Most buyers compare features. The smart ones compare architectures. Here are eight criteria that separate platforms that scale with you from platforms that hold you hostage.
Every ISP platform on the market checks the same boxes: "billing", "RADIUS", "customer portal", "ticketing". Two years in, you'll discover that how those modules talk to each other is what decides whether your team scales or drowns.
A 5,000-subscriber WISP picks a platform on feature parity. Eighteen months later, billing and helpdesk are technically the same vendor, but they're two databases synced over cron, so a suspended customer still appears active to the support agent for 8 minutes after the billing run. The agent troubleshoots a "connectivity issue" that's actually a credit hold. The customer escalates. Twice a week.
The vendor matrix said "integrated billing and CRM". The architecture said something else.
The criteria below are the ones that show up in your daily ops, long after the demo. Ask them. They are cheap to ask. Expensive to learn the answer in production.
Use these in your next vendor demo. The "ask" line under each criterion is what to say out loud while the slide is up.
If billing, network, and support sync via cron, your team makes decisions on data that is 0–N minutes stale. Suspended accounts appear active to support. Closed tickets reappear in finance. The support agent troubleshoots a credit hold disguised as a connectivity issue.
Ask: "Show the schema. One customer record across all modules, or multiple synced IDs?"
"Integrated network monitoring" usually means a button that opens a separate vendor console in a new tab. Real depth means signal levels live inside the customer record, ticket, and dispatch packet. The agent never leaves the ERP to see whether the ONU is healthy.
Ask: "Show this customer's optical signal in this UI, last 24 hours."
The honest question is not "self-hosted vs SaaS." It is whether the platform commits you irreversibly to one mode and one vendor. A platform that supports both deployment models, lets your data stay yours, and supports clean migration between modes preserves your renewal leverage. A SaaS-only platform with per-subscriber pricing scales linearly with your success and never gets cheaper.
Ask: "Can I start as SaaS and migrate to self-hosted later? Or run self-hosted now and shift to SaaS during a team transition? Same data, same architecture either way?"
A pluggable architecture, where any OLT with a management interface can be integrated as a new plugin, protects you at hardware-vendor renewal time. The opposite is a hardcoded list of vendors the platform was built around; adding a new one is a roadmap conversation, not an integration.
Ask: "How many OLT vendors run today, and what's the path to add a new one: engineering effort or 'wait for our roadmap'?"
Some platforms let admins configure workflows, status transitions, and views in the UI. Others bill professional-services hours for the same result. The difference is your operations team's autonomy.
Ask: "Add a custom field, status workflow, and Kanban view. No code, no PS fee."
An AI that can read your contracts, billing history, network telemetry, and support transcripts is a junior analyst on every shift. An AI that can only read what you paste into a chat box is autocomplete with extra steps.
Ask: "Summarise this account's state, network plus billing, without me typing context."
"GIS support" is the polite version of "we have markers on a map". Real fibre infrastructure means cable runs, core counts, splitter trees, manholes, splice points, OTDR file import, and service-path tracing from the ONT back to the OLT. The difference is millions in OpEx for any FTTH operator.
Ask: "Trace ONT to OLT through every splice. Then upload an OTDR file."
One-currency platforms break at the first border. Regional payment rails (mobile money, SEPA, banking APIs, regional bill-payment networks) decide your collection rate. A platform that adapts currency, tax regime, and payment rails through configuration, not code forks, is built for actual ISP economics.
Ask: "Two countries, two currencies, two tax regimes, one instance?"
Sales teams are good at showing the happy path. These signals tell you what's behind it.
We didn't write this framework to make ourselves look good. We wrote ISPCQ to answer these questions for ourselves first, then ran it in production across multiple geographies, currencies, and regulatory regimes from a single codebase.
Billing, network, helpdesk, GIS, inventory, HR, and AI read and write the same tables. No sync because no sync exists. The customer record the support agent sees is the same row finance sees, the same row the NOC sees, the same row Aelita reads.
The OLT-unified plugin model treats vendors as pluggable backends, not a hardcoded list. If an OLT has a management interface (CLI, API, SNMP, NETCONF), we can integrate it. VSOL and NETIS ship today via dedicated plugins; new vendors drop in as new plugins without touching core. Optical signal lives in the customer record. Service path traces from ONT to OLT inside the GIS.
Self-host on your hardware, your VPC, your region. Or run it as SaaS while you build the team to bring it in-house. Same architecture either side of the boundary, same data model, same migration path. Your data stays yours, exportable, never trapped. Your renewal leverage stays with you.
The assistant reads the same database your team reads. No copy-paste, no separate AI integration to configure, no screen-scraping middleware. Eighteen capabilities live in production today. Data sent to the Claude API is processed and returned without retention; only the generated summary is stored, not the source data.
Cables, cores, splitters, ducts, manholes. OTDR file import with fault localisation. Trace ONT to OLT inside the customer record. Cable-fault impact analysis showing every customer affected by a single splice failure.
Multiple countries. Multiple currencies. Multiple tax regimes. Six payment gateways including mobile-money rails. Same codebase, one deployment, zero per-tenant licensing. Adding a country is configuration, not a fork.
Architecture claims are cheap. Here's what's actually shipping in production right now.
Two ISPs running in production today across two countries. Live customers, live billing, live regulators. Not a sandbox.
Our full named-competitor comparison brief is published at ispcq.com/comparison-brief, sourced from each vendor's own public docs as of the date of issue. A feature-by-feature matrix of ISPCQ against the platform you're evaluating.
If you want a tailored follow-up (pricing context, migration scenarios, or integration questions specific to your stack), email us. We respond manually within one business day. No automated sales sequences.