The 'How big is the Amazon Marketplace' debate...with analysis
The other day I was discussing Amazon's third-party marketplace (3P) with someone 'in the biz' and the said..."I mean with Amazon now bigger than Walmart..."....whoa....whoa...whoa...hold the bananaphone (as my 6-yr old would say)!!!!!
I asked for some clarity on that statement and here was their logic:
- Amazon will do ~$50b in revenue this year
- 40% of their sales are 3P, round that up to 50%
- That means $25b are from 3P revenues, $25b are 1P GMV
- If you assume a 10% take-rate, then for that $25b in 3P REVS you end up with $250b in GMV
- $250b 3P GMV + $25b 1P GMV = $275b total GMV which is about the size of Wal-mart
Hmmm, compelling, but it makes one assumption that throws the whole thing off.
This has come up enough (both on the high and low side) that I thought it was worth digging into as we wait for Amazon's Q3 results next week. We're also seeing a lot of traditional manufacturers come in and ask about their Amazon strategy and this is one of the FAQs that comes up from those conversations.
What's wrong with this back-of-the-napkin analysis?
There are many statements in this analysis that are 'correct':
- Amazon's 2012 revenues are likely to be $50b
- Amazon's revenues do have a mix of 1P GMV and 3P Revenue
- 3P does have an average take rate of around 10%
However, there is one fundamental problem with this analysis:
- What amazon states (as of Q2) is that 40% of their PAID UNITS are from 3P. If you think about it, that's very different from 40% of revenue coming from 3P.
- Rounding from 40% to 50% is pretty generous - but that's a small one compared to the above.
What's the 'right' analysis? (or how big 'really' is Amazon's 3PM in 2012?)
I think it's really important that existing Amazon sellers and everyone else in the e-commerce ecosystem understand this because at ChannelAdvisor we think Amazon is one, if not the largest waves of change sweeping over the industry, so it warrants a bit of a deep dive.
First, it's important to note that Amazon doesn't give us everything we need to precisely figure out the size of their third-party marketplace. I suspect they do that on purpose to keep competitive information close to the vest and that's certainly smart on their part and it's clearly working.
With that caveat in mind, what we can do is build a model that makes some educated assumptions to get us in the ballpark and along the way learn some pretty interesting things about Amazon's business:
For the purposes of this discussion I'll walk you through this simple 10 line spreadsheet and reference the 'Line' column: (click to expand):
Looking at Lines 1, 6 and 10 - we have our 'constants' - let's assume Amazon will do $50b this year and that there's a $50 ASP and that the 3P mix stays at 40%. Also, there's an assumption in here that we have a 10% take rate (in Line 5's calculation).
The question we are trying to answer is essentially: What is the 3P level of GMV that will result in a 40% unit mix for 3P? You can solve for that with algebra, but to keep it simple let's look at how to solve it in a simple spreadsheet you can recreate on your own.
First, we need to normalize our data so we have apples to apples (GMV to GMV) and we also need to get to 'units' because that's the only metric we have (40% unit share for 3P).
On Line 2, we assume some amount of the $50b is 1P and is essentially 1P GMV. Then on line 7, we calculate the units by dividing that by $50 from line 6.
For 3P units, we take the remainder ($50m-1P GMV) and assume that is 3P revenue. We then gross that up 10x to reflect the 10% take rate. In line 5 we have $3.125b in 3P revenue and gross that up to $31.25b in GMV.
To get 3P units, we divide by $50 ASP (line 6 again) and get the units on line 8.
Now we are left with:
- Line 2 + Line 3 = Total GMV - or 1P GMV + 3P GMV = total GMV
- Line 7 + Line 8 = Total units - or 1P Units + 3P Units = Total UNITS
From here we can divide the 3P units by the total to get the % that are 3P and the answer works out to 40%
Drum roll please!
Based on this simple model, we approximate that Amazon's 2012 3P GMV will be $31.25B and when combined with 1P, the company has $78.125b in GMV. That's certainly a good bit smaller than the $275b if you take the wrong path.
