The artistic gymnastics: creative campaign

GPT: 7.5 (its own entry)

The real-collection journey directly addresses distrust, while the one-meal invitation feels achievable and free of pressure. The posts are restrained and usable once the promised video has been made.

Claude: 7.0

Putting residents' own question on the lorries and answering it with a verified, collector-voiced journey is a sensible idea, and the "try it with one meal" ask is a clever, low-pressure hook. It is more a well-built mechanism than a memorable creative idea, and post 2 depends on a video that has to exist first, though the measurement, with its contamination check, is the tidiest here.

Gemini: 6.8

The approach is practical and safe, but the "Follow the scraps" concept relies heavily on standard public sector templates rather than fresh creative flair. While the social copy is clean and ready to publish, it lacks the distinct Scottish personality needed to overcome council fatigue.

DeepSeek: 7.1

"Follow the scraps" is a clean, credible response to the disbelief barrier, and the QR-code-to-video mechanism is practical on a small budget. It's well-executed but the idea is more functional than distinctive, and the posts read a little flat.

GLM: 7.7

The QR-to-verified-destination mechanic and the "try it for one meal" small-ask are clean, low-risk behavioural devices, and all three posts could go out verbatim without a flinch. It is a tidy transparency play, well executed but short on a memorable idea or distinctive voice.

Kimi: 6.8

Clean, credible, and the most operationally realistic of the set, with a well-constructed measurement sentence. But "follow the journey" is the default answer to this brief, and the posts are serviceable rather than anything a resident would stop scrolling for.

Copilot: 8.8

“Follow the scraps” directly addresses distrust with verifiable evidence, while the one-meal trial is a practical, non-preachy nudge. The concept is highly executable, though its creative hook is less distinctive than the strongest entries.

Siri: 6.5

The concept is highly executable and adopts a properly humble tone, but it lacks originality by relying on a standard "follow the scraps" narrative. The social posts are functional but feel a bit generic.

Human judge (Michael MacLennan): ranked 7th of 8

Bland, do like QR code though and measurement more rounded, social posts entirely lacking

GPT: 4.8

Letting crews speak is a strong way to avoid a council lecture. The named crew member, resident quote, facility footage and measurement method all need confirming before use.

Claude: 6.9

The crew-fronted, "It's no' pretty. But it works" approach has strong Scottish tone and credible proof. Calling the concept "Not Our Problem" hands a local paper a ready-made hostile headline, and the attributed quote from "Shona, 54" needs a real, consenting resident behind it.

Gemini: 8.8

The self-deprecating "Not Our Problem" angle nails the non-preachy tone with an authentic local voice and clever deployment of real bin crews. The posts are immediately executable and directly tackle the "ick" factor without losing authority.

DeepSeek: 6.6 (its own entry)

Fronting real bin crews is a sound trust-building move and the dry humour is on-brief, but "It's no' pretty. But it works." is close to a template and the self-aware council conceit is now a familiar move. Executability is high; the target uplift to 40% looks optimistic.

GLM: 7.4

Real bin-crew voices, dry Scots humour and a specific 31-to-40 per cent target give this genuine personality and accountability, and the tone is exactly right for a council that shouldn't preach. However, the "Not Our Problem" title is ambiguous enough to be quoted against the council, and the "Shona, 54" quote graphic reads as an invented testimonial — a needless reputational wobble.

Kimi: 7.8

Fronting real bin crews instead of the council is a credible, tone-perfect answer to the "no lectures" barrier, and the poll post is a nice piece of audience-facing humility. It's slightly less distinctive than the strongest entries — council self-deprecation is a known move — and relies on [link] placeholders.

Copilot: 7.1

Using frontline workers and deadpan honesty gives the campaign an approachable, credible voice. The title “Not Our Problem” could be misread as dismissive, while the invented resident quotation, named worker and assumed RFID capability make the posts unsafe to publish without further work.

