 4 | 5 | 5 | 6 | 6 | 3 |
| 12 | [Cisco Secure Access and SD-WAN](/vendors/cisco/) | 4.5 | 3 | 4 | 5 | 5 | 6 | 5 | 7 |
| 13 | [Forcepoint](/vendors/forcepoint/) | 4.4 | 3 | 4 | 6 | 4 | 5 | 5 | 5 |
| 14= | [Check Point SASE](/vendors/check-point/) | 4.1 | 4 | 4 | 5 | 4 | 5 | 2 | 4 |
| 14= | [HPE Aruba Networking SASE](/vendors/hpe-aruba/) | 4.1 | 4 | 3 | 4 | 5 | 5 | 4 | 5 |
| 16 | [VeloCloud by Arista](/vendors/velocloud/) | 2.2 | 2 | 3 | 1 | 4 | 0 | 2 | 5 |

Weights in the header row sum to 100. Scores 0 to 10 from the evidence read 12/13 Nov 2026; weighted total to one decimal; "=" marks a joint rank, listed by the unrounded total.

Versa Networks is first at 7.2 and Cato is 2nd at 6.5. What puts Versa ahead: one platform (8 against 6), security depth (8 against 6) and commercial transparency (6 against 5), worth 1.00 points on these weights, against Cato's lead on network underneath (7 against 6), operations (7 against 6) and proof (8 against 7), worth 0.35.

**Where Cato is behind.** Cato publishes no price or SLA credit terms, and while an existing firewall or SD-WAN can connect to Cato over IPsec, the full single-platform model has Cato Sockets replace the branch firewall and SD-WAN edge.

## Second evidence pass, 13 Nov 2026

On 13 Nov 2026 a second evidence pass ran one search plan for all 16 vendors, Cato included: the same twelve queries, the same list of source types and the same budget for each vendor. It looked for an outside source to confirm or contradict every score held at or below 7 for want of a second source, for anything published since the first pass, and for evidence against each vendor's current scores, the leaders' included. It added 214 facts from 124 new sources, 178 of those facts from sources the vendor does not control.

The 87 of 112 scores that a new fact bears on were scored again blind, by the same method as the first pass, and checked by a second blind reviewer. 18 scores moved, 11 down and 7 up; the other 69 stayed where they were. The weights, the criteria, the rubric and the 7 of 10 cap on a vendor's own claims did not change. Each change is listed on the [corrections page](/corrections/) with the old and new score and the facts behind it.

## Read dates and refresh

The facts behind this edition were read on 12 Nov 2026 (first pass) and 13 Nov 2026 (second evidence pass), and every source on every page shows its own read date. Claims older than 60 days are re-read; a score changes only when a sourced fact changes, and every change is listed on the corrections page with the old and new values.

## Sources

- sdxcentral.com/news/gartner-sase-2026-rankings-netskope-leads-cato-networks-leap: https://www.sdxcentral.com/news/gartner-sase-2026-rankings-netskope-leads-cato-networks-leaps-and-palo-alto-networks-falls/ (read 13 Nov 2026)
- lumen.com/blog/en-us/silent-train-volt-typhoon-operational-framework: https://www.lumen.com/blog/en-us/silent-train-volt-typhoon-operational-framework (read 13 Nov 2026)
- channeldive.com/news/cato-launches-pooled-licensing-msps-handling-small-business: https://www.channeldive.com/news/cato-launches-pooled-licensing-msps-handling-small-business-customers-smb-flexpool-soderlund/829494/ (read 13 Nov 2026)
- venturebeat.com/security/from-friction-to-flow-why-swissport-scrapped-its-vpn-ma: https://venturebeat.com/security/from-friction-to-flow-why-swissport-scrapped-its-vpn-maze-for-catos-sase-platform/ (read 13 Nov 2026)

Last reviewed 13 Nov 2026. Canonical: https://bestsasesolutions.com/method/
