How We Compare

Every comparison on VersusVerdict follows the same published rules. This page explains who builds the comparisons, how the data is collected and checked, how winners and scores are calculated, and why the rules are set up the way they are. If something here doesn’t match what you see on a comparison page, please tell us.

Where the data comes from

The specifications come from manufacturers: official spec sheets, product pages, press materials and, for cars, official ratings such as EPA fuel economy estimates. We compile them into one structured dataset, with the same fields and units for every product in a category.

• Checked before publishing. Every product row is checked against the manufacturer’s official spec sheet before it goes live.
• No guesses. If we can’t confirm a value, the page shows “N/A” instead of an estimate. A missing value never counts for or against a product.
• Ratings are labeled. Figures that the manufacturer or an official body rates rather than measures directly, such as battery hours, EPA mpg, 0 to 60 mph times or promised software updates, are labeled as manufacturer ratings or estimates. Your results in daily use can differ.
• No lab tests. We don’t run our own lab tests, benchmarks or test drives. VersusVerdict compares published specifications.

Units are stored in one standard form and converted for display, so a car listed in horsepower and another in kilowatts are compared on the same scale.

Which products we compare, and why

A comparison is only useful if both products are realistic alternatives for the same buyer. So before a pair is published, it has to pass a set of checks we call the segment lock:

1. Same category. Phones are compared with phones, graphics cards with graphics cards.
2. Same segment. Each product belongs to a segment within its category, such as Flagship Smartphone, Ultraportable Laptop, Premium Wireless Earbuds or Compact SUV. Both products must be in the same segment.
3. Compatible body type. The form factor has to match. A compact SUV is compared with another compact SUV or crossover, not with a midsize sedan. A book-style foldable is compared with another foldable, not with a regular phone, and over-ear headphones are compared with over-ear headphones, not with earbuds.
4. Similar price. Using launch or starting prices, the cheaper product must cost at least 60% of the more expensive one. A $600 product is not paired with a $1,200 product in the same segment, because most buyers aren’t choosing between them.
5. Similar age. The two products can be no more than three model years apart.

The family rule. Models from the same product family that are no more than one generation apart are treated as belonging together even if they sit in different segments, for example a flagship phone and the lower-priced version released alongside it, or a console and its Pro edition. This is a choice real buyers make. A direct successor (the next generation of the same model) is always allowed, even if the price gap or the time between releases is larger, because “should I upgrade?” is one of the most common questions.

Cross-class comparisons. Occasionally, people search for a pair that the rules would normally reject, such as two products from neighboring segments. When we publish one of these on purpose, the page says clearly at the top and in the Executive Summary that the two products are in different classes, so you can weigh the result accordingly.

How differences are classified

For every specification both products have, we calculate the difference and decide whether it matters.

• Percentage difference. For numbers, the difference is measured relative to the smaller value: (higher − lower) ÷ lower × 100.
• Threshold. Each specification has its own threshold, the smallest difference that is worth mentioning. It is 5% for smartphone battery capacity and car power, 4% for phone weight and 10% for charging speed, for example. A difference below the threshold is a tie.
• Minor or major. A difference at or above the threshold is a minor difference. A difference of at least three times the threshold is a major difference. With a 5% threshold, a 5,000 mAh battery against a 4,500 mAh battery (11% more) is a minor difference, while 5,000 mAh against 4,000 mAh (25% more) is a major one.
• Yes/no features such as wireless charging or active noise cancelling count as a major difference for high-priority features and a minor one for the rest.
• Ratings on a scale, such as water and dust resistance, are compared by their position on the scale. Two or more steps apart is a major difference, one step is minor.
• Direction. Each specification states whether higher or lower is better. More battery capacity is better; more weight is not. Some values, such as a car’s length or a processor’s name, have no better direction. They are shown for reference but don’t decide a winner.
• Same-brand-only specs. Some numbers can’t be compared across brands because manufacturers count them differently. Graphics card shader units are a clear case: NVIDIA and AMD designs are not equivalent unit for unit. These values are shown side by side, but they only decide a winner when both products come from the same brand.

“At a glance” winners

Each comparison groups specifications into areas, such as Display, Performance, Camera and Battery & Charging for phones, or Engine & Performance, Fuel Economy & Range and Space & Practicality for cars. The “At a glance” panel shows who comes out ahead in each area:

• Every specification a product wins earns points equal to that specification’s weight (at least 0.5), doubled if the difference is major.
• If the product with fewer points still has at least 34% of the area’s points, the area is a trade-off: each product has real strengths there.
• If no specification in the area crosses its threshold, the area is a tie.
• Otherwise, the product with more points wins the area.

