RJ: R2 order conversion rate is "much higher" than internal expectations

[TTedP]

RJ: Hey world, the R2 matches the efficiency of the MY. World: Oh, ok that's awesome! Lets see some 3rd party testing about that. Yikes.

RJ: Hey world, the R2 pre-order conversion rate is higher than we expected here at Rivian. World: Oh, ok. Hmmmm.

So true. The internet is so cynical.

Like my German mom when I would come home with an A- “yeah but you didn’t get an A+”

Sponsored

 

[tivoboy]

Do we actually know what the conversion rates are what was expected or reported? Did they actually say some number that we don’t have a reference point for?

Historically a 10% conversion rate is considered good. And achievable target as well. They might have pushed an early achievable conversion target for some defined population as 15%. Higher than a normal possible target.
And yet they MAY have achieved a 20% conversion for at least the first target 15-25k expected orders.

 

[WoodChips]

I never even thought about that. Good point!!!

There has been some major road construction in my area and I saw someone in a Tesla ride through a new intersection with their hands behind their head. At first I was really caught off guard but as the days went on I continued to think about it. Every Tesla that drives through that area is uploading new data. Who knows how many times it takes to learn a new road pattern. A single person might go through it twice a day but there might be 50 Teslas all uploading their information...

The Rivian delivery vans I see all over the place have a lot of availability data, it's going to be a huge advantage to help improve their hands free driving.

 

[tivoboy]

This is very misleading. The 15% is derived from the entire sales history of Tesla. FSD has only become good this year. Take rates for FSD in 2026 are near 50%. The price for used Tesla vehicles is drastically different for hardware 3 versus hardware 4 simply because the latest FSD needs HW4.

Well, to be fair I doubt the actual “take rate” for FSD in 2026 is 50%.. the “usage or trial rate” is most certainly POSSIBLY that amount and the number of ppl who start a trial and forget to cancel and get charge for a month before cancelling is also going to push that number up, POSSIBLY to 50%, but thats more a marketing and finance scam than anything else - and by design I might add.. there is significant literature and analysis on this in the business community - not specific to Tesla..

The prior to ~ 2020 “take rate” for purchased FSD was somewhere between 4-5%… and that was only derived several years loter. So it landed somewhere in the 100-185K range.. It was North America only at the time.

Then, they had the “one time only” FSD transfers - that actually happened at least FIVE Times, basically it was a rolling window from 2022-2025..and I think even into early 2026.. they really did want ppl to transfer while buying a new car.

Cars crashed and weren’t repurchased or repaired, the FSD died with the vehicle.

And then the specified “unsupervised” vs “supervised” language changed in March 2019.. so anyone prior could say FSD should be level 4, hands free, eyes off, (Robotaxi wasn’t really a term of art yet).

So after ALL of that, cars with true FSD are now probably about 40-50K WW. Everything now is just variable subscription model, and that can certainly vary based on use case, efficacy and price.

 

[windblowlc]

So true. The internet is so cynical.

Like my German mom when I would come home with an A- “yeah but you didn’t get an A+”

Asian parents - "We expect you to get straight AAA+"

 

[rbdavis808]

Historically a 10% conversion rate is considered good.

I suppose my own completely uninformed spitballing would put the R2 conversion rate right about there. My understanding is that the last released figure for reservation count was a bit above 200,000 in early '25. My reservation of 9/25 would then perhaps make me about #250,000? I got an invite estimate in June for 2/27, and I read no reports of people getting estimates for any date later than that. So, does that mean they are expecting the order queue of 250,000+ to be cleared around the time they have built R2 #30,000? That would be in the ballpark of 10%. So, if they were expecting 10% and are now calculating 11%, RJ's not lying.

Yes, it would be nice to see the actual numbers to satisfy my own idle curiousity, and no, I can't imagine we ever will as Rivian won't want to chance a stupid internet debate on what it all means that might damage their marketing, for which I could hardly blame them.

 

[tivoboy]

I suppose my own completely uninformed spitballing would put the R2 conversion rate right about there. My understanding is that the last released figure for reservation count was a bit above 200,000 in early '25. My reservation of 9/25 would then perhaps make me about #250,000? I got an invite estimate in June for 2/27, and I read no reports of people getting estimates for any date later than that. So, does that mean they are expecting the order queue of 250,000+ to be cleared around the time they have built R2 #30,000? That would be in the ballpark of 10%. So, if they were expecting 10% and are now calculating 11%, RJ's not lying.

Yes, it would be nice to see the actual numbers to satisfy my own idle curiousity, and no, I can't imagine we ever will as Rivian won't want to chance a stupid internet debate on what it all means that might damage their marketing, for which I could hardly blame them.

That’s some good baseline datapoints, but I think the run rate needs updating.

I have projected at least 1750 weekly production by EOY 27.. I might actually up that to 2000 by start of 2027…

A 2/27 could be anything from week 1 to week 4, but still lets say thats nine weeks of 2027 and a run rate of ~ 1900 so ~ 17100 by then in 2027, add the most likely at least 25K this year (previous Rivian estimate that I think is a low ball) and we’re pretty close to 45K..

From the original 200K reservations, that we could at least 15%-20% depending on total reservations.

So I think 10% might have been the bogey (standard target, nobody gets fired for buying IBM) and they thought MAYBE 15%, and now it looks that THAT is doable, so 15-20% conversions (this is now a converted intent at least, not purchases which should be much higher for this short current period)

 

[rbdavis808]

I think the biggest unknown is the total reservation count as of June when the invite estimates went out. With all the flurries of imminent launch excitement through this spring, could that count now be over 300,000? I think there's a decent amount of elasticity in all these guesstimates.

