Researching What People Buy on Amazon
Amazon exposes exactly what people are buying through three public metrics: the Best Sellers Rank, review velocity, and customer questions. You do not need expensive software to read these numbers. You just need to understand the relationship between rank, inventory, and buyer feedback. Building a product without looking at these numbers is a choice to fly blind.
Decoding the Best Sellers Rank
The Best Sellers Rank, or BSR, is a relative metric updated hourly. A BSR of 1 means the product is the top selling item in that specific category right now. A BSR of 10,000 means 9,999 products are currently selling better. Because it is relative, BSR fluctuates based on what other products are doing. A product might sell exactly 10 units every day, but its BSR will drop if a competitor runs a massive discount and sells 500 units.
BSR is split into main categories and subcategories. The main category BSR is the only one that matters for estimating sales. A product might be number 1 in manual garlic presses but number 45,000 in the kitchen department. The subcategory rank is a vanity metric used by sellers to claim a bestseller badge. The main category rank is tied directly to the total volume of the entire department.
To turn a main category BSR into a concrete sales estimate without paying for a subscription tool, you must establish a baseline using the inventory cart method. Find a product in your target category. Add it to your cart. Change the quantity to 999. The site will usually restrict you to the actual available inventory, returning an error stating the seller only has 432 available. Record this number. Return exactly 24 hours later and repeat the process. If the available inventory is now 410, the seller moved 22 units.
Do this for five products at different BSR tiers in your target category. Track a product at BSR 1,000, BSR 5,000, and BSR 10,000 for seven days. You now have a custom, highly accurate formula for that specific category. You might find that BSR 5,000 in the patio department equals roughly 15 sales a day. Once you have this baseline, you can instantly estimate the sales of any competing product just by glancing at its rank.
There are limitations to the cart method. Many sellers set a maximum order quantity of 5 or 10 to prevent competitors from tying up their inventory. If you hit a maximum order limit, you must move on to another product in the same BSR range. You only need a few successful tracking targets to build your baseline.
Calculating Review Velocity
BSR tells you what is happening today. Review velocity tells you what happened over the last month. Not every buyer leaves a review. The standard e-commerce estimate is a 1 percent to 2 percent review rate. If a product gets 15 new written reviews in October, you can estimate they sold between 750 and 1,500 units that month.
To calculate review velocity, navigate to the product reviews and sort by the most recent option. Count the number of reviews left in the last 30 days. Multiply that number by 50 to get the high end of estimated sales, or multiply by 100 to get the low end.
This is a trailing indicator. It is highly valuable because it filters out temporary BSR spikes. A seller might run a 90 percent off lightning deal, spiking their BSR to 500 for a single afternoon. If you only look at BSR, you will overestimate their sustained demand. Review velocity smooths out these anomalies. A product with a great BSR but only two reviews in the last month is riding a temporary traffic spike. A product with a steady 30 reviews a month has consistent, validated demand.
You must exclude ratings that do not include a written review. Look only at written feedback with a date stamp. Buyers can leave a star rating with one click, which happens at a much higher frequency than written reviews. Using the 1 percent rule on total ratings will wildly inflate your sales estimates.
Harvesting Demand Signals from Questions
High sales volume proves a market exists. The customer questions and written reviews show what the market is missing. This is where you find what to build next.
Search the product reviews for phrases like 'wish it had', 'too small', or 'broke when'. These are direct demand signals. If a top selling microphone moves 1,000 units a month but has a consistent 3-star rating because the USB cable is too short, the opportunity is not a completely new microphone. The opportunity is a bundle that includes a three meter cable, or a version with a reinforced port.
The customer questions section is a goldmine for product development. Buyers use this section to ask if a product solves a specific edge case before they spend money. Someone will ask if a particular router supports a specific VPN protocol. If the seller answers no, and ten people upvote that question, you have a validated demand signal for a router that does.
Finding these gaps manually takes time. Unthinkable is a web, iPhone and Android app that surfaces business and app ideas from real demand signals across the web. It is a discovery tool that helps you spot these gaps before you commit to building a solution.
