The single most valuable conversion mechanism on this site is page load performance, and it is bleeding value through extreme lab-measured LCP and TBT on key mobile pages (e.g., /en-us/news LCP 12.2s, TBT 1,140ms; /en-us/sports/nba LCP 9.1s, TBT 1,890ms), which cause visitors to bounce before content appears or becomes interactive.

Stories load in under 2.5s.
Lab LCP on /en-us/news is 12.2s, far above the 2.5s good threshold, causing most mobile visitors to bounce before content appears.

Live scores, schedules, and news for the NBA season.
Lab TBT on /en-us/sports/nba is 1,890ms, blocking main thread interactivity for nearly 2 seconds, causing taps and scrolls to freeze.

Live scores, schedules, and breaking news for Major League Baseball.
Lab TBT on /en-us/sports/mlb is 1,690ms, blocking main thread interactivity and causing unresponsive taps during load.

Performance
Loading…
Lab LCP on /en-us/sports/mlb is 9.3s, far above the 2.5s good threshold, causing most mobile visitors to bounce before content appears.

Live scores now.
Lab LCP on /en-us/sports/nba is 9.1s, far above the 2.5s good threshold, causing most mobile visitors to bounce before content appears.


Lab TBT on /en-us/weather is 1,500ms, blocking main thread interactivity and causing unresponsive taps during load.
Gathered from third-party reviews, forums and video — the site’s own pages are excluded. Every quote was verified word-for-word against the page it links to; anything that could not be verified was discarded rather than paraphrased.
Widespread customer complaints about no real person to talk to and poor customer service.
Many reports of account lockouts, email access...
MSN has been deleting a lot of my comments they don't like and they're violating my 1st amendment right and my right to freedom of speech.
The "guidelines" are ambiguous and if you are blocked, you are not told what "guidelines" you broke and you are never unblocked.
digital content often relies on misleading clickbait titles that do not reflect the actual subject matter.
I got a charge from Microsoft for 19.99 on a different card I've never given them. The charge says recurring.
Each surface starts at 100. The worst issue costs its full penalty — Urgent 25, High 12, Medium 5 — and each further issue costs 30% less than the one before, because the fifth problem on a page does not damage it as much as the first.
| Surface | Urgent | High | Score | Grade |
|---|---|---|---|---|
| Performance | 5 | 4 | 26 | F |
| Trust & Proof | 0 | 1 | 88 | B |
Grades: A 90+, B 80+, C 70+, D 60+, F below 60.
The sum of every finding's individual range. Each row is traffic × conversion × assumed lift band × order value:
| Finding | Lift band | Range |
|---|---|---|
| LCP 12.2s on news | 2.0–8.0% | 0–0/1k |
| TBT 1,890ms on NBA | 2.0–8.0% | 0–0/1k |
| TBT 1,690ms on MLB | 2.0–8.0% | 0–0/1k |
| LCP 9.3s on MLB | 2.0–8.0% | 0–0/1k |
| LCP 9.1s on NBA | 2.0–8.0% | 0–0/1k |
| TBT 1,500ms on weather | 1.0–5.0% | 0–0/1k |
| TBT 1,140ms on news | 1.0–5.0% | 0–0/1k |
| TBT 1,130ms on article | 1.0–5.0% | 0–0/1k |
| LCP 5.4s on article | 1.0–5.0% | 0–0/1k |
| Missing trust signals | 1.0–5.0% | 0–0/1k |
No traffic or order-value figures were supplied, so the total is a per-1,000-visitor rate against an assumed 2% baseline. Supply real figures and this becomes a dollar range.
Caveat that applies to every row: Across 127,000 experiments, 12% produced a statistically significant improvement on the primary metric; a healthy programme win rate is 10-30%. These are prizes if changes work, not forecasts.
A verified finding was measured directly — real-user field data, security headers, accessibility defects counted in the markup — and is reproducible by anyone. Everything else is expert inspection: valuable, but a hypothesis until tested.