How the scores work

Phone score, 1 to 10

For sites with a valid lab run, the phone score is the rounded sum of three parts. A higher number means more room for a phone visit to improve:

• 0.4 × (100 − Lighthouse performance) / 10
• 0.2 × the page weight in megabytes × 1.25, capped at 10
• 0.4 × an evidence rating from 1 to 10

The evidence rating is a judgment from the linked complaints, articles, and regulator actions. It is not a measured number. Sites whose lab run was blocked use that evidence rating alone. For register.com, the score uses the lowest Lighthouse performance of the three pages measured.

Reach and Tranco

Every site has a rank from Tranco list Y83KG (opens in a new tab), built from Chrome UX Report, Farsight, Majestic, Cloudflare Radar, and Umbrella for September 4 through October 3, 2026. Tranco ranks registrable domains, so the California DMV rank is all of ca.gov.

Where a company or a government published a user or visit count, that figure is included with its source. Nothing was estimated, and no Similarweb numbers were used. Counts that say “Tranco rank only” have no sourced user total. Some company figures are non-unique or group-wide: the Weather Company 330 million is across its properties, Forbes 109 million uniques are global, and myWalgreens 124 million is loyalty members, not web users.

The reach tier, from 1 to 10, comes only from the Tranco rank:

• 250 or better: 10
• 1,000 or better: 9
• 2,500 or better: 8
• 5,000 or better: 7
• 10,000 or better: 6
• 20,000 or better: 5
• Otherwise: 4

Rebuild difficulty

Low, multiplier 3

The core jobs are public, or there’s an open data source.

Medium, multiplier 2

A public front plus a login area, or public pages with content-rights limits.

High, multiplier 1

The core jobs sit behind login, payments, regulated or health data, or aggressive anti-bot terms.

Opportunity score

Phone score × reach tier × doability. The maximum is 300. The research list is sorted by that score after the pinned first place. It’s a rough way to prioritize, not a precise measure. Weather.com and Fandom tie at 216. register.com scores 112 and is still first because the work has started.

On this site the ballot is ordered by votes. A tie keeps that research rank. The build queue is a separate plan: register.com, GoDaddy, Weather.com, Fandom, then USPS.

Lab measurements

Lighthouse 13.5.0, run locally in headless Chrome with a mobile form factor: an emulated Moto G Power, simulated slow 4G, and a 4× CPU throttle. One run per URL, the homepage unless noted, one at a time on October 4, 2026, around 04:50–05:09 CEST. StubHub’s clean run errored, so the blocked attempt from 03:51 CEST is the one recorded. Earlier parallel runs, made under CPU contention, were thrown out wherever a clean run exists.

The Google PageSpeed Insights API wasn’t usable. The unauthenticated daily quota returned HTTP 429.

Caveats

• Single runs move around, total blocking time most of all.
• The test used a US datacenter IP, so geo, consent, and ad decisions may differ from a real phone.
• Only homepages were tested, plus two Network Solutions pages for register.com. Heavy logged-in flows aren’t in the lab numbers.
• Bot walls blocked StubHub (a DataDome “access temporarily restricted” page) and American Airlines (a challenge that ended in HTTP 403).
• Spirit, UPS, and Realtor.com failed outright and were left off the wall.
• Complaint links were collected by web search on October 4, 2026. Several are Reddit or forum threads, which are anecdotal. Not every link was opened by hand, so spot-check before quoting them.

What stayed off the ballot

Forbes, Daily Mail, and CNN rank high (4th, 5th, and 7th in the research list). They’re copyrighted, ad-funded articles. A cleaner page would be a reader for those articles, which sits outside the phone tasks this queue is for. They’re shown on the wall, and they aren’t planned. Revisit them if an on-device reader is ever the right fit.

register.com is now a front end for Network Solutions. Search, login, help, and DNS redirect there, so one rebuild covers both brands. The public pass is done. The logged-in pass is still open.

Bot Constitution

Three bots work on this site. None of them outranks the owner.

• Development builds and fixes, and can’t approve its own work.
• Testing looks for bugs and reports them, and can’t change app code.
• Security looks for weaknesses and reports findings, and can’t change app code. It gets the final check before anything ships.

The owner has the final say.