An overall lead is only declared when one product wins the more important areas by a clear margin, using the same area weights as the Spec score below. Otherwise the page says the choice depends on your priorities. The “Which One Should You Buy?” section and the Executive Summary read from the same analysis, so they cannot contradict each other.

The Spec score

The Spec score is a number from 0 to 10 that places each product within its segment. It is calculated by formula from the specifications only. It is not an editor rating or a user rating, and it doesn’t account for things a spec sheet can’t show, such as build quality or reliability.

Step 1: normalize each specification within the segment

For every scored specification, the product’s value is placed between the lowest and highest value found among all products in the same segment in our dataset:

• Higher is better: score = (value − segment minimum) ÷ (segment maximum − segment minimum)
• Lower is better: score = (segment maximum − value) ÷ (segment maximum − segment minimum)
• Yes/no features: yes = 1, no = 0 (reversed where “no” is better)
• Ratings on a scale: the same min-max calculation over the scale positions

Each result is between 0 and 1. If every product in the segment has the same value, that specification is left out, since it can’t separate anything. Segment ranges are only used once a segment holds at least three products. When two products from the same family sit in different segments, the combined range of both segments is used.

Step 2: area scores

Area score = 10 × Σ(weight × normalized value) ÷ Σ(weight), counting only the specifications the product has data for. Each specification’s weight reflects how much it matters to buyers in that category; for example, battery capacity has a higher weight than wireless charging speed. An area score is shown only when both products have at least two scored specifications in that area; otherwise the page shows “N/A”.

Step 3: overall Spec score

The overall Spec score is the weighted average of the area scores both products have. It is shown only when at least three area scores are available. The area weights are:

Design, dimensions and price are shown in their own sections and are not part of the Spec score.

A worked example

Take a compact SUV in a segment where cargo space behind the rear seats ranges from 700 to 1,100 liters, maximum cargo space ranges from 1,500 to 2,000 liters, and rear legroom ranges from 900 to 1,000 mm. Every car in the segment seats five, so seating is left out.

• Cargo 1,000 liters (weight 3): (1,000 − 700) ÷ (1,100 − 700) = 0.75
• Maximum cargo 2,000 liters (weight 2): 1.00
• Rear legroom 950 mm (weight 2): (950 − 900) ÷ (1,000 − 900) = 0.50

Space score = 10 × (3 × 0.75 + 2 × 1.00 + 2 × 0.50) ÷ (3 + 2 + 2) = 10 × 5.25 ÷ 7 = 7.5.

If the same car scores 6.0 for Engine & Performance and 8.0 for Fuel Economy & Range, and those are the three areas available for both cars, the overall Spec score is (1 × 6.0 + 1.3 × 8.0 + 1.2 × 7.5) ÷ (1 + 1.3 + 1.2) = 25.4 ÷ 3.5 = 7.3.

Because scores are relative to the segment, a product’s score can shift when a new product with a higher or lower value joins the segment. Scores are rounded to one decimal place.

How automation is used

We want to be clear about this. The text on each comparison page is assembled by our own software from the differences calculated above. It uses a library of sentence templates prepared for this site (drafted with the help of AI writing tools and maintained by the editor), and a sentence only appears when the data difference it describes actually exists. That means every statement on the page is tied to a real value in the table, and a close comparison reads like a close comparison.

No text is generated by an AI service when a page is built or when you view it: the same data always produces the same page. The optional “Summarize with AI” buttons simply open an external AI tool with a link to the page; they don’t add anything to it.

The Executive Summary is the editor’s personal opinion format. It is generated from the same data and the same analysis as the rest of the page, using templates written to match the editor’s voice and priorities, such as value for money, daily use, software support, battery life and weight. It is template-based rather than written individually for each pair, and it follows the same rules as the rest of the page: it never claims hands-on testing, never uses hype, and never disagrees with the calculated result.

What “Last updated” means

The “Last updated” date on a comparison changes only when the content of that page actually changes, for example when a specification is corrected, a price is updated, a new product enters the segment and moves the ranges, or the wording of the page is revised. Rebuilding the site or changing its design does not change the date.

Corrections

Every comparison has a “Report a correction” link that leads to our contact page. Tell us the page, the specification, the correct value and a source, ideally the manufacturer’s own page. We check the report against the source, and if we were wrong, we fix the dataset, which updates every comparison that uses that product.

Who is behind the comparisons

VersusVerdict is built and edited by VersusVerdict Editor, who maintains the dataset, writes the sentence templates and the Executive Summary voice, and handles correction reports. You can read more about the site and how it is funded on the About page, and about its independence in the Disclaimer.