How we produce campaign-supporting content

Most agencies running paid promotion for contest entries make the same mistake: they pour budget into placements before anyone has written the words. We did this too, in early 2023, and watched a perfectly good entry get 3,100 impressions and 11 clicks because the creative was a screenshot of a contest page. We changed how we work after that.

This page describes our content production methodology in detail. It is opinionated. Some of the opinions are unpopular.

The 3-asset minimum, and why we refuse to go below it

Every campaign we run starts with three pieces of original creative built specifically for that entry. We call it the 3-asset minimum:

One hero image, produced at two crops (1:1 and 4:5) for feed placements, plus a 9:16 variant for Stories and Reels.
One caption pack of three variants: a short hook (under 90 characters), a medium read (180-220 characters), and a long-form descriptor for landing copy.
One short video, 15 seconds, with on-screen text legible at 320px width and a silent-friendly first three seconds.

Why three, not one? Because ad platforms reward variant testing, and because a single asset cannot serve a feed, a Story, and a search-intent landing page at the same time. Why not seven? Because 47% of our clients are individual contestants paying out of pocket, not brands with creative retainers. Seven assets would price them out. Three is the floor where the math works without compromising the work.

We will not run a campaign on a single static image. If a prospect insists, we point them toward a freelancer.

What the client provides

The split of work matters. We are not a stock photo agency, and we are not your photographer. The client provides the raw material; we shape it into ad-platform-compliant creative.

Here is what we ask for, in order of importance:

The original entry submission. Whatever you uploaded to the contest — the photo, the video, the written pitch, the project page. This is the source of truth.
Two to four supporting photographs. Phone camera quality is fine. We need behind-the-scenes context, not glossy product shots. If your entry is a recipe, we want a picture of the kitchen. If it is a song, we want a photo of the instrument.
A 60-to-90-second voice memo explaining why you entered, in your own words, in your first language. We translate. The memo is for tone, not transcript.
The contest rules link. We read them. Every time. More on that below.
One do-not-mention list. Employer, ex-partner, the school your kid attends — anything you want kept out of the creative.

What we explicitly do not need: a brand book, a Pinterest mood board, or a brief written in marketing language. If you send those, we read them politely and then ask for the voice memo anyway.

What we produce, in concrete terms

The 3-asset minimum expands into a specific deliverable list. Per campaign, the standard package contains:

Three caption variants

Variant A is conversational and first-person. Variant B leads with the stakes of the contest (deadline, prize, what winning means for the entrant). Variant C is question-led and built for comment-bait without resorting to engagement-bait tactics that Meta penalises under its Promotion Guidelines (last reviewed at time of writing in early 2025). We test all three, kill the worst performer after roughly 72 hours, and double down on the winner.

Two image edits

One edit is a colour-graded, crop-adjusted version of the entrant's original photo. The second is a composition with text overlay — the entry's name, the contest's name, and a single clear call to action. We do not generate fake faces, fake hands, or fake scenery with AI image tools. We do use generative fill sparingly for backdrop extension and dust removal, and we disclose that to the client in the asset handoff document.

Copy for three channels

The channels are: paid social ad copy (Meta and TikTok, formatted to each platform's character limits), a 240-character landing page hero block, and a short-form email or messenger blurb the client can send to their own network. We write all three. We do not write spam scripts, and we do not write content meant to be posted from accounts other than the client's own.

One 15-second video

Almost always a vertical 9:16 cut, assembled from the client's own footage and stills, with captions burned in. Why captions burned in? Because Meta's autoplay-muted default means a video without text loses roughly the first second of attention before sound kicks in, and that second is where the decision happens.

The boundaries — what we will not produce

This is the part of our methodology we are most careful about, and the part most agencies refuse to write down.

We will not produce content that misrepresents the entry. That means:

No claims the client did not make in their original submission. If you entered a baking contest with a sourdough, we will not market it as artisanal heritage grain unless that is actually what it is.
No before-and-after edits that change the physical reality of the entry — no slimming filters on people, no colour-doctoring of products to look different from what was submitted.
No fabricated endorsements. No "as featured in" badges unless the feature actually happened, and we will ask for the link.
No "voting urgency" copy invented from thin air. If the contest closes 18 June 2026, we say 18 June 2026; we do not say "ends tonight" on 12 June.
No content impersonating the contest organiser. This is not just unethical, it violates Google's Misrepresentation policy and gets accounts disabled.

We will also not produce content for entries we believe are likely to lose voter trust. This is a judgement call. We have walked away from two projects in the past year where the entry was real but the framing the client wanted was so far from the submission that running ads on it would have felt like a small fraud. We refunded the deposit and explained why. Details on how that works live on our refund page.

Production timeline, honestly

Total turnaround on the 3-asset minimum is 5 to 7 working days from the moment we receive the full input pack. Here is the rough split:

Day 1. Intake call (30 minutes), rules read, do-not-mention list confirmed.
Day 2. Voice memo transcription, first caption draft, image selection.
Day 3. Caption variants A/B/C written, image edits drafted, video storyboard built.
Day 4. Client review round one. We expect changes. We budget two revision rounds into the price.
Day 5-6. Video assembly, final image grading, copy finalised across all three channels.
Day 7. Asset handoff document, including a written log of what was edited, what was AI-assisted, and the contest rules check.

If the contest deadline is sooner than seven days from intake, we say so honestly. We do not take on rush jobs that would force us to skip the rules-read step. A campaign disqualified for rule violation is worse than a campaign that started two days late.

A specific example, anonymised

A client from Wrocław entered a regional craft beer competition in October 2025. Her entry was a smoked porter brewed in a 40-litre batch at home. She came to us with a phone photo of the bottle, a voice memo about her grandfather's brewing notebook, and a deadline of 19 days.

We produced: a hero image showing the bottle with the open notebook in soft natural light (her own kitchen, not a studio); a 15-second video that opened on the notebook page and ended on the pour; three captions in Polish and English; and a single-paragraph landing block for her personal voting link.

The campaign ran for 11 days on Meta with a total ad spend of 320 EUR. She did not win — she placed fourth out of 84 entries. She also picked up 1,247 new Instagram followers for her brewing hobby, which she told us mattered more than the contest itself. We count that as a good outcome. We also count it as a campaign where the creative did its job because we resisted her initial request to add an "award-winning" label that no one had given her yet.

How this fits into the rest of our process

Content production is one stage. It sits between strategy intake and live ad management. The assets we produce here are the assets we run, full stop. We do not swap in stock images mid-campaign, and we do not let the client overwrite captions in the ad manager without a documented review. This is partly quality control, partly compliance: under the EU's Digital Services Act, which has been fully applicable to all platforms since 17 February 2024, traceability of ad creative matters more than it used to.

If you want to see what the assets cost as part of a full campaign, our pricing page has the current numbers. If you want to know how we got here as a small Minsk-based team serving Polish and EU contestants, the about page is honest about that too. And if you have a specific question we have not answered, our FAQ covers the recurring ones.

One last opinion

Good contest creative is not a function of budget. It is a function of how closely the creative matches the actual entry. Slick production on a misrepresented entry tends to underperform a phone photo with an honest caption — we have watched it happen often enough to believe it. The job of the content we produce is to make the entry findable, legible, and worth voting for. Not to make it look like something it is not.

That is the whole methodology, written down.