Wiki Model 3 Order Tracking Spreadsheet

[Troy]

@Sneakerbuddy,
We will find out the accuracy of the production and delivery estimates in a few days. Here are my estimates:

• 27,443 Model 3 production in Q2 2018 (Expected accuracy: 95%)
• 23,756 Model S+X production in Q2 2018 (Expected accuracy: 92%)

• 15,814 Model 3 deliveries in Q2 2018 (Expected accuracy: 85%)
• 11,269 Model S deliveries in Q2 2018 (Expected accuracy: 85%)
• 9,674 Model X deliveries in Q2 2018 (Expected accuracy: 85%)

• 14,512 Model 3 deliveries in the US in Q2 2018 (Expected accuracy: 85%)
• 1,302 Model 3 deliveries in Canada in Q2 2018 (Expected accuracy: 85%)

The actual numbers for the first 5 items will be released by Tesla on 3 or 4 July. I will calculate the accuracy and post it here. I don't think Tesla will release the last two items but I added those just in case they decide to release them too.

 

[Sneakerbuddy]

@Sneakerbuddy,
We will find out the accuracy of the production and delivery estimates in a few days. Here are my estimates:

• 27,443 Model 3 production in Q2 2018 (Expected accuracy: 95%)
• 23,756 Model S+X production in Q2 2018 (Expected accuracy: 92%)

• 15,814 Model 3 deliveries in Q2 2018 (Expected accuracy: 85%)
• 11,269 Model S deliveries in Q2 2018 (Expected accuracy: 85%)
• 9,674 Model X deliveries in Q2 2018 (Expected accuracy: 85%)

• 14,512 Model 3 deliveries in the US in Q2 2018 (Expected accuracy: 85%)
• 1,302 Model 3 deliveries in Canada in Q2 2018 (Expected accuracy: 85%)

The actual numbers for the first 5 items will be released by Tesla on 3 or 4 July. I will calculate the accuracy and post it here. I don't think Tesla will release the last two items but I added those just in case they decide to release them too.

Those numbers would indeed be brutally low for Tesla shareholders. I dont think even the bears have the delivery figures that low. 20% decline in MS/X deliveries. M3 deliveries just over 1,000 per week...

This all makes sense to me except for the M3 production vs. delivery numbers. Do you really think they have over 12,000 M3 in inventory/storage somewhere? That would be at retail $660M in badly needed cash tied up. Plus, wouldn’t the forums be filled with expected deliveries in early July? Wouldn’t the delivery centers be overrun with inventory ready to go? My personal guess is deliveries a little higher at 18K and production closer to 22k.

 

[Troy]

Here are the results of the 5 estimates I posted yesterday. The accuracy was as expected. Generally speaking, I'm happy with my estimates.

Both of my production estimates are based on VIN calculations. Besides the Model 3 spreadsheet, I also manage the Model S/X Order Tracker spreadsheet. In this spreadsheet, the S+X production estimation process is completely automated. Therefore hopefully it will display similarly accurate estimates for all quarters from now on. Tesla only publishes S+X production numbers but they publish the delivery numbers separately.

The interesting thing here is that Tesla didn't hold back on S/X deliveries. Therefore it will be interesting to find out what happened with the 200K situation. As you might know, I estimate that they have already crossed the 200K in Q2. If that's correct. $7,500 will last until for deliveries until the end of Q3 2018. Then it will be $3,750 in Q4 2018 and Q1 2019 and $1,875 in Q2 and Q3 2019.

If you are waiting for the standard range and if you end up getting $1,875 instead of $7,500, you would lose 7500-1875= $5,625. That means the larger battery would only cost $9,000-$5,625= $3,375 more than waiting for the standard range.

