Data Notes

I treat KEXP's own published lists as the record. Anywhere I've departed from them, it's noted below.

Every KEXP list behind this project has gone through some amount of manual handling: reconciling what aired against what KEXP eventually published, and making judgment calls where the data itself is ambiguous or inconsistent. Rather than bury a version of this note at the bottom of every post, it lives here once.

Data as of: KEXP’s Annual lists through 2025, and MusicBrainz metadata as saved between December 2025 and August 2026. I’ll refresh the metadata from time to time to pick up changes. Published posts stay as they were, and anything that changes gets its own update.

How the list data was captured

The Annual lists from 2007 on were captured live as albums were announced on air, then reconciled against KEXP's own published version once it went up. The 2001–2006 Annual lists predate that approach: KEXP staff graciously provided these to me directly, which I then cross-checked against historical playlists.

How the listener votes worked

KEXP’s Annual countdown lists are built entirely on listener votes, though the voting format has evolved significantly over time. Over the years, the station adjusted how many albums each person could select as well as whether voters chose from pre-populated lists or typed in custom entries. This history was compiled using archived KEXP voting pages, though a few years remain unaccounted for (and are blank) where historical records were unavailable.

In short: five picks a year through 2007, mostly typed in; ten picks a year from 2008 on, with two exceptions (5 in 2010, 20 in 2012) and a few years I couldn't verify.

The formats, ballot sizes, and picks for 2003 to 2016, 2018, 2022, and 2023 come from archived copies of KEXP's voting pages. The number of picks and the format for 2020 to 2025 were gathered from my own voting history. For 2019, I only have the voting page's title ("Top Ten"); the 2018 and 2020 votes were both 10 picks, so I've listed 2019 the same way. KEXP's 2014 voting page describes its list as "all the records we added to KEXP last year," so in these years the ballot was probably built from the music KEXP had added to its library, plus whatever listeners wrote in. The years marked "Not found" are ones where no archived voting page turned up.

Because the number of picks and the format changed several times, some of what looks like a trend in the lists could partly reflect a change in how the votes were collected. I'll point that out where it matters.

Albums on two lists. Four albums appear on two different Annual lists, each time at a notably better rank the second year: Elbow's Cast of Thousands (2003 #71, 2004 #39), Vampire Weekend's self-titled debut (2007 #22, 2008 #2), The Head and the Heart's self-titled debut (2010 #16, 2011 #4), and Of Monsters and Men's My Head Is an Animal (2011 #16, 2012 #10). They aren't errors; each was eligible in two voting years. Two were on the ballot both years: The Head and the Heart (2010 and 2011) and My Head Is an Animal (2011 and 2012), which KEXP notes was released in Iceland in 2011 and in the US in 2012. The other two came from write-ins: Cast of Thousands (2003 and 2004 were both write-in years), and Vampire Weekend's debut (typed in with suggestions in 2007, on the ballot in 2008). I keep each as KEXP listed it.

Artist and album names

Artist and album names are pulled directly from MusicBrainz by artist ID, so whatever MusicBrainz has is what shows up here, including whatever normalization they've already done on their end (their own "and" vs. "&" conventions, for instance). I'm not making manual naming calls of my own.

MusicBrainz is crowd-sourced

Beyond names and release dates, the details I include for artists and albums also come from MusicBrainz, a free music database built and curated by volunteers around the world. On the positive side, this resource provides an invaluable wealth of data. On the negative side, because it's entirely dependent on users' contributions, there can be gaps in the data or inaccuracies. I work from a saved copy rather than the live site, and where I have spotted a real error that matters I note it on this page. If you notice something wrong, the best fix is at the source: anyone can register for a free MusicBrainz account and correct it, and a richer MusicBrainz makes the data behind this blog better too.

Collaborations and shared credits

MusicBrainz isn't consistent about collaborations. Sometimes a person and the band they front are two separate artists, and sometimes a collaboration is credited to one person. I follow MusicBrainz: a different artist ID is a different artist here. Mark Lanegan and Mark Lanegan Band, for example, are counted as two artists, which suits them, since the solo records and the band's records are different projects. The flip side is that a few people show up both as a solo artist and as a band.

How I compare time periods

Most posts compare an early stretch of the lists with a recent one, for example 2001–2010 against 2016–2025. I do this because any single year is small: most lists have 91 albums, which can mean only a handful of artists in a given category, and one year can jump around for reasons that have nothing to do with a trend. Averaging over several lists smooths that out, so what's left is more likely to be a real change. The cost is that the start and end years are choices, and they aren't always the same from one chart or tile to the next. Here is how I make them.

