Brand-built vs platform-hosted contests: pick your battles

Every week someone messages us with a contest link and the same question: "Is it worth entering?" The honest answer almost always depends on which kind of contest it is. Not the prize. Not the brand. The plumbing underneath.

There are basically two species. The first is the brand-built contest — a custom landing page on WordPress, Tilda, Webflow, or a one-off Next.js build, where the brand wrote its own rules and runs its own counting. The second is the platform-hosted contest — Gleam.io, Woobox, ShortStack, Vyper (until it closed on 12 September 2024), or similar SaaS tools where the brand just configures a template. Both are legitimate. Both have very different traps.

We've supported entries in roughly 340 contests since 2022. About 61% were brand-built, the rest platform-hosted. The win rates diverge in ways that surprise newcomers. Here's how we think about it before agreeing to help.

What "brand-built" actually means

A brand-built contest is when the marketing team (or their agency) built the contest themselves. It usually lives at contest.brand.com or a campaign subdomain. The voting form is custom. The leaderboard, if there is one, was coded for this single campaign. Sometimes it's a Google Form glued to a Google Sheet. Sometimes it's a polished Tilda block with a JavaScript counter.

What this means in practice:

Detection of duplicate or suspicious votes is whatever the developer bothered to code. Often almost nothing.
The rules document was written by a marketer, not a legal team running a SaaS product. It tends to be vaguer.
The jury, if there is one, is internal. Sometimes the CEO's spouse votes too.
Disqualification happens by human judgment, not by an automated flag.

This sounds permissive. In some ways it is. We've seen brand-built contests where the same IP address could submit fifty form entries and the counter just ticked up. We've also seen brand-built contests where the marketing manager called a participant directly and asked her to explain why her vote graph spiked on a Tuesday afternoon. The variance is huge.

The jury bias problem

Here's the contrarian opinion we hold and clients sometimes don't want to hear: brand-built contests are often less fair than platform-hosted ones, not more. The transparency is worse. The decision-makers are inside the brand. If the marketing director's nephew enters, you are competing against the marketing director's nephew. We've watched a Polish skincare brand publicly announce a winner whose final-round score was visibly lower than two other finalists. No appeal mechanism. No published criteria weighting. The participant who asked for clarification got a polite non-answer.

This is not corruption in any criminal sense. It's just what happens when the contest organiser is also the contest judge and there is no neutral platform sitting between them. We tell every client who enters a brand-built contest: assume the jury phase is roughly 70% objective and 30% relationship-driven. Plan accordingly.

What "platform-hosted" actually means

Platform-hosted contests run on third-party software. Gleam.io is the most common one we encounter for European entrants. Woobox and ShortStack appear too. Their business model is selling contest infrastructure to brands, which means they need their detection and rules to be defensible. If their tooling lets cheaters win, their customers churn.

So they invest in detection. Gleam, for instance, publishes its eligibility and fraud documentation openly, and it's been steadily expanded since 2023. They check device fingerprints, look at IP velocity, cross-reference social account ages, and flag entries that match patterns from previous abuse. Woobox does similar things less visibly.

The rules are also tighter. A Gleam contest will typically state region eligibility, age minimums, account-age requirements for social entries, and what counts as a single entry, all on one page. It's harder to argue with because it was written once and tested across thousands of campaigns.

Where platform-hosted contests get hard

Differentiation. If you and 4,200 other people are entering the same Gleam giveaway, your individual contribution is almost invisible. The structure rewards two things: a large existing audience to mobilise, or a creative entry that lives outside the platform (a video, a blog post) which the platform then ranks. If you have neither, the math is rough. We've seen contests where the top 10 winners all had follower counts above 18,500. That isn't a coincidence.

Platform-hosted contests are also less forgiving of honest mistakes. If you accidentally violate the rules — say, you used a VPN because you were travelling and the platform flagged a country mismatch — there is no human to call. Your entry is just gone.

How we decide whether to help

This is the part where we have to be specific, because "it depends" is a cop-out.