 

[Sneakerbuddy]

@Sneakerbuddy,
We will find out the accuracy of the production and delivery estimates in a few days. Here are my estimates:

• 27,443 Model 3 production in Q2 2018 (Expected accuracy: 95%)
• 23,756 Model S+X production in Q2 2018 (Expected accuracy: 92%)

• 15,814 Model 3 deliveries in Q2 2018 (Expected accuracy: 85%)
• 11,269 Model S deliveries in Q2 2018 (Expected accuracy: 85%)
• 9,674 Model X deliveries in Q2 2018 (Expected accuracy: 85%)

• 14,512 Model 3 deliveries in the US in Q2 2018 (Expected accuracy: 85%)
• 1,302 Model 3 deliveries in Canada in Q2 2018 (Expected accuracy: 85%)

The actual numbers for the first 5 items will be released by Tesla on 3 or 4 July. I will calculate the accuracy and post it here. I don't think Tesla will release the last two items but I added those just in case they decide to release them too.

Bravo Troy! Your numbers were pretty much spot on. You pretty much nailed it on S/X deliveries and production. You were a little low on M3 deliveries which I think is due to the data degrading as they move away from the reservation list. Biggest miss was M3 production which was

Here are the results of the 5 estimates I posted yesterday. The accuracy was as expected. Generally speaking, I'm happy with my estimates.

Both of my production estimates are based on VIN calculations. Besides the Model 3 spreadsheet, I also manage the Model S/X Order Tracker spreadsheet. In this spreadsheet, the S+X production estimation process is completely automated. Therefore hopefully it will display similarly accurate estimates for all quarters from now on. Tesla only publishes S+X production numbers but they publish the delivery numbers separately.

The interesting thing here is that Tesla didn't hold back on S/X deliveries. Therefore it will be interesting to find out what happened with the 200K situation. As you might know, I estimate that they have already crossed the 200K in Q2. If that's correct. $7,500 will last until for deliveries until the end of Q3 2018. Then it will be $3,750 in Q4 2018 and Q1 2019 and $1,875 in Q2 and Q3 2019.

If you are waiting for the standard range and if you end up getting $1,875 instead of $7,500, you would lose 7500-1875= $5,625. That means the larger battery would only cost $9,000-$5,625= $3,375 more than waiting for the standard range.

Troy - Very impressive work with the estimates. You pretty much nailed it. A few questions:
1) How do the spreadsheets deal with the concept that people can now order a M3 with hardly a reservation and therefore may never report to you? This would seem to affect sample size issue and may cause you to under estimate future orders/deliveries.
2) How do you account for the huge difference between M3 production and delivery? I can see some vehicles on trains or trucks but even if you knock out the 5,000 in the last week, that still over 6,000 cars in transit and most of those could be sold in California (if not Fremont pickup.)

 

[Troy]

@Sneakerbuddy,

1) How do the spreadsheets deal with the concept that people can now order a M3 with hardly a reservation and therefore may never report to you? This would seem to affect sample size issue and may cause you to underestimate future orders/deliveries.

There are two aspects to your question:
Sample size: It's possible to use different methods to calculate deliveries without relying on the sample size. Next quarter, I will try that to improve my delivery estimate accuracy. Ideally, I want to automate all S/X/3 production and delivery estimates and hit 95%+ accuracy with production and 90%+ with delivery estimates.
Survey entries: Lots of people still submit new entries. See the screenshot here that shows the entries submitted in the last 24 hours. There are 40 entries. That means 1200/month new entries. Personally, I would be happy with 500/month. The reason total entries have stayed around 5K is that I keep deleting outdated entries with missing data to improve the quality.

After Tesla opened up the configurator to everyone, I have removed the reservation date, time, time zone columns from the spreadsheet and the form and I made the configuration date a required answer. Ideally, all entries should have a configuration date, VIN, VIN date and delivery date. By the way, the Model S/X don't have a reservation queue either but people still enter their data to the Model S/X Order Tracker spreadsheet.

2) How do you account for the huge difference between M3 production and delivery? I can see some vehicles on trains or trucks but even if you knock out the 5,000 in the last week, that still over 6,000 cars in transit and most of those could be sold in California (if not Fremont pickup.)