For many folks in the industry, the $31.25B number probably feels high as well because they think of Amazon as a $50b sales retailer, when in fact they have another $23b under reported in there hiding in plain site that 'really' makes them effectively a $78b retailer (from a retail economic impact perspective).
This is part of what is freaking out Wall St. lately - Amazon's revenue growth is coming down some, but the margins are expanding. It's more understandable when you think about this math we just walked through and you assume that 1P margins are low and 3P margins are very high. In fact, this is exactly what is happening because Amazon's 3P units are growing at ~65% and 1P is growing ~30%.
To illustrate this, let's say that Amazon grew to $60b in 2013 up from 2012's $50b - that would be 20% growth and feel like the company hit a wall. But let's assume that 3P goes to 55% of units over that time-frame under the hood. Note, I'm picking this number because it is absurdly low to illustrate my point, but here's what it would look like:
So while revenue grew a paultry 20%, GMV rocketed from 2012's total $78b (1P:$47b, 3P:$31b) to $118.7b (1P:$53b/$65b) - almost a double of impact and in fact 3P doubled in this model while 1P grew 20%.
Imagine if while this happened that 3P was 2-3-4-5X as profitable as 1P, you'd see a huge change in the profit profile as it happened - Amazon would essentially appear to slow (but not really from an economic impact standpoint) and their margins would go up - yet their prices would stay low. Feels weird, but when you peel the onion and really understand there are essentially two business inside of here, it starts to make sense.
Again, I use this example to illustrate that the 10% 'downshift' of 3P GMV into Amazon revenue effectively masks and understates a large amount of the current and possibly future impact of the 3P.
How would you build a more detailed model?
Earlier, I mentioned that this is a simple model. The natural next question is - well how can we get more precise? First, if you want to get really exact here are some broad assumptions in the model that should be challenged/explored:
- 10% take rate - Amazon charges 8% in CE and as much as 25% for Kindle accessories. If you had the data, you could model out a blended take rate. For example, a little less than half of Amazon's revenue is Media and half is EGM - you could build out a model here that maybe shows a 13% take rate or if EGM is heavily electronics, you could say 9%. It's not clear to me where payments is reported either, that's another 2 points that you may need to include here, so think of take rate as having a range of 9-15%
- ASP - We've assumed a constant ASP. Well if half the site is media and media is usually < $20, that's an aggressive assumption for that category.
- Annual vs. Quarterly - To really get this right, you should model it quarterly because the numbers are all moving rapidly - e.g. it wasn't too long ago that Amazon's 3P units were 33%.
- Category/3P mix in EGM / Media - In our experience, the EGM side of Amazon has a higher 3P share than 1P, that times the take rate impact times the ASP impact could swing things also.
- Other - Amazon has an other category with some non-GMV items like advertising, some Kindle items, etc. You could build a model to pull that out.
Here's a slightly more complex model that puts a different ASP in there for 1P and 3P to give you an idea of how far you can take this:
Here you can see that the one change of $20 in ASP between 1P and 3P has goosed up the 3P GMV to $50b (because the 1P units soared, which raises the 3P units significantly to meet 40% which when multiplied times the new ASP -boom.)
The real answer....How big is Amazon's 3P?
The real answer is....Only Amazon knows. Because we are given one data-point (40% orders) of about 10 we need, we can only make an educated guess at a range of the scale of Amazon's 3P. Our best guess generates a range of $25b-$55b with a midpoint of $40b for 3P which when added to 1P gives Amazon a $86b impact in 2012. While not quite yet the size of Wal-mart, that puts Amazon's 3P in the zipcode of eBay's $65b in GMV (ex-autos) and over when you consider 'all' of Amazon.
The bottom line: Amazon may look like a $50b retailer, but it's really a ~$86b GMV retailer+marketplace when you take into consideration the true impact of 3P.
This blog post was written by Scot Wingo, CEO of ChannelAdvisor.
I am long Amazon and Google. eBay is an investor in ChannelAdvisor where I am CEO.