I profoundly believe the Egg reviews are useful, but in qualified ways and with certain concepts in mind.
The first thing I do is to check the 1-star reviews -- more often or thoroughly if they exceed 10%. But even if less, I scan them.
Certain kinds of error descriptions may point to DOA or short life. Of course, if these are below 10%, you should remember that from a quality control perspective this could be a lower number for behavioral reasons: people who have a gripe are more likely to vocalize it in a review, while people who were equitably satisfied with a product may not bother to write a review.
For comparisons, I add up the 4 and 5-star frequency count, and if it exceeds 80% I mark the item as "promising" and write it down or add a new row to a spreadsheet.
For motherboards, I expect to see lower "very-good to excellent" 4 and 5 ratings. You might easily find such ratings -- 50 to 60% -- for top-end motherboards with many features. You can sort these out between "dufus-noobies who didn' know what dey doing," "specialized or ancillary feature not working," "conflicts between other devices," "Doesn't work under [some other] OS," and so on. If you're not planning to use Linux, and other respondents said it worked just fine in Windows, you'd discard the review showing "Doesn't work with Linux."
Now -- these are the mere ratings, sorted by score.
Something that should be most evidently important is the total frequency count, or the number of all customers who write reviews. If the number is large -- 3 or 4 digits -- it's likely to show either a 4-star or 5-star rating. You'd still analyze the 1-stars, 2-stars etc. but the frequency count total is a measure of popularity generated by reviews and test results in conjunct with a price point.
Just guessing for now, I'd think that a 5-star average with a 1,000-customer review-response will help you save money and feel blessed by bargains.