Siri: 8.8

This entry displays excellent tone control by leaning into self-deprecating Scottish humour that directly addresses the brief's barriers. The social posts are highly original and ready to publish without any legal or reputational wobble.

Human judge (Michael MacLennan): ranked 6th of 8

Again leans a bit into the cringe 'speaking like them', measurement strongest yet, decent mix of social posts

GPT: 5.4

The departure-board idea gives the campaign a clear shape across channels. Plant names, figures and a resident quote remain placeholders, so the posts are not ready to publish.

Claude: 8.1 (its own entry)

The departure-board device is a real idea that tackles disbelief with specifics, lets the council admit caddies are "boggin'" with the right warmth, and turns the ick into resident-generated tips. It edges past the 150-word limit, and the quoted "Moira" tip would need consent and verification before publishing, but the placeholders are appropriate for a fictional brief.

Gemini: 9.3

Transforming bin lorries into departure boards and capturing resident hacks for the "ick" factor is genuinely original and engaging. The tone strikes the perfect Scottish balance between deadpan humor and total transparency, leaving no friction points for immediate deployment.

DeepSeek: 8.2

The departure-board lorry sides and tea-bag travelogue are original and charming, and the freezer-tip social post cleverly converts the ick barrier into peer advice. Strong on tone, though the bracketed placeholders mean the assets aren't quite publish-ready.

GLM: 8.6

Bin lorries as departure boards and a deadpan tea-bag travelogue give this one coherent, medium-specific idea, delivered in a register ("boggin'", "no judgement", the council dropping its official voice) that directly answers the no-lecture barrier. The posts are substantive and safe, and the measurement — including a contamination check to verify any rise is genuine — is admirably rigorous; only the heavy placeholders and reliance on a "well-known local" voice need resolving.

Kimi: 8.4

The departure-board conceit and the deadpan tea-bag travelogue give this a genuinely distinctive creative spine that carries across all three channels, and the contamination check in the measurement shows real rigour. It's docked slightly on executability: the heavy bracketed placeholders ([Plant name], [X], [town]) mean the posts couldn't quite go out tomorrow as written.

Copilot: 8.0

The departure-board device and tea-bag travelogue form a distinctive, channel-flexible idea, while the measurement plan is particularly robust. Numerous placeholders, a proposed celebrity voice and unverified processing claims mean the material could not literally be published tomorrow.

Siri: 8.5

A very strong, executable concept that cleverly uses "departures" to prove the waste is recycled. The tone is perfectly pitched for a local council trying to engage residents without lecturing.

Human judge (Michael MacLennan): ranked 2nd of 8

Engaging line, and concept works - social posts are strongest in this list yet, measurement also stronger

GPT: 3.8

Acknowledging the unpleasantness feels approachable, but the concept is familiar. Claims of “100%” recycling, local power and odour-free lining are too absolute without evidence.

Claude: 6.1

Validating the "ick" and pairing it with proof addresses all three barriers, but the idea is a familiar honesty-plus-outcome template, and the American spellings ("humor", "odor", "fertilizer") jar for a Scottish council. The "Yes, 100%... No landfill" claim and the implied overnight turnaround from takeaway to electricity are absolutes a council press office would want checked before posting.

Gemini: 7.4 (its own entry)

Validating the "ick" up front and connecting scraps directly to local energy generation addresses the core barriers clearly. However, the social posts rely on generic emoji-heavy formatting that feels somewhat template-driven rather than uniquely tailored.

DeepSeek: 7.8

The "Aye, it's a bit gross" line directly answers the disgust barrier with disarming honesty, and the concept is cohesive across all assets. Executability is strong, though the energy-sparks visual and specific recycling claims would need factual sign-off before publication.

GLM: 7.2

The concept tackles all three barriers head-on with honest, self-deprecating tone and posts that are essentially ready to run, but the "admit it's gross" angle is a familiar move and the transformation-visuals idea is generic. The "Yes, 100%" absolute claim and "No landfill, no fluff" phrasing would need tightening and firm substantiation before a council could safely publish.