We read the official rules first. If a brand-built contest has no public rules document, or the document is shorter than 200 words, we usually decline. There's nothing for us to align our work with.
We check who counts the votes. If a brand-built contest says "the jury's decision is final and not subject to appeal" without naming the jury or publishing the criteria, we tell the client honestly: the public vote phase is winnable; the jury phase is a coin flip you can't influence.
We check the platform's detection posture. For platform-hosted contests, we read the platform's current fraud policy. Gleam's has changed three times since we started tracking it, most recently in early 2025.
We model the audience. A platform-hosted contest with 800 entries is winnable for a participant with a modest network. The same contest with 6,000+ entries usually isn't, no matter what we do legitimately.
We say no when the math says no. About 23% of inquiries we receive end with us declining to help. Usually because the contest is too saturated, the rules forbid the kind of promotion we do, or the structure is purely jury-driven.

You can see our process documented at /process.html and the work we actually offer at /services.html. We'd rather lose a sale than take money for work that won't move the needle.

A real example, anonymised

A client we'll call M. came to us in November 2024. She had entered two contests simultaneously: a brand-built one run by a Warsaw cosmetics company on a Tilda page, and a Gleam-hosted contest from a Berlin houseware brand. Same prize tier, similar promotional value to her.

The brand-built contest had 1,247 entries at the time she contacted us, no public counter showing votes per entry, and a final jury phase listed as "weighted 40% jury, 60% public vote". The Gleam contest had 3,890 entries and pure points-based ranking.

We helped her with the brand-built one and politely declined the Gleam one. Reasoning: on Gleam, the entries already in the top ten had referral-link counts she'd need a much larger network to match, and her budget didn't justify trying. On the Warsaw contest, the points spread was tight, the audience for the brand's niche was small enough that a focused Meta campaign could reach genuinely interested voters in Poland, and the jury phase was at least partly mathematical.

She finished third in the brand-built contest, which mattered because third place still got a prize. We could not have promised that outcome and didn't. What we could promise was that her entry would reach a relevant audience under Meta's Advertising Standards rules, and that we wouldn't do anything that would get her disqualified.

What we will not do, in either format

This list hasn't changed since we wrote our refund policy in 2023.

We will not buy votes from click farms, vote-trading services, or anyone selling "100% real votes from real people" for a fixed price per vote. Those votes are almost never real, and even when they technically are, they trip detection systems and get the participant disqualified.
We will not create fake social media accounts to vote. Ever. This violates Meta's account integrity policy and every platform's terms.
We will not access a contest platform on the client's behalf using their credentials. The risk to the client is too high and the legitimacy of the entry depends on it being theirs.
We will not promise specific vote counts or final standings. We can promise effort, reach, and ethical methods.
We will not run promotion that violates the contest's own rules, even when the rules are loose enough that we could get away with it.

Honest self-limitation: we got the rule-reading wrong once, in early 2023. A client's contest had a clause buried near the bottom forbidding paid social promotion of the entry page itself. We missed it. The entry was disqualified. We refunded in full and added a second-reader step to our intake. It's the kind of mistake that, in retrospect, was obvious. We don't think we'll make that specific one again. We'll make others.

Practical guidance for each format

If you're entering a brand-built contest

Read the rules twice. Screenshot them, because brand-built rules pages get edited mid-campaign more often than you'd expect. Look for the words "jury", "panel", "discretion", and "final". Those tell you how much of the outcome is actually within reach. If the public-vote weighting is below 50%, your promotional spend has diminishing returns above a modest threshold.

Ask the brand directly, by email, what counts as a single vote and whether referrals are permitted. Save the reply. If their answer contradicts what they later claim, you have something to point at.

If you're entering a platform-hosted contest

Spend an hour studying the platform before you do anything else. Gleam, Woobox, and ShortStack each weight different actions differently. A Twitter follow is often worth fewer points than a referral that produces a sign-up. Knowing the math upfront lets you concentrate effort where it counts.

Don't use a VPN. Don't switch devices mid-campaign without good reason. Don't ask family members in other countries to "just sign up" if the contest is region-restricted; it gets your entry stripped, not boosted.

The boring truth

Most contests are winnable by participants who put thought into their entry, mobilise a real network, and add a small amount of paid reach to relevant audiences. Most contests are not winnable by participants who are starting cold and hoping to outspend the field. The format matters less than that underlying reality.

If you're trying to decide whether to enter a specific contest and want a sanity check, our FAQ covers the common situations, and /contact.html is where we read every message ourselves. We'll tell you honestly if we think helping you makes sense. Sometimes the most useful thing we do is talk someone out of spending money on a contest they were going to lose either way.

The right contest for you is the one where the rules are clear, the math is favourable, and your effort can actually move you up the ranking. Format is a clue to those three things, not the answer.