Billboard Hot 100 Chart Formula and Point Calculations

[moonlite]

Pandora shares their #s, so programmed streams can be estimated from them. Just search for the artist on www.nextbigsound.com and go to the "releases" section. Ex. Bieber

And HDD's SRC is more accurate than their regular on-demand chart and gives very accurate sales as well for most of the top songs. But it doesn't update till a week after the tracking week is over

[FindingC7osure]

Thank you for creating this thread! It's very helpful to see everything outlined so clearly and in one place.

Nov 8, 2019 4:19:30 GMT -5 gs said:

I think this component may explain why songs like "Senorita" and "bad guy" are charting higher than they seemingly should be. For instance, "Truth Hurts" had more points for airplay, paid streams, YouTube and sales than Senorita, yet, was 1 position below it on the Hot 100. I may be mistaken though, and something else is at play.

I have been doing predictions for the last few weeks as well and have been struggling to understand how these songs consistently have more points than it seems they should. Some predictors have theorized that certain songs have additional YouTube User-Generated Content (UGC) that isn't being reflected by YouTube's weekly numbers (especially "Old Town Road" and "bad guy"), but I'm not sure if this is what has been making up the entire difference. Would you be willing to share your point breakdowns of "Señorita" and "bad guy" for this past week?

I think it's interesting to see that Simon is weighing free and passive streams equally. I remember reading that Billboard weighs passive streams the least of the three, so I'm curious to see if a minor adjustment in this weighing could make the formula slightly more accurate.

Thanks again!

[gs]

Thanks for the info! I've updated the original post.

FindingC7osure: Simon said he doesn't calculate "UGC" and just uses the official chart (since I believe YouTube takes all of that into account already). Looking at the Pandora source (I've updated the OP), "Truth Hurts" had virtually no passive streams (<140,000) while "Senorita" had over 3,750,000. This could explain the difference.

[gs]

I wonder if running tons of different formula tests could allow us to "hone in" on the most accurate one.

Billboard Hot 100: 11/09/19 (Simon Falk Formula; chart position difference in parentheses)

1. Lose You To Love Me (303) (0)
2. Someone You Loved (296) (0)
3. Circles (281) (0)
4. Truth Hurts (266) (+1)
5. Senorita (254) (-1)
6. Good As Hell (233) (0)
7. Follow God (232) (0)
8. No Guidance (228) (0)
9. bad guy (205) (+1) (This has HUGE Pandora spins; the 1.33x passive multiplier adjustment is probably too much)
10. Panini (198) (-1)
11. 10,000 Hours (198) (+1)
12. Memories (193) (-1)
13. Beautiful People (181) (+2)
14. HIGHEST IN THE ROOM (181) (0)
15. Trampoline (169) (+1)
16. Ransom (169) (-3)
17. Bandit (159) (+1)
18. Closed On Sunday (149) (-1)
19. Selah (147) (0)

Logically, this may mean:
• Airplay must be weighed less, and/or Streaming weighed more (explaining "Ransom" above "Trampoline"/"Beautiful People")
• Sales must be weighed more (explains "Follow God" below "Good As Hell" and "Memories" above "10,000 Hours", **despite** the above point)
• Passive is likely weighed less and split into its own category (partially explains "bad guy" below "Panini", but runs in contrary to this next point below)

However, "Truth Hurts" vs. "Senorita" is very tricky:

• +7,000 sales
• +1,272,000 paid streams
• +840,000 YouTube streams
• -4,733,000 passive streams
• +8,150,000 airplay

I can't adjust the formula in a way that makes "Truth Hurts" ever fall below "Senorita" without it becoming ridiculous. Every metric except "passive" would need to be weighted much less. This must mean some other metric is missing that is not being counted, or one of the data points is inaccurate.

