Type a shoe name, like Pegasus, Clifton or Kayano.

Will next year's version of your favorite running shoe be as good?

Brands revise their shoes every year or two. We examine whether shoes maintain their quality.

99

HOKA Clifton 9

99 / 100 Best price $150
View shoe
99

ASICS Novablast 5

99 / 100 Best price $117
View shoe
98

HOKA Speedgoat 6

98 / 100 Best price $120
View shoe
98

Nike Vaporfly 4

98 / 100 Best price $146
View shoe
98

Saucony Endorphin Speed 5

98 / 100 Best price $122
View shoe
98

Nike Dragonfly 2

98 / 100 Best price $95
View shoe
98

Saucony Triumph 22

98 / 100 Best price $105
View shoe
97

Brooks Ghost 16

97 / 100 Best price $110
View shoe
97

Nike Vomero 17

97 / 100 Best price $133
View shoe
97

HOKA Bondi 9

97 / 100 Best price $140
View shoe
94

Brooks Glycerin 22

94 / 100 Best price $90
View shoe
94

New Balance FuelCell Rebel v5

94 / 100 Best price $137
View shoe

Each product line's top-rated shoe for sale, from the lines we hold five or more shoes of.

Shoe brands constantly revise their shoes, including even those with a die-hard following. Sometimes, runners are rewarded, other times they get burned. Three examples show why. First, consider the Nike Vaporfly, which debuted with the Vaporfly 4%, improved substantially with its first revision, the Next%, and has never dropped since.

Nike Vaporfly ratings by version

Each version’s Better Shoes Rating, oldest first.

Line chart of five Nike Vaporfly versions from the Zoom Vaporfly 4% to the Vaporfly 4, every one rating 99 or above.
Figure 1. Every version has rated at the top of our scale, and no revision has moved the line by more than two points. The Vaporfly 4 has only three reviews so far, so its rating is the least settled of the five.

Compare this to the Nike Pegasus product line, which has continually frustrated runners with its inconsistency. For example, the Pegasus 36 was a top-rated shoe, with a score of 94. The next year, they released the Pegasus 37, which rated a dismal 64. Since then, the shoe's quality has recovered into the middling range.

Nike Pegasus ratings by version

Each version’s Better Shoes Rating, oldest first.

Line chart of twelve Nike Pegasus versions moving between the mid-60s and the mid-90s, with the Pegasus 36 highest and the 37 that followed it far below.
Figure 2. The Pegasus swings across about forty points of the scale from one version to the next. The 36 rates 95 and the 37 that followed it rates 56.

Lastly, consider the HOKA Clifton series. The original version was widely popular and well-reviewed, scoring a 98. Scores lagged a little, but the Clifton 4 proved to be a disaster compared to its reputation, scoring only a 70. Since then, the shoe has rated consistently well, with the Clifton 9 being one of the most popular and highly rated shoes in our entire database.

HOKA Clifton ratings by version

Each version’s Better Shoes Rating, oldest first.

Line chart of eleven HOKA Clifton versions opening at 98, dropping to 70 at the Clifton 4, and recovering to 99 at the Clifton 9.
Figure 3. The Clifton opens at 99, falls to 64 at the Clifton 4, and recovers to 99 at the Clifton 9. The Clifton 11 has only 3 reviews so far, so its rating is the least settled of the eleven.

These three case studies demonstrate what most runners have learned through experience: quality varies within a shoe's product line.

In this article, we test that intuition against the data. We hope to help shoppers have a better understanding of how much quality persists through most shoes' product lines so they can use this knowledge, alongside our ratings database, to improve their shoe-buying decisions. The product line, we will show, is the best quality indicator available for a new shoe whose quality is not yet known.

Consecutive versions of shoes rate alike

Across 77 pairs of consecutive versions where both versions carry at least 25 reviews, the correlation between one version's average rating and the next is 0.54. A version that rates in the top half of the catalog is followed by another top-half version 69% of the time, where chance alone would rate in the top half only half the time.

Relationship between star ratings of versions one, two, and three generations apart

Raw review averages, before any modeling.

Three scatter panels of a version’s average star rating against an earlier version’s, one to three generations apart, with the upward trend line flattening as the gap widens.
Figure 4. The upward slope means versions are correlated: good versions tend to follow good ones, and bad versions tend to follow bad ones. The relationship fades as versions separate, falling from r = 0.54 one generation apart to 0.20 three apart. Because these are plain review averages, the pattern comes from the shoes, and never from our model. Based upon version pairs where both shoes have at least 25 reviews.

Figure 4 also demonstrates what happens as versions drift apart. The correlation falls to 0.28 two generations apart and 0.20 three apart. A runner who liked the previous version has better evidence about the next revision than those generations apart.

Runners who rebought within a product line preferred the new version

Our cleanest evidence comes from 806 reviews where the previous shoe was an earlier version of the same product line and the runner rated which one was better. 54% of reviewers rated the new version better, while 34% rated the old one better, a net of +20 points in the new revision's favor.

Buying a shoe within the same product line leaves runners as or more satisfied than the prior version 66% of the time.

Verdicts from runners who rebought within a product line

Each dot represents the share of reviewers giving that verdict, from reviews where the previous shoe was an earlier version of the same product line.

Dot plot of three verdicts: 54% rated the new version better, 13% had no preference, and 34% rated the old version better.
Figure 5. 54% of reviewers rated the new version better and 34% rated the old one better, a net of 20 points in the new revision's favor. Based upon 806 reviews where the runner had run in an earlier version of the same product line.

Next year's version usually rates higher

Across those same 77 pairs, the new version rates higher than the one it replaced 57% of the time, and falls more than ten points from its previous score only 18% of the time. The usual outcome for a new shoe is a modest improvement in rating. Only about one revision in seven is a real step down in measured quality.

Last year's version is also a sound buy

Runners switching shoes report liking the new one most of the time, whatever it is, so raw switching numbers overstate how often a change succeeds. Figure 6 shows head-to-head verdicts grouped by how much ratings typically jump when switching from one version of a shoe in a product line to any other version.

Head-to-head verdicts by how far the switch jumped

From comparisons where one runner reports on both shoes, grouped by the release-year gap between them.

Dumbbell plot across six switch-distance groups, each showing a larger share rating the new shoe better than worse, with the largest margin for jumps of four or more years.
Figure 6. Every group nets positive, because runners mostly favor a shoe they just switched to. Read against that baseline, runners who switched to an older shoe did as well as runners who switched to anything newer.

This suggests that buying last year's version at a discount is often a sound move, and our best deals pages rank discounted shoes by the same score shown on every shoe page.

Conclusion

Our data show that shoes within a product line typically rate similarly and in fact improve more often than not. Moreover, our data also show that buyers who shop within a product line are overwhelmingly satisfied with their new shoes. For this reason, we are comfortable recommending to runners that if the previous version worked and the new version rates well, buying it again is often a good decision. If possible, the best move is not to be an early adopter of a shoe; wait and see how the reviews look. If they're positive in comparison, this provides even more basis to shop within the product line. Lastly, if the previous version worked and the new version rates poorly, a discounted purchase of the old revision makes a lot of sense.

Data are accurate as of September 8, 2026. Numbers may differ slightly as new reviews come in. We periodically update these articles to stay in sync.