The Median Is 6: What TPT's Power Law Means for New Sellers
We expanded our dataset to 23,000+ Teachers Pay Teachers products. The single most important number in the entire dataset: 6. That is the median number of downloads across all products. The average of 310 is a statistical illusion — a number that describes almost nobody.
The Distribution Nobody Talks About
When TPT sellers discuss performance in Facebook groups and forums, they talk about averages. "The average product gets 300+ downloads." That is technically true. It is also deeply misleading.
Here is what the actual distribution looks like across 23,000+ products:
| Download Range | % of Products | What It Means |
|---|---|---|
| 0 downloads | 29.5% | Nearly 1 in 3 products have never been downloaded. Not once. |
| 1–999 | 66.1% | The vast majority. Most of these are under 50. |
| 1,000–9,999 | 4.0% | Meaningful traction. You are in the top ~5%. |
| 10,000+ | 0.5% | The hits. These pull the "average" up to 310. |
That bottom row — the 0.5% — is doing all the work. A handful of products with 50,000, 100,000, or even 500,000 downloads inflate the average so dramatically that it bears no resemblance to the typical outcome. The average is 310. The median is 6. The gap between those two numbers tells you everything about the shape of this marketplace.
The Average Is a Lie
If you lined up all 23,000 products by download count, the product in the exact middle has 6 downloads. The "average" of 310 is pulled up by 0.5% of outlier hits. When someone tells you the average TPT product gets hundreds of downloads, they are technically correct and practically useless.
This Is a Power Law, Not a Bell Curve
TPT's download distribution is not a normal bell curve where most products cluster around the average. It follows a power law: a small number of products capture the vast majority of downloads, while most products get almost nothing.
This is not unique to TPT. Power law distributions appear in almost every marketplace and creative economy:
- App stores: The top 1% of apps capture over 90% of revenue. Most apps get fewer than 1,000 downloads.
- Amazon books: A small number of titles account for the majority of sales. Most self-published books sell fewer than 250 copies.
- YouTube: The median video gets a few hundred views. The average is pulled into the thousands by viral hits.
- Music streaming: The top 1% of artists earn roughly 90% of streaming revenue.
The pattern is universal because the underlying mechanism is universal: visibility compounds. A product that gets early downloads ranks higher in search, which generates more downloads, which ranks it higher still. Products that do not get that initial momentum stay invisible. The rich get richer. The invisible stay invisible.
What 29.5% Zero Downloads Really Means
Nearly one in three products on TPT has never been downloaded. Zero. Not even once.
These are not abandoned drafts or accidental uploads. Many of them have descriptions, thumbnails, and reasonable pricing. They simply never gained any visibility in TPT's search results, and the sellers did not have enough followers to generate notification-driven downloads.
The zero-download products are concentrated among sellers with fewer than 100 followers. Without an existing audience to trigger the initial downloads, and without the search ranking signals that come from those downloads, these products enter a visibility dead zone they may never escape.
The Cold Start Problem
TPT has a classic cold start problem. Products need downloads to rank in search. But they need search visibility to get downloads. The only way to break the cycle: an existing follower base that provides the initial push. This is why follower count matters so much (Part 2 of this series).
Products Compound — If They Survive
One of the more encouraging findings from the expanded dataset: products that do get traction continue accumulating downloads for years. This is not a flash-in-the-pan marketplace. It is a long-tail platform where resources compound.
| Year Published | Avg Downloads | Years on Platform |
|---|---|---|
| 2014 | 2,527 | ~12 years |
| 2016 | 1,834 | ~10 years |
| 2018 | 1,102 | ~8 years |
| 2020 | 743 | ~6 years |
| 2022 | 521 | ~4 years |
| 2024 | 187 | ~2 years |
| 2025–2026 | 42 | <1 year |
Products published in 2022 average 521 downloads after roughly 4 years. Products from 2014 average 2,527 after about 12 years. The growth is not linear — it decelerates as products age and newer competitors enter the same niches — but it is real and sustained.
The implication: every product you upload today that gains any traction at all is a compounding asset. It will continue earning downloads in 2028, 2030, and beyond. The question is whether it ever gets past the initial visibility hurdle.
The Practical Math for Year One
Let us be brutally honest about what the power law means for a new seller entering TPT today.
The Optimistic-Realistic Scenario
Suppose you commit to publishing one product per week for a year. That is 52 products — an ambitious but achievable pace if you are creating worksheet-style resources.
Based on the distribution data:
- ~15 products (29%) will likely get zero or near-zero downloads
- ~32 products (62%) will get somewhere between 1 and 100 downloads
- ~5 products (9%) will get meaningful traction — hundreds or low thousands of downloads
- ~0–1 products might break into the 1,000+ range in year one
Those 5 products with traction will account for the majority of your total downloads. They are the ones that found a niche, hit the right keywords, or caught a seasonal wave. You cannot predict in advance which 5 they will be.