I don't think there's a sensible formula that places all the songs accurately using the available data... so perhaps "UGC" may be the secret ingredient after all? That would also explain "Old Town Road" only having ~105 points in the chart I made above, yet it's at #21. Although Simon gave it 150+ points, and he doesn't count UGC, so I don't know :s

[moonlite]

Nov 9, 2019 1:39:10 GMT -5 gs said:

I wonder if running tons of different formula tests could allow us to "hone in" on the most accurate one.

Billboard Hot 100: 11/09/19 (Simon Falk Formula; chart position difference in parentheses)

1. Lose You To Love Me (303) (0)
2. Someone You Loved (296) (0)
3. Circles (281) (0)
4. Truth Hurts (266) (+1)
5. Senorita (254) (-1)
6. Good As Hell (233) (0)
7. Follow God (232) (0)
8. No Guidance (228) (0)
9. bad guy (205) (+1) (This has HUGE Pandora spins; the 1.33x passive multiplier adjustment is probably too much)
10. Panini (198) (-1)
11. 10,000 Hours (198) (+1)
12. Memories (193) (-1)
13. Beautiful People (181) (+2)
14. HIGHEST IN THE ROOM (181) (0)
15. Trampoline (169) (+1)
16. Ransom (169) (-3)
17. Bandit (159) (+1)
18. Closed On Sunday (149) (-1)
19. Selah (147) (0)

Logically, this may mean:
• Airplay must be weighed less, and/or Streaming weighed more (explaining "Ransom" above "Trampoline"/"Beautiful People")
• Sales must be weighed more (explains "Follow God" below "Good As Hell" and "Memories" above "10,000 Hours", **despite** the above point)
• Passive is likely weighed less and split into its own category (partially explains "bad guy" below "Panini", but runs in contrary to this next point below)

However, "Truth Hurts" vs. "Senorita" is very tricky:

• +7,000 sales
• +1,272,000 paid streams
• +840,000 YouTube streams
• -4,733,000 passive streams
• +8,150,000 airplay

I can't adjust the formula in a way that makes "Truth Hurts" ever fall below "Senorita" without it becoming ridiculous. Every metric except "passive" would need to be weighted much less. This must mean some other metric is missing that is not being counted, or one of the data points is inaccurate.

I don't think there's a sensible formula that places all the songs accurately using the available data... so perhaps "UGC" may be the secret ingredient after all? That would also explain "Old Town Road" only having ~105 points in the chart I made above, yet it's at #21. Although Simon gave it 150+ points, and he doesn't count UGC, so I don't know :s

No it's definitely UGC views. Senorita is doing 8-9 mil rn on YT with UGC included and OTR is doing 15-16 mil with UGC. Ransom and Bad Guy are also songs rn with significantly high UGC YT views

[gs]

Nov 9, 2019 1:57:20 GMT -5 moonlite said:

Nov 9, 2019 1:39:10 GMT -5 gs said:

No it's definitely UGC views. Senorita is doing 8-9 mil rn on YT with UGC included and OTR is doing 15-16 mil with UGC. Ransom and Bad Guy are also songs rn with significantly high UGC YT views

Makes more sense, is there a source for the UGC views though?

This is what Simon said to me in DM, so maybe he does include them and I misinterpreted it:

> I just use youtube. idk what they are using
> dont really have a direct source for passive (or radio streams as billboard calls them) streams. I always use my streaming estimate from the previous week as a starting point for the new estimate

Edit: I have added the concept of UGC streams to the "Free Streams" section since it seems most likely to be the explanation for the points gap. It seems you need to use the current H100 position to estimate how many UGC streams it probably has by boosting its YT streams until the points roughly matches its position, since there's no source(?). Then, if it drops around -10% on the YT Chart, it probably dropped -10% in UGC streams as well, allowing for a decent estimation of the new week's streams.

