Methodology
How Comparia works
Every comparison on Comparia follows the same process: identify what actually differentiates the options, score each against weighted criteria, surface the trade-offs honestly, and update the recommendation when the facts change. This page is the full record of how we do it and where the boundaries are.
Last reviewed: April 2026
Our principles
Four commitments that shape every comparison. These are non-negotiable — if you ever spot us breaking one, tell us and we'll fix it publicly.
We don't accept payment from the products we compare
No vendor pays to appear on Comparia. No vendor pays for a higher score. When we link to a retailer we note whether it's an affiliate link — that's our only revenue relationship, and it never changes which product wins a comparison.
We score against user priorities, not generic averages
The reason most comparison sites feel useless is that they average everyone's needs. Comparia asks what you actually care about, weights the criteria accordingly and produces the answer for your situation — not the median shopper.
We show our working
Every comparison shows the criteria, the weights, the scores for each option, and where our confidence is low. If you disagree with a score you can override it and see the winner update. The recommendation is never a black box.
We say when we don't know
AI-generated scores are flagged with confidence levels — high, medium or low. Low-confidence scores are visible as a small amber dot on the comparison table. Where we can't test first-hand, we say so.
How a comparison is built
Five steps, every time. The order matters: get the criteria right before scoring — the wrong criteria produce a confident but useless ranking.
- 01
Identify the domain
Travel, property, jobs, products, services, financial, or open-ended. Each has a different shortlist of what actually differentiates real options — travel turns on flight time and vibe, property on location and catchment, jobs on comp and growth. Using the right shortlist for the domain is the single biggest quality driver.
- 02
Suggest criteria that differentiate
A criterion earns its place by moving the ranking. 'Reliable' on a car comparison is too vague. 'MOT failure rate in years 4-7' differentiates. We prefer criteria that have measurable, observable answers over generic virtues.
- 03
Weight by your stated priorities
Each criterion is rated critical (50% weight), important (25%) or nice-to-have (10%). Weights are normalised across the set so the scores always sum to a comparable total. You can change any weight and the winner recalculates live.
- 04
Score each option 1-10 per criterion
Scores are generated from published data, expert reviews and first-hand testing where possible. Each score carries a confidence level. You can override any score you disagree with and the recommendation updates accordingly.
- 05
Surface trade-offs, not just a winner
Every Comparia recommendation includes the winner's genuine limitations and a 'choose X instead if…' for each runner-up. A one-sentence winner is rarely the truth.
Confidence levels
Not every score is equally reliable. Each score carries one of three confidence labels, surfaced in the comparison table as a small coloured dot.
High confidence
The fact is widely documented, independently verified or personally tested. Published specs, official pricing, measurable benchmarks.
Medium confidence
Based on general reputation, category patterns or indirect signals. Reasonable but not independently verified for this specific option.
Low confidence
Inferred from the option name, limited public data or a category default. We surface low-confidence scores so you can weigh them appropriately or override them with your own judgement.
How often we update
A comparison is only useful if it reflects current reality. Every page shows a visible "Last reviewed" date. The underlying commitments:
Major model or product release
Re-score and re-write within 7 days. Example: a new flagship AI model or a major Stripe fee change.
Minor release or feature update
Update the affected rows in the comparison table and any changed prices. Recommendation reviewed for flip.
Routine refresh
Every comparison reviewed at least quarterly whether or not anything has shifted. The 'Last reviewed' date on each page reflects the most recent check.
Market shift
If the competitive set changes materially (new entrant, provider exits the UK market), we re-evaluate whether the existing 3-way or 5-way comparison still represents the genuine shortlist.
Corrections and feedback
If a score, fact or pricing claim is wrong, let us know. We publish corrections openly and date them. Getting it right matters more than being first to publish.
Editorial stance
Comparia does not accept payment from the products we compare. Some outbound links to retailers or providers are affiliate links which may earn us a commission at no cost to you. Read our full affiliate disclosure for the details.