The six checks
| Check | What to establish | Where to look |
|---|---|---|
| 1. Billing rule | Quota, multiplier scope, reset boundary, what happens at the limit | The written terms, not a chat reply |
| 2. Exit verification | Observed exit addresses, and whether your target services accept them | Your own request plus two IP data sources |
| 3. Peak behaviour | Sustained throughput at the hour you use it, with a no-proxy control | Your own measurement on a weekday evening |
| 4. Terms | Refunds, renewal, fair use, and whether rules can change mid-term | The published terms, saved at purchase |
| 5. Support | Whether a question gets a substantive answer, and how fast | Ask one real question before paying |
| 6. Payment risk | How much a failure would cost you, given the term you are buying | Your own decision, made before purchase |
Why this order
Billing comes first because it determines whether the plan can do your job at all. A generous but mispriced quota — a high multiplier on the route you need — makes every later measurement irrelevant. Exit verification comes second because a plan whose exits your services refuse has no value regardless of speed.
Peak behaviour third, because it is the most common disappointment: everything works in the trial, and the first busy evening is the real test. Terms fourth, because they determine what happens when any of the first three fails.
Support fifth, because it is cheap to test and expensive to discover late. Payment risk last, because it is the conclusion, not an input.
A one-week evaluation
- Day 1: buy the shortest term. Record quota, multiplier scope and reset boundary. Import one profile and verify one real request.
- Day 2: read the observed exit from two sources; test the services you actually use.
- Day 3: send one support question and record the response time and whether the answer addresses the question.
- Day 4–5: measure peak-hour throughput on a weekday evening, three runs, with a no-proxy control.
- Day 6: read the terms as published and note the refund and renewal clauses.
- Day 7: decide. If any of the first three checks failed, do not extend the term.
Signals that deserve weight
- Terms that are written, dated and internally consistent.
- Announcements that are archived rather than edited after the fact.
- Support answers that name a cause and a next step.
- Published measurements that state their conditions, including the hour and the target.
Signals that deserve less weight: unqualified superlatives, “unlimited” without a policy, review lists with affiliate links, and speed claims with no timestamp.
Signals that should stop the purchase
- A requirement to install a root certificate to use the service.
- Payment instructions that arrive only by direct message, or a different payment address because “the site is down”.
- No written terms at all.
- Pressure to buy a long term before any trial.
What to record
- The six checks, each with a date and a one-line result.
- The observed exit addresses and the sources used.
- Peak-hour numbers, all runs, plus the control.
- Terms as saved at purchase.
- The maximum you were willing to lose.
Scope
This page provides an evaluation method. It names no provider, ranks nothing, and publishes no test results for any service. Its only claim is structural: these six checks cover the decisions that a ranking cannot make for you.