[Lukas]

Question: Are you sure that the formula is paid streams/160 and free streams/260? I believe that the following article explains how the streaming tiers are weighted in terms of the streaming component:

www.billboard.com/articles/news/8427967/billboard-changes-streaming-weighting-hot-100-billboard-200

[gs]

Nov 10, 2019 19:08:21 GMT -5 Lukas said:

Question: Are you sure that the formula is paid streams/160 and free streams/260? I believe that the following article explains how the streaming tiers are weighted in terms of the streaming component:

www.billboard.com/articles/news/8427967/billboard-changes-streaming-weighting-hot-100-billboard-200

Updated the post to split them, I agree Simon's formula seems simplified. The ratio I found is 1:0.67:0.5 for paid:free:passive, but I can't read that article since it's paywalled. Let me know if that's the real ratio and/or the divisors seem accurate :)

That said it doesn't make much of a difference for most songs, like 0-2 points using 240k+320k vs. 260k grouped.

[Lukas]

Nov 11, 2019 3:14:14 GMT -5 gs said:

Nov 10, 2019 19:08:21 GMT -5 Lukas said:

Question: Are you sure that the formula is paid streams/160 and free streams/260? I believe that the following article explains how the streaming tiers are weighted in terms of the streaming component:

www.billboard.com/articles/news/8427967/billboard-changes-streaming-weighting-hot-100-billboard-200

Updated the post to split them, I agree Simon's formula seems simplified. The ratio I found is 1:0.67:0.5 for paid:free:passive, but I can't read that article since it's paywalled. Let me know if that's the real ratio and/or the divisors seem accurate :)

That said it doesn't make much of a difference for most songs, like 0-2 points using 240k+320k vs. 260k grouped.

Yes, that is the real ratio for paid:free:passive. And I also agree that it doesn't really make much of a difference in the calculations, but it would still be useful knowing and putting up there.

[Lukas]

Been about a month, but I felt like these were notable:

"bdsradio.com/pdweb/pdweb.dll/login (may be more accurate and calibrated to the Radio Songs numbers - you need to add up each format's AI)"

The problem with this site is that, for most formats, you can only see a song's AI on a specific chart if it is actually on the main chart. You can't see if it is in bubbling under or recurrent. And from what I know, AI IS calculated for those charts; songs like Sucker and DWAS stayed roughly the same on the chart when they went recurrent on the Mainstream Top 40.

Also, you should add digitalsalesdata.com/ to digital sales since it estimates exact iTunes sales. Definitely more accurate than Kworb which only gives you ratios to the #1 song.

Also can a mod stick this message to the top of the board - this thread contains insanely useful information.

[Lukas]

With the new formula change we need a revamp of this thread. I’ll make a draft covering the new points here tomorrow.

[Lukas]

Jan 13, 2020 16:48:47 GMT -5 Lukas said:

With the new formula change we need a revamp of this thread. I’ll make a draft covering the new points here tomorrow.

So I lied and this apparently didn't happen.

However, I think that this post on a different forum from the OP explains it incredibly well.

[gs]

I've updated the post to account for the new YouTube streaming and Song Revenue Chart sales oddities since the rule change and added some more details.

I haven't changed the formula yet as my estimates after the fact have all worked out (although very close). Sales could be 5.5 rather than 6 now, but not certain. If BB announces a formula change, I'll update it, or if it requires it.

[gs]

[85la]

^ That formula leaked by 6ix9ine cannot be labeled as "real" or "official" though, as it cannot be verified as actually coming from Billboard, and they addressed this in their article responding to 6ix9ine's allegations. Yeah, they could be lying, but we don't know for sure. If most of the rankings match up when the numbers are plugged in, however, it might be a fairly accurate estimate.