The Compounding Argument
Here is where the math gets interesting. Those 5 products that got traction in year one? They continue compounding. By year three, if the 2022 vintage data holds, they could each be averaging 500+ downloads. Meanwhile, you have published another 104 products in years two and three, and some of those will also find their groove.
The sellers who reach 10,000+ followers and 1,000+ average downloads per product did not get there by launching a single viral hit. They got there through volume, patience, and the mathematics of compounding returns on a power law distribution.
The Volume Strategy
You cannot predict which products will succeed. But you can control how many you publish. On a power law distribution, the optimal strategy is to maximize the number of "at bats" while maintaining quality. Publish frequently, learn from what works, and let the compounding do its job over years. Most products will underperform. A few will carry the catalog. That is not failure — it is how power laws work.
Why This Pattern Persists
You might expect TPT to try to flatten this distribution — to give new products a boost, or to limit the visibility of products that already have thousands of downloads. Some marketplaces do this (Etsy, for example, has experimented with "new seller" boosts).
But TPT's incentives point the other direction. The platform earns a commission on every transaction. Products with proven track records convert browsers into buyers at higher rates. Showing a new, unrated product instead of a proven product with 5,000 downloads and 4.8 stars is a risk to TPT's revenue. The algorithm rationally favors incumbents.
This means the power law is not a bug. It is a structural feature of the marketplace that serves TPT's business model. Understanding this is not cynicism — it is strategic clarity.
Comparing to Other Marketplaces
Is TPT's distribution unusually harsh? Not really. If anything, it is slightly more forgiving than some alternatives:
| Marketplace | Median Outcome | Top 1% Share |
|---|---|---|
| TPT (our data) | 6 downloads | ~65% of total DL |
| Apple App Store (industry reports) | <1,000 lifetime DL | ~95% of revenue |
| Amazon Kindle (industry estimates) | <250 copies sold | ~80% of revenue |
| Udemy (instructor data) | <50 enrollments | ~90% of revenue |
The power law is the default state of any marketplace with open entry. TPT is not uniquely unfair. It is normally distributed for a platform of its type. Knowing this in advance is an advantage — most new sellers enter expecting the average outcome, not the median one.
What to Do With This Knowledge
- Expect the median, plan for the average. Your typical product will get single-digit downloads. Your portfolio, given enough products and time, will trend toward the average. The unit of analysis is your catalog, not any single product.
- Publish weekly. On a power law, volume matters. One product per week gives you 52 chances per year for something to gain traction. Each chance is small. Fifty-two chances is meaningful.
- Do not over-invest in any single product. Since you cannot predict hits, spreading effort across many products is more efficient than perfecting a single product. Good enough, shipped weekly, beats perfect, shipped monthly.
- Use free products to break the cold start. Free products average 1,765 downloads vs. 76 for cheap paid products (Part 1 data). They are the most reliable way to build the follower base that powers everything else.
- Think in years, not months. Products from 2022 now average 521 downloads. Products from 2014 average 2,527. The compounding is slow but real. A three-year commitment is the minimum viable strategy.
- Study the 5–10 products that work. When something catches, double down. Make variations, sequels, and bundles. One hit can seed an entire product line.
The Honest Take
If you publish 52 products in year one, maybe 5–10 will get meaningful traction. The other 42–47 will sit quietly with single-digit or zero downloads. That is not failure. That is how every marketplace with a power law distribution works. The question is not "will most of my products underperform?" (they will) but "am I publishing enough to find the ones that compound?" (you need to be).
Methodology Notes
This analysis uses an expanded dataset of 23,000+ products scraped from TeachersPayTeachers.com, up from the 20,761 products analyzed in Part 1. The additional products were collected using the same Apollo state extraction technique across broader search categories.
Download count distributions were calculated from the full dataset. Year-over-year compounding estimates use average downloads grouped by publication year. Comparisons to other marketplaces use publicly available industry reports and should be treated as approximate rather than exact.
As always: this is a sample, not a census. TPT has millions of products. Our data captures products surfaced through search results, which may skew toward higher-visibility products. The actual median across all TPT products may be even lower than 6.
Up Next in This Series
Part 5 is a correction. We got something wrong in the original series — our claim that DOC files outperform PDF turned out to be a confound. At 23,000 products, the truth is more nuanced and more interesting. Read Part 5: We Were Wrong About DOC vs PDF
This is Part 4 of an ongoing series on TPT marketplace research. Part 1: the big picture data | Part 2: the follower multiplier | Part 3: file formats and pricing | Part 5: DOC vs PDF correction | Part 6: the three-part flywheel. Follow my journey as I learn new skills and build tools with Brian at Actyra.