Formula
The Billboard Hot 100 ranks songs by weighting the number of sales, streams, and radio airplay audience they receive throughout a tracking week. Billboard gets their numbers from Nielsen, which is highly accurate, but whose information is not freely available to the public.
We can estimate the positions of the next Hot 100 before Billboard publishes it by estimating each song's chart points using a variety of sources and an estimated formula. We can also far more accurately estimate a song's points when Billboard reveals the official numbers (which is always the case for the #1 song).
Billboard splits streams into "paid", "free", and "programmed/passive" streams with a weighting ratio of around 1:0.67:0.5.
Sales / 5
Paid streams / 1,250
Free streams / 1,875
Programmed or UGC streams / 2,500
Airplay / 6,000 (using Billboard's Radio Songs numbers)
However, estimating paid vs free is very difficult regarding on-demand streaming. So to make it easier, you can use:
Sales / 5
On-demand streams / 1,333 (includes "free" streams, so weighted a bit lower)
Official YouTube streams / 1,875
UGC YouTube streams / 2,500
Programmed streams / 2,500
Airplay / 6,000
Depending on the song's streaming sources, the 1,333 on-demand divisor can vary. For songs strong on Apple/Amazon Music, it's lower due to a signficantly higher proportion of paid streams. A value around 1,300 may be suitable for such songs. For songs stronger on Spotify, which includes free streams, it may be 1,350.
Sources
Sales
- hitsdailydouble.com/song_revenue_chart weekly (accurate)
- kworb.net/pop/ iTunes live
On-demand streams
- hitsdailydouble.com/song_revenue_chart weekly (very accurate, but updates the following week)
- hitsdailydouble.com/streaming_songs weekly
- spotifycharts.com/regional/us/daily/latest Spotify
- kworb.net/charts/apple_s/us.html Apple Music
Pop = 50-55% Spotify
Urban = 30-45% Spotify
These numbers are only rough guidelines for new songs - check the ratio between Spotify Weekly Chart and HDD Chart for existing songs for accuracy.
YouTube streams
- charts.youtube.com/charts/TopSongs/us YouTube weekly
- kworb.net/youtube/insights/us_daily.html YouTube past 2 days
- kworb.net/youtube/realtime_anglo.html YouTube live (single video)
This component is made up of two separately weighted factors: official videos (uploaded by the artist or label, e.g. Audio, Video, Lyric), and UGC videos (uploaded by fan/lyric channels, etc).
NOTE: Billboard appears to discard many YouTube streams currently, including official streams. For example, "The Box" had 41.4m streams on the 03/28/2020 chart, with an estimated 27.8m on-demand streams, 20.9m YouTube streams (at least 10.6m of those being official), and 3.5m programmed streams. However, only about 10.1m (48%) YouTube streams appeared to count. Theories include YouTube counting some invalid UGC (e.g. live performances), autoplays not counting at all, stricter criteria for length of streaming, among other factors. Therefore the YouTube chart is not reliable at face value for estimating streams that Nielsen tracks.
My own technique is to multiply estimated official streams by 0.6, and the remaining streams (UGC) by 0.3-0.4. Songs that are more likely to get posted by random lyric channels with high views (where autoplay seems to be a large factor) are closer to 0.3.
"I Love Me" had 14.2m streams according to Billboard, with 10.7m on-demand streams, 5.3m official YouTube streams, negligible UGC streams, and 0.3m programmed streams. This means 3.2m (60%) streams were counted on YouTube, providing credence to the 0.6x "official stream" multiplier.
Programmed streams
- www.nextbigsound.com/ Pandora (requires sign up. Search for the artist and choose "Pandora Radio Spins"). Likely accounts for vast majority of programmed streams.
Radio airplay
- kworb.net/radio/
Airplay has a separate tracking week (Mon-Sun) than Sales & Streaming (Fri-Thu).
Use Monday numbers.
For debuting songs' airplay estimates, be aware of the "8th day effect" where the cumulative audience drops on the 8th day due to first-day radio deals. e.g. if a lead single debuted with 14m audience and was gaining 3m AI every day, then its 8th day effect will be around -11.
NOTE: Billboard's Radio Songs numbers are much smaller than Mediabase numbers.
The multiplier can be anywhere from 0.65–0.80, with an average of around 0.72. The range is unfortunately large, so you need to adjust it based on the Radio Songs chart provided by Billboard, and it can fluctuate for a certain song week to week, but usually by less than 0.01.
e.g. 80,000,000 on Mediabase will be around 58,000,000 (0.725x) by Billboard's Radio Songs numbers. But some songs may be closer to 0.74, and others 0.71, or lower/higher. You need to refer to the previous charts or numbers given out by Billboard. In addition, format tracking changes can influence the multiplier. If a song gains a format on Kworb's chart, its multiplier will likely be lower from that point on.
Here are some debuting songs' multipliers vs. Mediabase:
"Stupid Love": 23.7m (Billboard) vs. 33.60m (Monday Mediabase) = 0.705
"I Love Me": 15.9m (Billboard) vs. 22.49m (Monday Mediabase) = 0.707 (gained a format on that day)
Example
Billboard Hot 100: 03/28/2020
"The Box"
Sales = 8.0k (via Billboard)
Streams = 41.4m (via Billboard)
Radio airplay = 65.0m (via Billboard)
On-demand streams = 27.8m (via Song Revenue Chart)
Official YouTube streams = 6.5m (0.6x Official)
UGC YouTube streams = 3.6m (0.35x Remaining)
Programmed streams = 3.5m (via Pandora Radio Spins)
8,000/5 + 27,800,000/1,300 + 6,500,000/1,875 + 3,600,000/2,500 + 3,500,000/2,500 + 65,000,000/6,000
40,100 points
No comments:
Post a Comment