Methodology: how NetSaver checks facts and how the tools work

How we check facts

Every guide was checked against primary sources on 8 November 2026: Ofcom, GOV.UK and Building Digital UK, the Competition and Markets Authority, Ofgem, UK planning legislation, and the providers' and manufacturers' own websites and contract documents. Where only a trade-press report was available — for example, some Starlink prices seen by ISPreview's readers — the guide names the source and says it is reported rather than published by the provider.

We keep these labels distinct in the guides:

• Confirmed — stated by the provider, regulator or legislation.
• Operator's own claim — a figure from a company about itself, such as coverage or average speed. We say so.
• Reported — from trade press or reader reports, not the provider.
• Estimate or rule of thumb — our own working, shown with its assumptions.
• Conflicting or unknown — we say so, rather than guess. Not finding something doesn't mean the answer is "no".

Prices change often and many depend on your address or a promotion. Where we give a price, we give its date and context. Always confirm the price for your exact address with the provider before you order.

What we have not done

• We don't run an address checker. NetSaver doesn't hold licensed address-level availability data, so we don't guess what reaches your home from a postcode. We link to the official and network checkers instead.
• We don't measure your connection. The 4G/5G review only works with figures you type in; there is no built-in speed test or signal reading.
• We haven't tested equipment or connections ourselves. There are no NetSaver speed panels, router reviews, UPS runtime tests or failover tests behind these guides.
• We don't run a deals feed, price tracker or ranking of offers, and we don't take commission.

Household needs estimate

The estimate adds up planning figures for each activity happening at the same moment, then doubles the total for Wi-Fi losses and short peaks:

Results are rounded up to the next whole megabit. To convert, 1 megabyte per second (MB/s) is 8 megabits per second (Mbps). This is a planning estimate, not a guarantee: services change their recommendations, and latency and stability matter as much as speed for calls and games.

4G/5G test review

The review compares the speed-test results you type in with the household need you enter. It gives a verdict only when there are at least three results including one between 8pm and 10pm, because evening is when masts are busiest.

• Unlikely to be enough — the evening result and the middle (median) result are both below your download need.
• Borderline — at least one result fell short of your download or upload need, latency went above 100ms or varied by more than 50ms between tests, or download speed varied by more than half between the slowest and fastest test.
• Promising — every result met your need, with steady latency.

These thresholds are our own rules of thumb, not industry standards, and we deliberately don't set a "good signal" level. The review can't tell whether a shortfall comes from weak signal or a busy mast, so it never recommends buying an antenna on the strength of the numbers alone.

Monthly data estimate

Up to 3GB an hour for HD video and 7GB an hour for 4K (Netflix's figures for its highest settings), and about 3.1GB an hour for an HD group video call — derived from Zoom's 3.0Mbps down plus 3.8Mbps up, where 1Mbps for an hour is 0.45GB. Daily totals are multiplied by 30.4 days for an average month, plus any other downloads you enter.

Cost calculator

For each option, the total over the period you choose (24 months by default) is:

• the regular monthly price for every month, except promotional months at the promotional price;
• plus each price rise written into the contract, added to the monthly price from the month it starts (rises stack: two £3.50 rises make £7 a month more after the second);
• plus one-off costs (set-up, installation, delivery);
• plus equipment, either bought once or rented every month — never both;
• plus electricity, if you give an average power draw and a unit rate: watts × 24 hours × 30.4 days × months ÷ 1,000 gives kWh, multiplied by your rate in pence.

All sums are done in whole pence and the electricity cost is rounded to the nearest penny. The calculator doesn't include early exit fees, credits, cashback or price changes that aren't written into a contract. A rise dated before the starting month is rejected (it should already be in the monthly price); a rise after the end of the period is ignored and flagged.

The examples you can load are labelled as illustrative and dated: 5G home broadband at £17 a month rising by £3.50 each April (Three and Vodafone, 8 November 2026); Starlink Residential 100Mbps at its £40 standard price with a reported £10 kit rental and £19 shipping; and 26.32p per kWh, Ofgem's price cap average for electricity from October to December 2026. In the backup version, the optional outage scenario multiplies your own estimate of days lost a year by your cost per day, pro rata for the period. It is your scenario, not a promised saving.

Decision trees

The two decision trees follow the logic in the guides, reviewed against the sources. Each question has an "I don't know" answer: where a missing fact makes an honest suggestion impossible, the tree tells you what to check instead. The backup tree asks first whether anyone relies on the line for telecare, alarms or emergency calls, and adds a caution to every result if so. Results are starting points based on your answers: they can't confirm that a service is available at your address or guarantee that a backup survives every failure. A text version of each tree sits underneath it.

Corrections and updates

These guides were written and checked once, on the dates shown. We don't promise a fixed update schedule. If something has changed, the provider's or regulator's own page is the authority.