 

[tivoboy]

I think the biggest unknown is the total reservation count as of June when the invite estimates went out. With all the flurries of imminent launch excitement through this spring, could that count now be over 300,000? I think there's a decent amount of elasticity in all these guesstimates.

And I also don’t’ really know is WE know what they mean by “conversion”

It might be “reservations” that have now clicked on their interest and purchase intent, but nothing more..

Or, possibly more likely it’s ppl who had reservations, indicated interest in Launch Edition, were GIVEN an order email for a launch edition AND put down the additional $500 order fee and/or took delivery.. that might be the metric they are tracking for “conversion”

The total reservation number is always a hard one to qualify and evaluate after actual launch..

Are there ppl who just put in a reservation NOW because they know they won’t get anything for a year, so why not and see what comes, it’s only $100 and refundable?

OR is it ppl who are putting in a reservation NOW because they are now actually very interested in the car?

That’s very hard to tell and define and quantify.

 

[mdpa]

There has been some major road construction in my area and I saw someone in a Tesla ride through a new intersection with their hands behind their head. At first I was really caught off guard but as the days went on I continued to think about it. Every Tesla that drives through that area is uploading new data. Who knows how many times it takes to learn a new road pattern. A single person might go through it twice a day but there might be 50 Teslas all uploading their information...

The Rivian delivery vans I see all over the place have a lot of availability data, it's going to be a huge advantage to help improve their hands free driving.

Yes, to some extent this is true. Also for about 90 days Tesla has added a popup asking for confirmation on "why" when FSD is disabled. A little annoying, but I figure it is temporary and helping them fast track improvements.

However, my understanding is the main difference is the "vision" based system. Unlike Super Cruise or Blue Cruise that uses mapped roads, it is using AI to interpret the environment, which is why it will work in parking lots or even on a temporary road, construction, or even a field.

Is it as reliable as when it compares it to mapped data? No, but pretty darn good. Probably drops from 99.8% to 88%. In fact, I can't remember having to correct a construction zone. It will respond to hand gestures or the turning of a stop/slow sign without any intervention.

 

[tivoboy]

​

Yes, to some extent this is true. Also for about 90 days Tesla has added a popup asking for confirmation on "why" when FSD is disabled. A little annoying, but I figure it is temporary and helping them fast track improvements.​

​

However, my understanding is the main difference is the "vision" based system. Unlike Super Cruise or Blue Cruise that uses mapped roads, it is using AI to interpret the environment, which is why it will work in parking lots or even on a temporary road, construction, or even a field.​

​

Is it as reliable as when it compares it to mapped data? No, but pretty darn good. Probably drops from 99.8% to 88%. In fact, I can't remember having to correct a construction zone. It will respond to hand gestures or the turning of a stop/slow sign without any intervention.​

​

​

In my 2018 TM3 that pop up has been there for ooohh about 6 years!​

 

[SANZC02]

I’m curious if it was more that they were expecting delayed orders with people waiting for RAP1/Lidar versions and are finding many people converting when they get the time to order e-mail.

 

[emroch]

Yes, to some extent this is true. Also for about 90 days Tesla has added a popup asking for confirmation on "why" when FSD is disabled. A little annoying, but I figure it is temporary and helping them fast track improvements.

However, my understanding is the main difference is the "vision" based system. Unlike Super Cruise or Blue Cruise that uses mapped roads, it is using AI to interpret the environment, which is why it will work in parking lots or even on a temporary road, construction, or even a field.

Is it as reliable as when it compares it to mapped data? No, but pretty darn good. Probably drops from 99.8% to 88%. In fact, I can't remember having to correct a construction zone. It will respond to hand gestures or the turning of a stop/slow sign without any intervention.

One of the interviews with the Rivian folks mentioned that they are intentionally focusing on "SD maps" instead of "HD maps" for exactly this reason. SD is effectively GPS/street view quality while HD is precise LiDAR mapping.
HD goes stale very quickly and reliance on it means any change like construction can completely throw off any autonomy. If you can get by with SD then you can be much more robust and react to changes in real time.

 

[kurtlikevonnegut]

I suppose my own completely uninformed spitballing would put the R2 conversion rate right about there. My understanding is that the last released figure for reservation count was a bit above 200,000 in early '25. My reservation of 9/25 would then perhaps make me about #250,000? I got an invite estimate in June for 2/27, and I read no reports of people getting estimates for any date later than that. So, does that mean they are expecting the order queue of 250,000+ to be cleared around the time they have built R2 #30,000? That would be in the ballpark of 10%. So, if they were expecting 10% and are now calculating 11%, RJ's not lying.

Yes, it would be nice to see the actual numbers to satisfy my own idle curiousity, and no, I can't imagine we ever will as Rivian won't want to chance a stupid internet debate on what it all means that might damage their marketing, for which I could hardly blame them.

First, RJ specifically mentions the conversation rate of launch edition vehicles. I think they are surprised at how many people are opting for the higher ASP vehicle, not the % who are converting orders.

Second, just because Rivian is saying you'll be invited to place your order in June doesn't mean you'll get a vehicle in June. Your actual delivery date could be December or even into 2028. They are clearly staggering order processing to avoid some of the teething problems from R1, but it's important to distinguish between "order invite" and "fulfillment."

 

[UnsungZero_OldTimeAdMan]

Asian parents - "We expect you to get straight AAA+"

And not top of class. And not like so and so's kids. Yah, heard it all. Have PTSD from it.

Sponsored