Do startup leaderboards rank products or audiences? Early Arena data

What those four numbers mean

“Completed battles” counts leaderboard battles finalised with a valid result during the UTC month. It does not count active battles or No Contests. “Arena Votes” counts currently valid Arena-sourced votes cast during that month, including votes in battles that may still have been live at capture time. The two numbers therefore have different populations: dividing 1,332 by 183 would not give a valid average votes-per-completed-battle.

“Distinct products moved” counts products whose leaderboard position changed in rank events during the month, including products pushed down when a challenger climbs. “Ladder moves” adds the positions affected by qualifying rank events. It is not a count of unique products, battles, or people. These definitions follow the code that produces the public summary.

A closer look at the 100 newest public battles

We also captured the 100 newest public leaderboard battles at 21:52 UTC. At that moment, 96 were completed and four were still active or in extra time. The 96 completed battles had a median of seven Arena Votes. Eight of those 96 displayed at least one Support Vote. Challengers won 68; defenders won 28.

This is a recent, creation-ordered slice covering battles started from 24 to 29 September, not a random sample of the full month. The seven-vote median also reflects the platform’s result threshold. Challenger wins in this slice do not establish a general challenger advantage: founders choose an eligible rival within the rules, and the matchups were not randomly assigned. The archived 100-battle CSV contains the public scores used here.

Why a founder’s audience is shown separately

Founder Battle separates votes by their route into a battle. A person who discovers a matchup inside the Arena can cast an Arena Vote that counts toward the result. A vote through a founder’s shared battle link is a Support Vote. Support is visible, but cannot meet the Arena minimum, decide the result, break a tie, or move rank. Paid boosts promote both sides of a battle equally and do not buy votes or rank. The full thresholds and cooldowns are on How ranking works.

The Lyntiq AI vs ProposalForge AI result makes the separation concrete. At the snapshot, Lyntiq AI had five Arena Votes against ProposalForge AI’s two and won the battle. ProposalForge AI had three Support Votes from shared traffic; Lyntiq AI had none. Those Support Votes did not reverse the Arena result.

What the data does not establish

A battle is a choice made by participating voters after seeing two public products. It is not a random sample of buyers, a controlled usability study, a sales comparison, or a claim that the winner is best for every task. A seven-vote result can be valid under the published rules while still being a small signal. Voters may know a brand already, misunderstand a product, or be better suited to one category than another.

Separating Support Votes removes the most direct route for a founder’s shared audience to determine a result. It does not eliminate every possible influence on the Arena judge pool. We should assess that as participation grows, report anomalies, and keep the rules and vote-integrity boundaries visible. We will not describe these early counts as proof that product quality has been objectively measured.

How to read any product leaderboard

1. Check whether rank means launch-day popularity, a rolling vote total, revenue, an editorial choice, or wins in direct matchups.
2. Look for the denominator: how many people voted and during what window?
3. Ask whether a creator’s own traffic can affect the competitive result.
4. Open the underlying product pages and try the products for your own use case.
5. Treat a narrow or small-sample result as a prompt to investigate, not a universal verdict.

For founders, the best next question is not “Did I win?” but “What did people understand, what did they miss, and which user task should I test next?” Our feedback guide shows how to turn that signal into a product decision. You can also inspect current and finished battles yourself.

Method and update policy

We captured the unauthenticated public Arena summary, the 100-record public battle list, and a completed public battle response on 29 September 2026 at 21:52 UTC. The summary groups completed leaderboard battles by finalisation time and valid Arena Votes by vote time from 1 September UTC. Rank movement comes from rank-change events in the same window. For the 100-record slice, we counted statuses, added each completed battle’s Arena scores, and counted completed battles with either Support score above zero. The example shows publicly reported scores at capture time. The snapshot manifest records the totals, cutoff, and source URLs. Moderation or vote invalidation may change a later live view, so this article keeps its original cutoff rather than silently changing the numbers.

This is an observational report about Founder Battle’s own early activity. We operate the platform and benefit when founders use it. We will date any substantive revision and link to public source pages so readers can check the current record.