+2,000% growth can start with 50 visits
Jason T. · 2026-10-03
+2,000% growth can mean going from 50 visits to 1,050. A tiny starting base can produce a large percentage without a large absolute gain. Before using a "fastest-growing AI tools" list to pick a site to investigate, check how much traffic it actually added. That example is hypothetical, not a real site. New Site Radar uses third-party monthly traffic estimates from TabAPI, published about a month later; its current snapshot describes the previous full month. The methodology explains how those estimates are used.
Put the percentage beside the gain
Here is the whole comparison, using two hypothetical sites: 50 → 1,050 visits is +2,000%, a gain of 1,000 visits. Going from 20,000 → 30,000 visits is +50%, a gain of 10,000 visits. The second site gained ten times as much traffic. Sort those sites by percentage alone and the first comes out ahead. Sort them by added visits and the order reverses. Neither calculation is wrong. They answer different questions. For an indie hacker deciding what to research, I would keep both answers visible. The first site changed more relative to its starting size. The second added more traffic. A percentage-only headline hides the distinction that would help me decide where to spend my time.
Why I use NEW instead
I built New Site Radar, and this is the rule I use: if a site had fewer than 1,000 visits in the previous month, I compute no growth percentage and mark it NEW. A jump of +9,900% or more is also treated as a launch and shown as NEW, even from a base of at least 1,000 visits. It is never shown as a capped percentage. That puts a limit on what I ask a percentage to communicate. At a small starting base, the denominator can dominate the impression. With a launch-sized increase, I would rather show a label than present the movement as an ordinary monthly growth comparison. The rule concerns traffic levels and changes. That is the meaning I would carry into any comparison.
Separate ranking eligibility from growth
A site needs at least 3 months of traffic history and at least 5,000 visits in the latest month to be ranked by New Site Radar. Those requirements answer a different question from the NEW rule: does the site qualify for the ranking? The distinction matters when a list is doing the work of a research shortlist. There is also a useful limit to the opening example. A comparison can clarify percentage growth without representing two actual ranked entries.
Read the estimate and the month
When I read a traffic ranking, I want to know whether the figures are estimates or a site's own counts. Here, they are estimates. Timing deserves the same care. For this ranking, the publication delay means I am looking back at a completed month. Coverage is another part of the interpretation. New Site Radar labels sites covered by only one traffic provider single-source. I read that as information about the estimate's coverage. I would keep the unit intact, too. The traffic figures are monthly visit estimates.
A short checklist for any growth ranking
Before using a ranking to choose an AI tool to investigate, I would make these checks:
- Read the base and calculate the gain. Put the previous and latest traffic figures beside the percentage. Subtract the starting figure from the ending figure.
- Check the estimates. Find who estimated the traffic and which month it covers. Check whether a single-source label applies.
- Read the entry and launch rules. Check the history and traffic requirements, then find out how the ranking handles tiny bases and launch-sized jumps.
I would do this before opening a string of product pages. It is a short way to decide what the ranking actually tells me, rather than letting its sort order make that decision. When using the ranking, start with the site that interests you and write down its starting traffic and absolute gain beside the percentage. Then decide whether that movement deserves a closer look. In the hypothetical comparison, the +50% site added ten times as much traffic as the +2,000% site. That is the number I would keep beside the headline.
More notes
- Which AI tool categories are gaining traffic in August 2026?
- What rising 1980s AI photo searches ask for
- Two kinds of AI sites need two different traffic checks
- Why I won't rank an AI site with only weeks of traffic history
- How every number on the site is checked