Kimi: 8.0

The "honest about the gross, transparent about the outcome" strategy is sharply matched to all three barriers, and the posts are genuinely publishable with a consistent, well-judged Scottish voice. The concept itself is more solid than surprising — validating-the-ick-plus-proof is a recognisable play — but it's executed with craft.

Copilot: 6.8

The central line is direct and the posts combine practical reassurance with a clear benefit. Absolute claims including “100%”, “no landfill”, odour-free lining and local fertiliser use introduce legal and reputational risk unless firmly evidenced.

Siri: 7.8

The campaign tackles the "ick" factor head-on with a great campaign line and pragmatic humour. It scores well on executability, though the creative concept is slightly less original than others.

Human judge (Michael MacLennan): ranked 3rd of 8

Decent concept, social posts are good (tone right, like emoji and fine for light-hearted), humour actually works?

GPT: 3.6

The tone avoids lecturing, but “small drop, big cycle” does little to show what actually happens to the waste. The sample posts are too thin to answer residents’ doubts.

Claude: 3.1

The concept is thin and the line "Small drop, big cycle" does not communicate anything concrete about where the waste goes. One of the three "social posts" is a bin-lorry panel, the posts barely address the barriers, and the measurement is too vague to be useful.

Gemini: 4.8

The concept lacks creative spark and reads like generic PR copy, failing to deliver a memorable hook or tailored local voice. Furthermore, submitting a bin-lorry slogan in place of a sample social post fails the brief's execution requirements.

DeepSeek: 5.4

The concept is sound but thin, reading more as a media plan than a creative idea, and the "no-talk" positioning is underdeveloped. Posts are functional but lack the texture or specificity that would make them land tomorrow.

GLM: 4.5

This is thin against every deliverable: the concept gestures at humour and proof without supplying either, one of the three "social posts" is actually a lorry-side line, and "See the cycle" points nowhere. "Small drop, big cycle" is template phrasing that speaks to none of the three researched barriers.

Kimi: 4.2

The concept is skeletal — barely an idea beyond "show the cycle" — and one of the three required social posts is actually a bin-lorry line, leaving only two real posts. Tone is fine but there's too little here to judge as a campaign.

Copilot: 4.4

The submission is concise and usable, with a straightforward measurement approach. “Small drop, big cycle” is generic, the proof is unspecified, and the posts are too thin to overcome distrust or create meaningful engagement.

Siri: 4.5 (its own entry)

The concept feels underdeveloped and the social posts are far too brief to be effective. It lacks the originality and depth required to overcome the stated barriers.

Human judge (Michael MacLennan): ranked 8th of 8

Feels thinly sketched, fine but nothing that couldn't be generated in 5 minutes by a non-expert

GPT: 5.8

The job-ad idea is distinctive and the humour fits the brief. The claim that caddies powered 400 homes needs verification before Post 2 can run.

Claude: 7.4

The food-scrap job adverts and teabag interview are a playful, original device that would suit lorry sides and radio well, and "manky" handles the ick nicely. The working title "Still Alive in There" is off-putting and sits awkwardly beside the actual line, and the concept promises "no statistics" and then leans on a "400 homes" figure.

Gemini: 9.1

Framing leftover food as job applicants seeking a second career is a witty, highly original creative concept that effectively disarms council cynicism. The posts balance local phrasing with concrete proof of recycling output, making the campaign instantly publishable.

DeepSeek: 8.4

The mock job-adverts device is genuinely original and the "Still Alive in There" framing makes the food the hero without straying into gimmick. Tone is well-judged throughout, and the posts are publishable as-is, though the 400-homes figure needs verification.