• The default is ten lists against ten. Where I can, I compare the first ten lists (2001–2010) with the most recent ten (2016–2025). Equal lengths keep the comparison fair. The five lists in between (2011–2015) aren't in either period. Ten is a round number I picked, and nothing special happens in 2010 or 2016.
• Sometimes I use only the last five lists. When a change is recent or sharp, a ten-year window can blur it, so a few comparisons use 2021–2025 instead. Five lists is a smaller sample and a noisier average. The tile captions and chart notes say which years they use. The one place the two windows sit side by side is the solo albums-per-year tiles on the Who's on the list? post: they use 2021–2025 on purpose, because solo men's bump ended around 2020, and a 2016–2025 window would have blurred it.
• Anything about new or returning artists starts in 2011. The lists begin in 2001, so in the early years everyone looks new, including artists who had been making records for years. Comparisons about first-timers, comebacks, and retention therefore start in 2011, after a decade of history, and use five-list windows throughout (2011–2015 against 2021–2025). For the retention chart I follow each artist for five years after they first appear, since five years catches the large majority of comebacks. That is also why the last group counted is the one first listed in 2020, the latest with five full years of follow-up through 2025.
• A middle period, when two endpoints would mislead. Solo men averaged about 15 albums a year in 2001–2010 and again in 2021–2025, but about 20 in between. Comparing only the ends would call that no change. Where that happens I add a third period or say so in the text.
• Smoothing. The share-by-year chart draws a three-year centered average, and hovering shows the actual numbers for each year.

To check that these choices aren't doing the work, I reran several headline comparisons with the periods shifted by a few years in either direction. The band share of entries fell from 73% to 76% down to 58% to 60% in every version, the typical artist went from 6 to 10 or 11 years into their career, and solo women's share rose from about 6% or 7% to somewhere between 17% and 23%, depending on the window. The direction and rough size of those held up. Solo men did not: depending on the windows, their share ended anywhere from about half a point lower to about four points higher than it started. That is why I describe their line as roughly level with a bump in the middle rather than as a trend.

Two more things affect any comparison across years. Listeners voted for five, ten, or twenty albums depending on the year (see How the listener votes worked above), and the lists aren't always the same length: most have 91 albums, but 2012 has 120 and 2024 and 2025 have 100. Both can nudge a single year, which is one more reason I lean on several lists at a time. When one post uses different years from another, it says which, in the tile caption or under the chart.

Corrections: what I believe to be real KEXP errors

None that I've found so far.

Curation calls: my own judgment on ambiguous entries

• Best of 2005: The Postal Service's We Will Become Silhouettes (#29) is a 4-track single.
• Best of 2008: The Decemberists' Always the Bridesmaid: A Single Series (#73) is a single.

Reissues, special editions, and release dates

Lists are matched to MusicBrainz by album, so a few entries carry a release date or position that isn’t quite what you’d expect. Everything is counted as KEXP listed it; these are the cases where that takes a little explaining.

• A reissue that shares the original’s entry. Radiohead’s OKNOTOK 1997 2017 made the 2017 list at #43. MusicBrainz groups it with OK Computer, so it appears here as OK Computer: an album from 1997 with a 2017 list appearance.
• A reissue that is its own album. Jeff Buckley’s Live at Sin-é was first released in 1993 as an EP. The expanded edition that made the 2003 list (#46) is a separate album in MusicBrainz, so it’s counted as its own album, dated 2003.
• Anniversary packages and live albums. Radiohead’s KID A MNESIA (2021) and I Might Be Wrong: Live Recordings (2001) are counted as separate albums in Radiohead’s total, because that’s how they were listed.
• Remixes. The xx’s Basic Space (Remixes) (#43 in 2010) counts as a list appearance, but as a remix release it isn’t part of the album-order counts.
• A wrong date. MusicBrainz dates Dear John Letters’ Stories of Our Lives (#73 in 2003) as 2008. AllMusic, Discogs, and a physical copy all say 2003, so I corrected the date to 2003.

One more thing worth being upfront about: a lot of this was gathered in real time: capturing albums as they were announced on air, then checking against KEXP's own published list where I could. There were times a list was posted out of order and corrected later, or a DJ mentioned on-air that something had played out of sequence. It's possible some of what's here isn't quite right. If anyone from KEXP, or anyone with better information, comes across this and wants to set the record straight, I'd genuinely welcome it, so please contact me.