Affiliate Contests and Leaderboards That Work

Every quiet affiliate program eventually gets the same suggestion: run a contest. It sounds cheap, you already have the tooling, and a prize makes a program feel alive in a way that another welcome email does not.

Then you run one. The winner is someone you would not have picked, your best affiliate finishes eleventh, a coupon site takes second, and clicks triple while revenue does not move. The contest worked exactly as designed, and the design was the problem. Three decisions decide the outcome: who is allowed in, what the leaderboard scores, and what the losers take away.

TL;DR

• A contest selects for whatever you score. Score clicks and you get click farmers. Score refunded orders and you get refunds.
• Publish the scoring formula before day one. A leaderboard revealed afterwards is a mood, not a competition.
• Cash attracts the wrong entrants. A commission multiplier for everyone above a floor plus a small top three pool beats one big cheque.
• Three weeks is the sweet spot. Under two weeks only finished content can win. Over four weeks the standings freeze and the leaders coast.
• The rest of the field decides whether you get a second contest. Publish full standings and give everyone above the floor something.

Why a contest attracts the wrong affiliates

A contest is a filter, and filters act on whoever shows up. Announcing a prize pool to an unqualified list does not find your best partner. It finds the people who are best at entering contests, which is a different skill from having an audience that trusts them.

Five of those six are bad news, and the reason they win is structural, not moral. Winning a burst contest requires committing to a burst, and affiliates with a real audience will not spend three weeks spamming their readers. Your contest quietly penalised the people you spent a year recruiting.

A burst also creates a short window where cheating pays and honesty is slow: ideal conditions for self-referrals, coupon stacking and the other fraud patterns that cost you money before anyone notices. Contests are still worth running. The gate just has to be narrower than "anyone with a link", and the score must be something a burst of junk cannot buy.

The leaderboard that rewards volume, and why it loses

The default leaderboard is the one your tooling makes easy: sort by clicks, or signups, or links generated. It has three attractions. The number is easy to compute, it moves fast so the board looks alive, and nobody waits to see their name rise. All three are cosmetic.

• Clicks. The cheapest metric to inflate on the internet. Bot traffic, redirect farms, a link in a channel with 20,000 members and no buyers.
• Signups. Barely better. A trial that never converts drags activation down while looking like progress.
• Posts published. Rewards whoever has the most channels rather than the best one, and punishes the affiliate who writes one thorough comparison piece.
• Reach or impressions. Self reported and unverifiable, so the standings mean nothing.

The deeper problem is that a volume board inverts your economics. You reward the person who produced the most activity rather than the most revenue, and those are frequently not the same person. Deal sites are the obvious example: enormous click volume, honest conversion, and an audience taught to wait for a discount.

A volume leaderboard is defensible in one case only, when volume itself is scarce and the unit cannot be faked, such as newsletter sends from a list you can verify. Otherwise sort by the metric that pays your bills.

Scoring a leaderboard that rewards conversion

The version that works is a weighted score, published in full, where the base unit is a verified paid conversion. It needs to be short enough that an affiliate reads it once and can predict their own behaviour from it.

The weights below are an example to edit. What matters is that they are written down before the start date.

Four details make this score work.

1. Refunds subtract. Ignoring them lets a bad affiliate run up a score and settle later, so subtraction is the cheapest fraud control you will ever build.
2. Renewals are worth less than new customers. Momentum matters, but a contest exists to bring in people who were not there before.
3. The floor is public. Any score above zero gets listed. Being seventh of forty is information. Being unlisted is a rejection.
4. Late conversions have a stated rule. Your cookie window defaults to 90 days, so freeze the board on the last day.

Prize design: multiplier, cash, or software credit

Prize structure follows score design. Once the unit is a new customer, the question is how to reward one in a way that attracts the right entrant.

Cash alone recruits people for whom the prize is the income, so you buy volume from a channel you would never choose organically. A multiplier alone does the opposite, rewarding the people already winning. The structure we keep returning to has three tiers. Everyone above the floor gets the multiplier for one or two billing cycles, which is nearly free because it only costs money when it works. A modest cash pool goes to the top three, smaller than you think, because a prize generous enough to tempt someone with no audience attracts exactly that person. Product shaped rewards go to the middle of the field: a free annual plan, a featured slot in the partner directory, a case study about their workflow.

Then describe the mechanics before day one: whether the multiplier covers renewals, whether it is capped, whether refunds claw back a paid prize, when it lands. A vague prize ends in a dispute, and disputes turn a promotion into a terms argument.

Duration: why three weeks beats a quarter

Duration is a fairness setting disguised as a schedule: the window decides which affiliate types can possibly win.

• One week or less. Only affiliates with a finished asset and an existing audience can compete. New writing has no time to be published, indexed or read.
• Two to four weeks. Long enough for one asset to be written, published, shared and converted, so the widest set of affiliate types has a real shot.
• One quarter. The standings freeze around week two. Leaders coast and the bottom of the field stops checking.
• Anything measured in seasons. Half the participants forget the rules and the other half think you already paid them.

Three weeks with a heartbeat in the middle holds tension without becoming a background task. Day zero: the rules, the formula and the standings page, which removes every excuse to wait. Day seven: a standings update plus the leader's best asset, so the field can see what winning looks like. Day fourteen: each affiliate's own numbers, because personal progress beats the public ranking for the middle of the field. Day eighteen: announce something small they can write about. Day twenty one: final standings, confirmed prizes, winner's playbook.

The contest window and your attribution window are different clocks. A click from the final day can convert after you awarded prizes, so decide the freeze moment in writing and put it in the rules where it will be read in time.

The rules you must publish before day one

A contest without published rules is not a generous gesture, it is a liability with a prize attached. Everything below belongs on one page, linked from your program terms and live before the start date.