It could be cars in transit as Tesla says. In the Google form, there is a question about the month of manufacture. If Tesla was stockpiling cars somewhere, we would find out that after those people take delivery and report the month of manufacture.

 

[MissyMcBee]

@Troy I Noticed that your spreadsheet states that a vin number will be automatically calculated after 74 days but some of us are close to that so does that mean it will show that we have one even if we don’t? I imagine you did that because some folks may not go in and update the spreadsheet?

 

[Troy]

@MissyMcBee, this is the message users see in column S on the All Entries tab here:

Because no VIN date was entered, a date was automatically calculated 74 days after configuration. If you have a VIN date, click 'Edit' next to your username to add it. If you don't have a VIN date, click 'Edit' next to your username and submit the form with or without any changes. The new timestamp will postpone the calculated VIN date for another 10 days. Thanks

In other words, if you simply open the form and click submit again, you will avoid the auto-calculated VIN date for another 10 days. The idea is to understand whether the user is still around. Only 2% of buyers wait 74 days or longer for a VIN.

 

[bgbolton]

@MissyMcBee, this is the message users see in column S on the All Entries tab here:


In other words, if you simply open the form and click submit even without making any changes, you will avoid the auto-calculated VIN date for another 10 days. The idea is to understand whether the user is still around. Only 2 of people wait 74 days or longer for a VIN.

Nah only 2 people bothered to submit form again. The number waiting 74 days or longer is 3 at the least

 

[Troy]

I want to talk about a trend change in VIN assignments. In the last 7 days, 50 people reported that they have received their VIN within two days of configuring. 50 entries represent about 700 people. I don't have an explanation for this but it's an interesting change I've noticed. You can find the live version of this table on the VINs tab here.

By the way, I'm making some changes to improve the quality of the data. It might look like the number of entries stays the same at a little over 5,000 but that's because I'm removing outdated and incomplete entries.

If you are waiting for the Standard range option, open the All Entries tab here, find your username and look at column M (Trim level) and make sure it says Model 3 SR or Model 3 SR AWD. If you are simply waiting because you are not ready to order yet, click Edit next to your username, enter 1st Jan 2017 into the configuration box and re-submit the form. The reason for 1st Jan 2017 is that new entries without a configuration date are not accepted anymore.

This will reduce abandoned entries in the future. However, making the configuration date a required answer means people who are waiting for SR can't resubmit the form because they don't have a configuration date. Therefore I wrote a formula that ignores 1st Jan 2017. When you submit the form, it will refresh the last update date. These days I'm deleting 151-day old entries that have a configuration date but no VIN or delivery date. I will slowly move towards 100 days while at the same time keeping the total number of entries over 5K. Also, I'm deleting 100 days or older entries where the person selected Model 3 LR but didn't enter a configuration date yet.

 

[daver6640]

@MissyMcBee, this is the message users see in column S on the All Entries tab here:


In other words, if you simply open the form and click submit again, you will avoid the auto-calculated VIN date for another 10 days. The idea is to understand whether the user is still around. Only 2% of buyers wait 74 days or longer for a VIN.

@MissyMcBee, this is the message users see in column S on the All Entries tab here:

@MissyMcBee, this is the message users see in column S on the All Entries tab here:


In other words, if you simply open the form and click submit again, you will avoid the auto-calculated VIN date for another 10 days. The idea is to understand whether the user is still around. Only 2% of buyers wait 74 days or longer for a VIN.

Oh I see the form fills in a VIN date after 74 days. Thats why on July 1st there were reports of tons of VINs assigned but not really? It was auto calculated?

The 74 day rule sounded like it was applicable in the past but shouldn't be moved out a bit further. I mean to take into account Tesla's deliberate attempt to avoid hitting 200K delivered car(s) in Q2. Some of us April configs were delayed intentionally and are still waiting for a VIN.

Thanks Troy!

 

[Troy]

Oh I see the form fills in a VIN date after 74 days. That's why on July 1st there were reports of tons of VINs assigned but not really? It was auto calculated?

