Amazon Sales Estimation
This is a guest post from Avi Eisenberg.
Disclaimer: There are endless potential provisions to the statements below. This article is not the epitome of all Amazon sales, and these tactics don’t work for every seller. Do your own research, make your own decisions, don’t take any estimate as certain. Remember that past sales do not always mean future sales. There is always more to find out about your Amazon sales, so any comments or contributions are encouraged.
Below are a few tips that can help you with Amazon sales estimation. There are several “tiers” an item can fall into. Let’s segment them as shown below:
- Top tier, >100 sales/day 3k/month
- Second tier, >10 sales/day 300/month
- Third tier, >2 sales/day 60/month
- Slow tier, 1 sale/day, 10-50/month
- Slower tier, <1 sale/day, 1<10/month
- Slowest tier, <1/month
There are also some curveball items that are difficult to evaluate with standard Amazon seller tools. There are primarily items with subcategory rank and items with no rank. I will provide some tactics on how to estimate sales for those later.
The difference between the top tier and the lowest tier is quite obvious. Things become more difficult when looking more specifically, and figuring out where an item might fall within a tier. You can subdivide items differently, and you might only be interested in specific tiers – I will discuss all of these briefly.
One note of caution that qualifies everything below: some categories, like clothing, have only a single rank for all the variation items listed under the same parent ASIN. Do not assume that your item sells as well as the listing sells overall. This requires more experience, and every seller must take this information with a grain of salt if clothing makes up the majority of inventory. Also, the seasons can have a serious impact on your listing’s rank. Definitely don’t take review counts from January, which likely reflect December sales, and expect that to predict March sales.
Defining the Sales Tiers
Starting from the lowest tier, number 6, long tail items. You can count sales quite easily if keepa tracks it, every rank drop is huge and is on sale. It’s important to check the buy box price during those sales because you can’t rely on selling it for more than that. Lots of books are great examples here, they may sell 3 copies a year at $50. If it costs you $3, they might be great to send in FBA and forget about. These items may have zero reviews – if the chart shows sales, don’t worry about it. If the item is in this category, rank is useless, there’s no use knowing it is 1 million vs 5 million. The only valuable piece of info here is the number of drops in keepa.
Tier 5 and better, we see multiple drops per month in keepa. Typically, if you look at a months period and see less than 10 drops, each drop is one sale. You can do a sanity test with review count. Take this general rule of thumb: review rate is between 2 and 10%, so reviews times 10 and times 50 are rough upper and lower bounds for sales. Again, check for significant price changes over time that can affect sales.
These two tiers are pretty simple. Each drop corresponds to one sale, for the most part, the sales don’t come closer together, so you can literally count the number of sales. One thing to consider is the in-stock percentage. If an item was sold out for much of the time period, then the fact that it didn’t sell gives you no info either way, so it’s possible keeping it in stock will lead to more sales.
When we consider Tier 4 and up, it gets complicated. You can’t simply count rank drops, because keepa doesn’t sample often enough to capture each rank drop, plus not all sales cause a rank drop. Here, it is very useful to have some background knowledge of other items in the same category. For example, I sell an item ranked about 10k in beauty. Over the last 30 days, I have had 41% of the buy box and 235 sales. That means, roughly, that it sells 235÷.41=~573 units a month. This is very helpful if I was wondering how many items to buy of an item ranked 20k in beauty – if I expect to be one of 3 sellers, getting 33 percent of the buy box, I would probably expect to sell about 100 a month (573/3=191, then round a bit down because the rank is worse). Do not compare ranks between categories, a rank of 20k in e.g. software is going to be tier 6 (see e.g. B0002VAPDE).
Tools for Sales Estimation
You can use estimator software. One Amazon sales estimation tool that comes highly recommended is Market Intelligence from Viral Launch, which is $20/month. They are good for some categories, not so great for others. I use them to get an order-of-magnitude estimate, which then can be refined using some tricks we will talk about in a bit. Note that Market Intelligence, as well as almost all software tools, will only give estimates for items with main ranks, not subranks – we’ll give a couple of tricks for handling those later as well. There’s also Tactical Arbitrage, Jungle Scout, and other software that use different techniques and can give you ideas.
Looking at tiers 2,3, and 4 together now: software should give you a rough idea where you are. If market intelligence says it sells 50/month, probably it sells between 25 and 100. I wouldn’t be too surprised if I sold 200 in a month, but it shouldn’t be counted on it either. If sellers need more specific estimates, then it’s very helpful to sell something in that category already. Take the 30 and 90-day averages, find something you sell in a similar rank range, and round up or down a bit based on that. Use reviews with the formula provided above to get better bounds as well. The more pieces of info you have, the better idea you have. Also, if you sell items in the same niche, you may have better data on typical review rate for that niche (count sales over a long period vs. reviews).
Once you get above 1k sales /month, it gets more difficult to maintain accuracy. If you’re only buying 100 units, knowing it sells over 1k/month is all you need to feel secure with the purchase. But if you want to make large quantity buys, you should seek out accurate info. At this scale, it may be worth it just buying 100, maybe at a higher price, shipping them in and selling them to see how many move. Then, take buy box share and do simple math to get an estimate for monthly sales. Again, selling other items in the same category can really help.
Outlined below are some techniques for dealing with subranks or no ranks, and that may help on other items as well. If there are no reviews and no rank, the item almost surely doesn’t sell. Ignore it. If there are reviews, it might just be a glitch that it doesn’t have a rank. You can estimate sales based on the reviews, but don’t count on accuracy there (that said, if it has to say 30 reviews in the last month you can figure it is moving pretty well.)
The Storefront Method vs. The Subrank Method
The two techniques I will describe are the storefront method and the subrank method.
The storefront method takes advantage of a simple fact: the order of products on a seller’s storefront is roughly ordered by sales. Some sellers have noticed it is not the same exact algorithm as rank – for example, we have seen items in the same category ranked lower that are earlier in the storefront. However, it is close enough.
The Storefront Method: Go to any seller on the item in question. Click the seller name and go to their storefront. If they have thousands of items, this won’t be practical. However, if they have a reasonable number of items, scroll through the storefront until you find the item you are looking at. Now, estimate sales on the items before and after this item. If you are lucky, those will be easier to estimate – they will be in a category you are more familiar with, they will have a rank/main rank, etc. From here, you can assume your item sells somewhere in between the one before and the one after. If the seller does not have a storefront that makes it practical to do this, just look at other sellers on the listing until you find one that does. Another thing you can do with the storefront method – list it as merchant-fulfilled at a high price, then see where it shows up in your storefront. You will know how well your other items next to it sell because you are the one selling it (and if there’s a shared buy box, divide by buy box % to approximate total sales). If you have enough items that cover various ranges of velocity, this could get you more accurate data than any other method.
The Subrank Method: Suppose an item has a subrank of 75. Sellers can click the subrank and identify the items with subranks between, say, 70 and 80. It is very possible one of them has a main rank as well as a subrank. If so, you can use that to estimate sales using the estimator tools, and extrapolate back to your item. If your item has a subrank above 100, you can look at the item that is 100 in the subrank as an upper bound. If you’re 200 in the subrank, then you’re probably fairly close to 100, if you’re 20k in the subrank, then you’re probably very far away from the 100 subrank item.
Note that these methods can be chained: for example, you can use the similar subrank method to find another item, then estimate that item’s sales using the storefront method. Ultimately, it is about relating things you don’t know to items or categories you do know.
Some of these techniques might be overkill for your purposes. Sometimes, though, buying the right amount and avoiding overstock 2-3 times as much as needed, can be vitally important. The more information you have, the better. The sellers with the best data always win.