• Eligibility. Who can enter, whether employees are excluded, one entry per person, whether new accounts qualify. If your payout rails cannot reach a country, say so before someone there does three weeks of work.
• Qualifying actions. Exactly which events score, and what does not: not clicks, not trials, not free accounts, not refunded orders.
• Self-referrals. Sled already refuses to credit a conversion when the buyer email matches the affiliate email. Say that, then name the surrounding cases: household accounts, disposable cards, a second address on your own link.
• Fraud and consequence. One sentence an affiliate can understand, plus the penalty: forfeited points, removal from the contest, possibly the program.
• Tie handling. A published tie-break order, applied without judgement calls.
• Prize terms. What it is, when it is paid, whether the multiplier covers renewals, what currency, who handles tax. A prize below your minimum, 2,000 cents by default, may sit pending until it clears.
• Close time and time zone. Pick one, including on your dashboard.
• Disputes and changes. How a problem is raised, in what window, and whether you may amend or cancel. Keep that last one narrow, because "we can change anything" makes the page meaningless.

One design rule keeps you clear of a heavier legal regime: do not add chance to the contest. A competition where skill decides the winner under published criteria is a competition. A random draw or a lucky bonus entry is a sweepstakes, with obligations that vary by country. If you want random rewards, run them separately and never let chance decide the standings. This is not legal advice; if the prizes are large or the audience spans countries, have the page reviewed.

Ties, refunds and the edge cases that ruin payout day

The awkward cases are predictable, so decide them in advance rather than at 11pm on closing night.

Ties. Weighted totals tie more often than you expect, because most scores move in whole and half points. Publish a chain: earliest to the final score, then more new customers rather than renewals, then the higher refund adjusted total, then split the prize.

Refunds after the freeze. Say whether a prize survives a later refund and whether future commission absorbs it.

One person, two accounts. A disqualification, not a merge, so a second identity cannot become a bonus entry.

A champion who turns out to be a problem. Say that prizes can be withheld pending a review of qualifying conversions. It is almost never used, and its existence prevents the argument.

The near miss. Someone finishes fourth by half a point holding a screenshot of a checkout that never reaches your conversions. It is the most common dispute, and the conversions page and CSV export settle it. A number beats an argument.

Keeping the losers motivated

Almost everybody loses, and how they lose decides whether you can run this again. The failure mode is common: a winner is announced, a screenshot goes out, and the other thirty nine people who wrote about you for three weeks receive nothing. They do not become better affiliates next quarter. They become quieter.

1. Give everyone above the floor something. A multiplier bump, a free annual plan, a bonus that is small but real. Participation should not be strictly worthless.
2. Publish full standings, not the podium. Somebody in eleventh place learns they were two conversions away. Somebody unranked learns nothing.
3. Send each affiliate their own numbers, not the leaderboard. Their clicks, conversions and score against their own last three months. Most affiliates do not know their own baseline, and the baseline is what they can act on. It is also the best reactivation moment of the year, because affiliates who never promote lack feedback rather than motivation.
4. Rotate the metric. One round on conversions, the next on new content published, the next on activation. Different partner types win different rounds.
5. Share the winner's playbook. With permission, publish what the top three actually did. The field learns more from that than from the prize announcement, and the winner enjoys the credit.

Skip the public loser list and any framing where participants are told they underperformed. Encouragement in public, numbers in private.

Running the whole thing inside Sled

You do not need a spreadsheet circus. The pages that matter are already there.

• The affiliates page is your roster and week one check, with clicks and pending balance side by side, so you can see who has published. An entrant with nothing live by day seven needs the personal nudge, the move that fixes a silent affiliate cohort.
• The conversions page is the scoreboard's source of truth. Filter to the contest window, then by provider, order type and status. Order type matters most, because it separates a new customer from a renewal and your weights depend on that.
• The CSV export at /api/export/conversions is what you hand to a scoring script, read only and safe to audit.
• The payouts page shows whether a cash prize clears your minimum, 2,000 cents by default.
• The share kit at /{tenant}/affiliate/{code}/share gives every entrant a starting point, which turns the contest from a filter for people who already had content into an opportunity for everyone else.
• The read only MCP server is included on every plan, enough to ask an assistant to compute standings without letting it change one.

Two practical notes. Nine real participants test the format better than forty names you cannot follow up with, so start small even though there is no affiliate cap. Sled is 19 dollars a month or 99 a year after a 10-day free trial, and one extra conversion decides it. If your program has barely launched, run the launch playbook first, because a contest on an empty program only advertises emptiness.

FAQ

Will a contest attract affiliates who are only in it for the prize? Yes, always, and the fix is structural. Narrow eligibility, score paid conversions with refunds subtracted, keep the cash pool modest, and make the everyone-above-the-floor multiplier the real reward. Contest hunters optimise for whatever you score.

Should the leaderboard show clicks or conversions? Conversions, with renewals weighted below new customers and refunds subtracted. Clicks are decent coaching material in week one and useless as a ranking, because they are trivially inflated.

How long should an affiliate contest run? Two to four weeks, with three weeks the usual best fit. Shorter favours affiliates who had content ready, longer lets the standings freeze and the leaders coast. Announce one close time in one time zone and freeze the board there.

Do I need a lawyer to run a contest? Not for a modest prize pool among your own approved affiliates with published skill criteria. You do need published eligibility, qualifying actions, fraud consequences and tie-break rules, and you should avoid random draws, because chance moves you into sweepstakes territory. If prizes are large or entrants are international, get the rules reviewed.

Running a contest is a good excuse to find out which affiliates actually convert: start a program on Sled and let the conversions page keep score, or go straight to signup and have your first standings up this week.