How RateGate works

What we're doing with your bills, how we're using your data, and how plans are compared.

How a bill becomes data

Every bill goes through a processing pipeline before it reaches the chart.

First, we work out who your retailer is by reading the bill itself. Once we know the vendor, we use a parser built for that exact layout to pull out the numbers that matter: your rates, your billing dates, your usage, your ICP, and a few other details. Note that we don't read your personal details (name, street address, payment details, or any other private information) because we don't need it.

Those extracted numbers then get checked against a strict set of rules. If something looks off or doesn't line up, the bill is set aside rather than shown with a guessed-at figure. We'd rather have a gap than a wrong number.

Bills that "pass" get stored, and then we clean up the address (suburb, city, postcode) so bills from the same area can be grouped together properly. From there, duplicate rate entries on the same bill get merged into one clean figure per rate type.

Only bills that make it through every one of those steps show up in comparisons. Nothing on the chart is estimated or filled in to cover a gap. It's either extracted straight from a real bill, or it's not there.

A few things we capture along the way:

• Your tariff type (flat, day-night, or time-of-use)
• Whether your meter is a smart meter, and whether you have a controlled circuit (usually hot water) - this helps us show you better alternative plans
• Your daily charge and per-unit rates, broken down by rate period
• Your service address, so we can compare you to people nearby

All rates are stored GST-inclusive and include the EA levy, since that's the amount you actually pay as a consumer.

How we combine multiple rates into one number

Loads of plans don't have one simple rate. You might pay one rate during the day and a cheaper one at night, or separate peak, shoulder, and off-peak rates if you're on time-of-use. Many plans also show rate changes in the middle of a billing period, even if the plan itself hasn't been changed.

That makes plans hard to compare side by side, so we calculate a blended rate: one cents-per-kWh figure built from your actual usage on that bill. We take how many units you used in each rate period, weight each rate by that amount, average it out, then add the small Electricity Authority levy on top.

Because it's based on your real usage split, not a guess like "half day, half night," the blended rate reflects how you actually use power. We also carry out a similar averaging process for the daily rate (a normally-constant amount you pay per day on your plan) to get a single figure we can factor in there too.

What the benchmark cost on the chart actually shows

Next to the blended rate, each plan on the chart shows a benchmark cost: what that plan would cost over 28 days given your usage. The sum is simple: 28 times the daily charge, plus a set number of units times your blended rate.

If you've uploaded your own bill and we can read your actual usage, every plan on the chart is priced against your usage. This is important to understand: because of the way your blended rate and daily rate interact, some plans will be cheaper at some usage levels than at others. Plans that look cheap at a certain usage level could be quite expensive at others. Using the actual usage from your bill allows us to compare all plans on an equal footing, so the comparison stays fair and relevant.

How we find and score cheaper plans (Insights email)

When you upload a power bill and sign up to receive power price insights, there are a couple of things that have to happen before you get your first email.

First, we wait until there are at least 5 other power plans uploaded in your area to compare your rates to. Because interpreting all these different plans can be tricky, we don't want to jump the gun on sending you an email with one or two comparison plans that are not a good fit for you.

When RateGate suggests a plan that might save you money, it starts by looking at other real bills from your locality. By default that's the suburb or suburb-less town you're in, and if there isn't enough data there yet, we widen the search to your whole city.

Every suggestion also gets a confidence rating: High, Medium, or Low. That's not a prediction of how much you'll save - it's how much reason there is to doubt the comparison is a fair one. We knock a plan down a level for each of these:

• The comparison data is old (e.g. plans over a year old are less accurate than plans that are 6 to 12 months old)
• The suggested plan has a different tariff structure to yours (e.g. flat versus time-of-use plans)
• A time-of-use or day-night plan is suggested but your bill doesn't show you have the right meter for it
• Your controlled load setup doesn't match
• etc.

We try to present you with the highest-confidence, most cost-effective plans in your area that we have in our dataset. If you receive an email that just doesn't ring true with you, definitely let us know.

What we don't do yet

We'd rather be upfront about the gaps than let you assume RateGate sees everything.

• We only know what we've actually extracted from real bills people have uploaded. We don't cover every vendor or every plan on the market, only what we've seen and successfully parsed. Check the banner on the home page for the number of vendors we're indexing.
• If a bill's layout is one we haven't built a parser for yet, it gets set aside rather than guessed at. So a gap usually means "we haven't seen this format yet," not a deliberate omission. We see this alert and will write a parser almost immediately.
• Area comparisons use whichever of town, suburb, city, or postcode we can read from your bill. We don't need your street address, but if we can't see your town/suburb/city in your bill, your recommendations will be mostly meaningless.
• If your bill shows a discount, we record it, but it isn't reflected in the rates or costs we show (this is by design so you're comparing rates like-for-like). So a plan on RateGate might look pricier than what you're actually paying if you've got a personal discount applied.