When mining the questions section, look for recurring themes. One person asking for a left-handed version of a tool is an anomaly. Twelve people asking for it over six months is a market segment. Extracting these questions gives you a feature list prioritized by actual buyers with credit cards in hand.
Avoiding the Variation Trap
The most common mistake when researching sales is misreading product variations. The platform groups product variations together to consolidate reviews and rank. A t-shirt listing with ten colors and five sizes shares the exact same BSR and the exact same review pool.
If you look at the BSR of the neon green shirt in extra small, you are seeing the BSR of the entire listing. The black large shirt might be driving 90 percent of the sales, while the neon green version sells one unit a month. If you source a neon green shirt based on the overall listing BSR, you will end up with dead inventory.
To find out which variation is actually selling, you must audit the reviews. Buyers often state the size and color they bought within the text, or the site tags the review with the specific variation purchased. Count the last 100 reviews and tally the variations. If 60 reviews are tagged with the black large variation, assume the black large drives 60 percent of the estimated sales volume.
This requires manual counting, but it is the only free way to isolate variation demand. If a listing has 50 variations, you might find that only three of them account for 80 percent of the revenue. This data allows you to launch a competing product with only the most profitable variations, minimizing your upfront capital risk.
Cross-Referencing Frequently Bought Together
Another powerful indicator of unmet needs is the frequently bought together module. This section is generated by algorithms tracking actual checkout behavior. It shows you exactly what a product is missing out of the box.
If a popular flash drive is constantly bought with a specific adapter, the market is telling you that the flash drive is physically incompatible with modern laptops. The demand signal is clear: build a dual-interface flash drive. If a standing desk is always purchased with a specific cable management tray, the desk lacks adequate routing options.
You can quantify this demand by looking at the review velocity of the accessory. If the cable management tray has a high review velocity and its reviews explicitly mention the brand of the standing desk, you have a proven cross-sell opportunity. You can enter the market by bundling the core product with the frequently bought accessory, capturing the revenue that the original seller is leaving on the table.
This strategy reduces your marketing costs. When you build a product that solves the exact problem created by a top seller, you can target your advertising directly at buyers searching for the competitor. You offer a complete solution while the competitor offers a partial one.
Factoring in Seasonality and Price
A BSR snapshot in November is useless for predicting sales in April if the product is a snow shovel. Always check the date stamps on reviews to identify seasonal trends. If 90 percent of the reviews are left between December and February, you are looking at a highly seasonal product.
Price changes also distort BSR. A product might rank well at 15 dollars and plummet at 20 dollars. When tracking your baseline using the inventory cart method, record the price alongside the available inventory. If the seller raises the price, watch how the daily sales volume reacts. This gives you direct insight into the price elasticity of the market.
If a competitor raises their price by 20 percent and their sales volume drops by 50 percent, the market is highly price sensitive. If they raise the price and volume remains steady, the market values the specific features of that product over cost. You can use this data to position your own product.
Researching what people buy requires discipline. The numbers are public, but they are noisy. By tracking inventory changes, calculating review velocity, parsing the questions section for demand signals, and isolating variations, you strip away the noise. You are left with a clear picture of what sells, what fails, and what buyers are still waiting for someone to build.
Questions
How accurate is the 999 cart method for estimating sales?
The cart method is highly accurate for products with fewer than 999 units in stock. It measures exact inventory depletion over a 24-hour period. However, it fails if a seller imposes a maximum order limit or holds more than 999 units in their fulfillment center.
Why does a product with a high Best Sellers Rank have very few reviews?
A product with a high rank and low review count is usually experiencing a temporary traffic spike. This happens during lightning deals, flash sales, or viral social media campaigns. The rank reflects immediate sales velocity, while reviews take weeks to accumulate.
Can I use the subcategory Best Sellers Rank to estimate sales volume?
No, subcategory ranks are unreliable for estimating sales volume. They are easily manipulated and do not correlate consistently with overall marketplace traffic. You must use the main category rank to build an accurate sales baseline.
How do I find out which product variation sells the most?
You must manually audit the most recent written reviews. Tally the specific sizes or colors mentioned in the review text or tagged by the platform. The distribution of these tags directly mirrors the distribution of sales across the variations.
