How we score
Every article ends with a score out of 10. This is where it comes from, what it includes, and what it can’t tell you.
Last updated 2 Oct 2026
It’s a research score
We don’t run products in a lab. We read the vendor’s pricing and documentation, independent reviews and benchmarks, and recent changelogs, then judge each product on four criteria for its desk. Every source is linked with the date we read it.
Four criteria per desk
| Desk | Criteria |
|---|---|
| Software & AI | Cost efficiency, Time saved, Output quality, Admin & security |
| Remote Hardware | Performance, Build quality, Value, Setup |
| Systems & Workflows | Time saved, Setup effort, Tool cost, Adoption |
Each criterion scores 0 to 10, to one decimal. The overall score is the plain average, rounded to one decimal. No hidden weighting. Example: the Plugable TBT-UDT3 scored 8.3, 8.0, 9.0 and 8.6, which averages 8.475 and publishes as 8.5.
The cost model
- Time costs $65 an hour, a loaded cost for a salaried team member. Work out yours.
- Hardware is spread over 220 working days a year, so a $300 dock sits next to a $20 subscription.
- Prices are list prices on the day we checked, each with its date, all collected in the Ledger.
- Hours saved come from a stated assumption, shown in the article, and every article lets you rerun it.
Reading a score
| Score | Meaning |
|---|---|
| 9.0 and up | Rare. The clear default for most teams. |
| 8.0 to 8.9 | A strong pick, with the trade-offs named. |
| 7.0 to 7.9 | Right in a specific situation, not as a default. |
| Below 7.0 | We’d look elsewhere first. |
Scores compare products within one article. An 8.5 dock and an 8.5 AI assistant were judged on different criteria.
What a score can’t cover
- How a tool fits your workflow and existing contracts.
- Long-term reliability, unless an independent source has reported on it.
- Price changes after the date in the article.