Originally I wanted to count the auto-assigned VIN dates too when making production calculations but then I changed my mind because the invite dates are not spread across evenly. There was a huge invites batch on 18 April. That means we would see a spike 74 days after 18 April. Therefore I decided not to use it where the count matters. In the end, the production estimate worked pretty well.

All tables in the VINs tab and the table on top of VIN Queue 1 exclude auto assigned VIN dates. These tables are focused on the count. The bottom table in VIN Queue-1 and the tables in VIN Queue 2 include it. These tables provide a visual representation of the current situation. I think the quality of the VIN Queue 2 tab right now is pretty good.

 

[daver6640]

Originally I wanted to count the auto-assigned VIN dates too when making production calculations but then I changed my mind because the invite dates are not spread across evenly. There was a huge invites batch on 18 April. That means we would see a spike 74 days after 18 April. Therefore I decided not to use it where the count matters. In the end, the production estimate worked pretty well.

All tables in the VINs tab and the table on top of VIN Queue 1 exclude auto assigned VIN dates. The bottom table in VIN Queue-1 and the tables in VIN Queue 2 include it. I think the quality of the VIN Queue 2 tab right now is pretty good.

Thanks for explaining that. I will take a look at it.

 

[thx1139]

Day 1 online reservation.
Ordered Sunday 7/1/2018.
Was able to edit remainder of Sunday and Monday. Now the option to edit is now gone. Does that mean anything? We are ordering the AWD version.

 

[njsrikar]

@Troy do you account for people who edited their configurations? For example, I configured an AWD on 5/20 but changed color to obsidian black on 28th June. Should I update the tracker sheet to 6/28? There are confusing reports whether this change in configuration would affect the delivery date (initially it had said 12-16 weeks, now it says Sept-Nov 2018).

 

[Troy]

@njsrikar, I don't think you need to update your configuration date in the Google form when you edit your configuration.

 

[ras2645]

@Troy Hi, I'm new here. I configured my AWD with19 inch rims, midnight metallic grey, auto pilot and full self driving on 5/24. I was a day one line reservation. When I configured I was told 2 to 3 months. Are you able to estimate when I may get my car and what place I may be in for AWD production/ delivery?

 

[Troy]

@ras2645, I'm using 7.2% for the sample rate these days. 60 people have reported ordering a Performance version and 334 people a non-P AWD in the US and Canada. That means, 60/7.2%= 833 P versions and then 4,639 non-P AWD will enter production. They will start with the Performance version and then you are on top of the non-Performance AWD list. We don't know when Performance production will start. All I can see is that they acquired 2285 AWD VINs on 28 June from NHTSA and then 436 more AWD VINs on 5 July.

The difference between these two dates is exactly 7 days. That could mean 436/week production rate at the beginning and then maybe they increase it to 1000/week. This is only for the GA4 (General Assembly 4 production line aka tent line) that will make the Performance/AWD versions. I think they are using a software that calculates how many NHTSA VINs to acquire 3 weeks in advance.

To make this more interesting, here are my predictions:

• Performance production will start on 19 July 2018
• Non-Performance AWD production will start on 3 Aug 2018

 

[ras2645]

I really appreciate the estimates, your work is very impressive! I live in New York, if I'm lucky I may be looking at middle to end of September delivery which would be about 4 months from configure which seems plausible because its a month later than my estimate at configure time.

 

[Sneakerbuddy]

Troy - Any comment on the apparent lack of deliveries being delivered and scheduled in July? Tesla reported 11,000 cars in transit as of 6/30 which is a huge number - almost twice their previous largest month of deliveries. Combine this with what should be increased production in July and I would expect they could deliver up to 20,000 M3 in July alone (which would be on the order of 3x prior months volume.) But, spreadsheets are showing delivery volume much closer to May. Any insights or guesses?

 

[kwintone]

Thank you for putting all of this together, very helpful!

MS owner in NorCal
6/29 reservation
6/30 configured M3P, DBM, EAP + FSD
6/30 MVPA