A filter is a decision you live with for years — it touches every lesson, every device, and your E-Rate certification. This guide gives technology directors and school leaders a structured way to evaluate options: the criteria that actually separate products, the questions that expose weak ones, and a scoring method that keeps the decision honest.
You do not need a procurement department to do this well. You need a clear picture of your environment, a short list of must-haves, and the discipline to test with your own traffic instead of a vendor's demo sites.
These are the dimensions where products genuinely differ. Weight them for your situation — a 1:1 district should weight off-campus enforcement heavily; a single school with no take-home program can relax it.
How much of the web does the filter actually know? A database spanning well over a hundred million domains makes uncategorized sites rare; a small one leaves students browsing in the gaps. Ask for the number and how it is measured.
Wrong categories cut both ways: a health-education site labeled adult blocks a lesson, a gambling site labeled games slips through. Accuracy also means multi-category awareness, since one domain can legitimately be both video and adult.
If devices go home, policy must go with them. Verify that the same rules apply on a kitchen-table Wi-Fi network as in the library — and that off-campus activity still appears in your reports.
Essay writers, homework solvers, image generators, deepfake and voice-cloning tools, and AI companion chat all need policy decisions now. Look for a dedicated, daily-updated AI category set rather than a promise to "add it soon."
When E-Rate paperwork comes due, you need category-level evidence that obscene material and content harmful to minors is blocked, with logs to back it. Reporting built for auditors saves days of manual assembly.
Cloud filtering needs no appliance and suits schools without server rooms; on-premise keeps traffic inside your network where policy demands it. The right product offers both so the choice is yours, not the vendor's. See how cloud-based web filtering for schools typically rolls out.
Many schools run filtering with a fraction of one person's time. Policy by category and grade band, sensible defaults, and quick per-site exceptions matter far more than a hundred settings nobody will touch.
Search results and image tabs are where filtered content most often leaks through. Confirm the filter forces safe search on the major engines and video platforms rather than merely recommending it.
You should be able to project three years of cost from the quote alone. Per-student pricing with the essentials included beats a low base price that meters reporting, off-campus coverage, or support as add-ons. Compare against our published pricing.
Filtering failures happen during class time. You want a vendor who answers while the incident is live, speaks to schools regularly, and treats a mis-categorization report as same-week work, not a backlog ticket.
Use this table during demos. For each criterion, one column describes a strong answer; the other describes the answer that should lower a score.
| Criterion | What good looks like | Warning sign |
|---|---|---|
| Database | 100M+ domains, updated daily, new domains classified as they appear | "Millions of sites" with no number and no update cadence |
| Accuracy | Multi-category labels; your fifty-site list handled correctly live | One label per domain; demo restricted to vendor-chosen examples |
| Off-campus | Identical policy and logging on any network | Works "behind the firewall" only, or off-campus costs extra |
| AI tools | Thousands of AI domains in dedicated categories, refreshed daily | A handful of chatbot URLs on a static list |
| Reporting | Category-level reports an E-Rate auditor accepts as-is | Raw logs you must export and explain yourself |
| Management | Policy per grade band; exceptions in seconds | Every change is a support ticket or a config file |
| Pricing | Predictable per-student cost, essentials bundled | Base price plus a surcharge for each capability you actually need |
Score us on the same criteria as everyone else. This is the data foundation you would be evaluating — and we will run your fifty-site list against it on the first call.
Count what you actually run: how many students and staff, which grade bands, Chromebooks versus Windows versus iPads, how many devices go home, and whether you have on-site server capacity or want everything hosted. A filter that fits a laptop-cart middle school will not automatically fit a 1:1 district with three buildings.
Before seeing any product, list the school content filter requirements you cannot compromise on — typically CIPA-required blocking, off-campus enforcement for school-issued devices, and audit-ready reporting. Everything else is a preference. Deciding this in advance stops a polished demo from redefining your priorities.
Bring fifty real URLs to every demo: the databases your librarians rely on, the video and news sites teachers assign, plus the problem sites from last year's incident log. Ask each vendor to show, live, how their categories treat your list — not their rehearsed examples.
Run the leading candidate with real students and teachers for two to four weeks. Track two numbers: legitimate resources blocked (over-blocking) and inappropriate content reached (under-blocking). A pilot surfaces in days what a demo hides for months.
Rate each product one to five on every criterion, multiply by the weight you assigned, and total it. The arithmetic is less important than the honesty it forces — a weak audit trail can no longer hide behind a charming account rep.
Call two schools of similar size and ask what happened the last time something broke during the school day. Response time on a Tuesday morning in October tells you more about a vendor than anything in the proposal.
Demos are choreographed; questions break the choreography. Each of these has a concrete, checkable answer, and a vendor who dodges any of them is telling you something useful. Write the answers down — they feed directly into your scoring grid.
How many domains does your database categorize, and how do you count them?
A site registered this morning — when is it classified, and what happens before that?
Can a single domain carry multiple categories, and how does policy resolve the conflict?
Walk me through policy following a school-issued Chromebook onto home Wi-Fi.
Show me the exact report you would hand an E-Rate auditor.
How are AI tools categorized, and how often does that list update?
A teacher needs one blocked site for tomorrow's lesson — what are the clicks?
How is HTTPS traffic categorized without breaking the sites students use?
The difference is who does the work. In the first answer the database absorbs the web's daily churn; in the second, your staff does — one submitted URL at a time.
Ask when a specific, recently launched site was categorized. If the vendor cannot answer, or the honest answer is "when someone reports it," students will spend their days on the youngest, least-categorized part of the web — precisely where new risks appear. A database refreshed daily is the minimum for a population that finds new sites faster than any adult.
Some quotes look attractive because reporting, off-campus coverage, SafeSearch enforcement, or support are metered separately. By renewal, the "budget" option costs more than the transparent one — and cutting a module you discover you need mid-year is politically painful. Insist on a quote that covers your must-have list end to end.
If the answer to "what happens at home?" involves a different product, a future roadmap item, or a shrug, the filter only solves half your problem. For any school with take-home devices, the unfiltered evening hours carry the highest risk and the clearest duty-of-care exposure.
A wall of raw logs is not evidence; it is homework. You want a report that states, in category terms, what is blocked for whom, and that an E-Rate reviewer can read without your interpretation. If producing that report during the demo takes more than a minute, imagine producing it under deadline.
When a page is blocked, the filter should say which category triggered it. "It was on the list" satisfies nobody — not the teacher who lost a lesson, not the parent on the phone, not the board member asking about over-blocking. Explainable decisions are what make a filter defensible.
Turn the ten criteria into a one-page grid. Give each criterion a weight from one to three — three for must-haves like CIPA blocking, off-campus enforcement, and audit reporting; two for strong preferences like AI coverage and ease of management; one for nice-to-haves. Then score every candidate one to five per criterion based on what you observed in demos and the pilot, multiply score by weight, and sum.
A worked example: if off-campus enforcement is weighted three and a candidate scores two because coverage requires an extra product, that single row costs it six points against a rival scoring five — a fifteen-point row. The grid makes such gaps impossible to talk around. Share the completed grid with your superintendent or board; a documented, criteria-based selection is also exactly the paper trail you want behind an E-Rate-funded purchase.
If you want a head start, our overview of the best web filter for schools maps each of these criteria to how our own filtering answers them — useful as a reference column while you score the rest of the field.
Bring your fifty-site list and your criteria sheet. We will show you live category lookups, off-campus policy on a take-home device, and the exact report you would hand an auditor — then you score what you saw.