From their article:

"The Hot 100 forecast 6ix9ine referenced in his Instagram video: The chart forecast referenced was not created nor provided by Billboard to the industry. Those with access to sales, streaming and radio data from various sources often create their own chart models and update them at their own frequency. Billboard does not distribute any Hot 100 ranking forecast to labels, management or artists."

www.billboard.com/articles/business/chart-beat/9380654/billboard-hot-100-number-one-calculations-6ix9ine

[mikerivera]

Leave it to 6ix9ine to try and snitch on Billboard

[kanimal]

May 20, 2020 12:34:02 GMT -5 85la said:

^ That formula leaked by 6ix9ine cannot be labeled as "real" or "official" though, as it cannot be verified as actually coming from Billboard, and they addressed this in their article responding to 6ix9ine's allegations. Yeah, they could be lying, but we don't know for sure. If most of the rankings match up when the numbers are plugged in, however, it might be a fairly accurate estimate.

From their article:

"The Hot 100 forecast 6ix9ine referenced in his Instagram video: The chart forecast referenced was not created nor provided by Billboard to the industry. Those with access to sales, streaming and radio data from various sources often create their own chart models and update them at their own frequency. Billboard does not distribute any Hot 100 ranking forecast to labels, management or artists."

www.billboard.com/articles/business/chart-beat/9380654/billboard-hot-100-number-one-calculations-6ix9ine

Variety reported the same ratio, citing a "label source."

variety.com/2020/music/news/6ix9ine-ariana-grande-billboard-chart-feud-1234609142/

[mikerivera]

I’m fairly certain the only reason Billboard doesn’t share the formula is so that they don’t have to be held accountable for whatever results they publish. Which is kinda dumb, because we’ve had multiple people at Pulse who have been able to find the formula for years now. Hell, I was even able to figure out the formula for the pre-streaming digital era. It was sales/11, radio/11,000.

[gs]

The new formula is likely the actual one since labels are using it, even though it's not "confirmed". It makes things align better than the other one at the very least anyway.

In the old points it's:

Sales / 6
Paid / 1,500
Free / 2,250
Programmed / 3,000
Radio / 7,200 <-- main difference

So we adjusted sales too high incorrectly earlier in the year (/5.5), the only thing we had wrong was radio weighted too high (/7,000) originally.

This helps explain why Pain 1993 was above Circles at #7, Stupid Love above ROXANNE at #5, etc.

[gs]

Updated with the new radio recalibration multiplier

[degen]

May 20, 2020 15:36:47 GMT -5 mikerivera said:

I’m fairly certain the only reason Billboard doesn’t share the formula is so that they don’t have to be held accountable for whatever results they publish. Which is kinda dumb, because we’ve had multiple people at Pulse who have been able to find the formula for years now. Hell, I was even able to figure out the formula for the pre-streaming digital era. It was sales/11, radio/11,000.

Lol I’m pretty sure they don’t share their formula so they can prevent labels and managers from manipulating the charts during the tracking week. Even though it’s no secret that industry insiders pretty much know the data 95% of the time, it’s still wise for Billboard to not officially share it so they don’t get accused of helping certain labels/artists.

[gs]

For all of us wondering why YouTube views started barely counting in January:

Not only did they remove non-song UGC, they also removed guest views from counting.

[kacpi]

So TikTok videos don't count as streams anymore?

[somelikeitwhen]

Aug 23, 2020 8:14:43 GMT -5 kacpi said:

So TikTok videos don't count as streams anymore?

They never did

EDIT: unless you mean compilations on Youtube which haven't counted since January

[gracies]

May 20, 2020 19:33:44 GMT -5 gs said:

The new formula is likely the actual one since labels are using it, even though it's not "confirmed". It makes things align better than the other one at the very least anyway.

In the old points it's:

Sales / 6
Paid / 1,500
Free / 2,250
Programmed / 3,000
Radio / 7,200 <-- main difference

So we adjusted sales too high incorrectly earlier in the year (/5.5), the only thing we had wrong was radio weighted too high (/7,000) originally.

This helps explain why Pain 1993 was above Circles at #7, Stupid Love above ROXANNE at #5, etc.

How much does 160K units contribute to the charts? Doesn't that ensure a top 10 debut?