Amazon product trend analysis done programmatically.

What Volatilo is

Volatilo, simply put, is a trend analysis and data aggregation tool for selected product categories. There are excellent tools that provide detailed pricing data for every product on Amazon ever sold down to the hour. These tools use tremendous amount of data but much of it is offered at a raw state most users cannot make sense of very easily.

Volatilo bridges this gap. We — here meaning a set of Python scripts — analyze the price and sales momentum of category leaders (100 top sellers in each sub-niche to be precise) to show whether the product is gaining prominence among its peers, whether it can hold its ranking without significant markdowns, and most crucially, whether the seller has a predictable habit of offering lower prices throughout the year. We think this is valuable information as many customers would prefer to take advantage of a markdown event. We try to make an educated guess about when the next price event takes place or if we are observing a marked down price point we try to guess whether the prospective buyers should hurry in making their decisions.

What it is not

Volatilo is not a price tracker. We do not record, track, or publish current price. There are excellent price trackers. Keepa even offers a Google Chrome extension that allows you to track prices of products and notify you when a price drops. We do not provide real time price alerts.

It should go without saying but we are not giving users advice on purchases or timing. We stand by our analyses but these are statistical analyses based on past data, and predictions and guesses based on past data always come with a certain amount of uncertainty. And it should also be clear that neither our data partners nor Amazon is responsible for what you see on these pages. Both analyses and interpretations belong to us and they are in no way binding for Amazon or Amazon sellers. We only provide data and some guesses. Decision is up to you.

Who is “we”?

I built Volatilo working on my own (and as would be expected these days) with the help of various AI assistants. Much of the coding is Claude and Gemini, but concept, analysis and signal design, data collection, and database schemes all are my brainchildren. The weekly data updates run more or less automatically, but I write an occasional analysis or comparison article. And one more thing, when you send an email to info@volatilo.com, it reaches my inbox, not an automated queue.

The role of AI has made authorship of digital products very slippery. In an age of anonymous and faceless webpages, I wanted to put a name on record. Much of the code here came from AI tools. But as design and concept belong to me, I also take responsibility for any errors and mistakes.

Financing Volatilo

Volatilo is a free tool to use and will always remain so. It might earn revenues through affiliate links and display ads. But we will never sell a spot on our pages or change the analyses presented here in exchange of any financial incentive. In other words, we will always be honest and upfront about the data we use, the products we choose to present on these pages, and predictions we will make. These will always be determined by a script that takes only observed data into account and nothing else. In even simpler terms, we will never take money to feature a product or add or remove a promotion event. We will only respond to content issues when we are demonstrably and factually wrong. We will not alter any interpretation or analyses (unless these are affected by factually wrong data).

Independence

No product is on this site because a brand asked for it, and none has been described differently because one did. What appears is whatever sat in the top 100 of a tracked category on the day I read it, plus what recently left one of those lists and is still being watched. I take no submissions, grant no editorial favours, and have no relationship with any seller whose products appear.

Volatilo is not affiliated with, endorsed by, or operated by Amazon or Keepa. The underlying observations are retrieved through the Keepa API and published here as derived weekly signals with Keepa's written consent.

Where AI is used, and where it is not

Claude and Gemini wrote large sections of the data collection, analysis, and frontend code. I designed the framework for analyses and overall frontend principles, but actual TypeScript and Python code mostly came out of AI assistants.

While AI is extensively used in the design, there is no LLM in the dynamic text used on category and product pages. Each analysis is strictly and programmatically tied to the underlying data. Product pages are updated weekly on an automated schedule, but what is presented on page is not generic AI slop. It's data interpreted in plain English, tied to strict rules based on trends, momentum, gaps, and statistics. I made every effort to make it feel less repetitive, but being precise and accurate is important for this project.

Since I designed how each tracked piece of statistical analyses is translated into words, I take responsibility for it. If you see something that makes no sense, it's a material mistake or one of my scripts failing, not an LLM hallucinating.

If something is wrong

A figure that was wrong when it was read is worth telling me about, and so is a product record you want taken down. Both reach the same mailbox; the contact page says what to include so a reading can be traced back to the row it came from.