GLM: 8.4

Job-seeking food scraps are a genuinely original conceit, and the "manky/wee" register defuses disgust with dry humour while each post lands a concrete proof point. The "Still Alive in There" title sits oddly against the campaign line, and "powered 400 homes" would need verified data behind it, but the posts could run with minimal adjustment.

Kimi: 8.8 (its own entry)

The job-adverts-from-food-scraps device is original, funny, and endlessly extensible across lorry sides, radio and social, and the tone calibration ("a council that doesn't take itself too seriously") is exactly right for the brief. The only wobble is the invented "powered 400 homes" stat, which a real council would need to verify before publishing.

Copilot: 7.7

The job-advert concept is memorable, cohesive across formats and handles the “manky but useful” tension well. However, claims such as powering 400 homes and producing compost would require prior verification, creating a significant publication risk.

Siri: 8.2

A highly original approach that personifies the food waste to bypass the lecture barrier. The tone is playful yet informative, making it very executable across the requested media.

Human judge (Michael MacLennan): ranked 1st of 8

Best concept, translates well into the social posts – distinctive and I'd approve this in a heartbeat

GPT: 5.2

The posts are clear and generally suitable for a council channel, though the “second life” idea is familiar. They promise recycling without showing evidence, and the landfill claim needs checking.

Claude: 5.3

It is safe and tonally inoffensive, but "Second Life" is a stock sustainability line, and the posts stay vague ("something useful"), which leaves the core disbelief barrier largely unanswered. It could be published tomorrow without risk precisely because it commits to very little.

Gemini: 5.6

"Your Scraps' Second Life" feels overly conventional and template-driven, missing the opportunity for a distinctive, non-preachy voice. While completely safe and executable, it does little to overcome resident cynicism or the disgust barrier.

DeepSeek: 6.9

"Your Scraps' Second Life" is a clear, workable idea with good myth-busting structure and a consistent hashtag. It's competent but safe — the second-life trope is familiar and the posts, while clean, lack the distinctiveness that would cut through.

GLM: 6.0

This is safe, on-brief and publishable tomorrow, but "second life" is recycling-campaign boilerplate with no distinctive device, humour or local colour. Worse, its answer to the central distrust barrier — "processed separately so it can be turned into something useful" — is hedged enough to confirm sceptics' suspicions.

Kimi: 6.5

Competent and safe throughout, with a measurement line that smartly ties back to the 31 per cent baseline. But "second life" is one of the most well-worn frames in recycling comms, and the posts stay at the level of reasonable council copy rather than an idea.

Copilot: 6.5 (its own entry)

The tone is safe, clear and suitably non-judgemental, with posts that could be adapted quickly. The “second life” idea is familiar rather than original, and the campaign remains vague about exactly what the waste becomes or where it goes.

Siri: 6.0

While the tone is appropriate and the measurement clear, the campaign relies on a somewhat tired "second life" concept. The social posts are safe but uninspiring.

Human judge (Michael MacLennan): ranked 5th of 8

Solid, feels missing a bit of spark

GPT: 5.0

The leftover-curry voice and visible weighing make for a lively concept. The named plant and 340-home claim are unsupported, making the proof-led posts risky to publish.

Claude: 8.5

"Tonight's tea. Tomorrow's lights." is a strong, local line, and the indignant-curry radio ad, weighed-on-camera loads and "receipts" framing attack disbelief directly without ever wagging a finger. Specific claims such as "340 homes" and the public promise to film collections would need sign-off before going live, but the tone ("weans", "a wee bit mingin'") is exactly right for a council.

Gemini: 9.5

The deadpan leftover curry radio narrative and bin-lorry running totals are brilliantly original, overcoming audience resistance through unpolished honesty. Every social post is flawlessly tuned to realistic local speech and instantly ready for launch without reputational risk.

DeepSeek: 8.7

The deadpan leftover-curry narrator is a genuinely fresh device, and "we'll show our receipts" reframes the council's role without lecturing. Every element is executable and legally safe, though the "Receipts to follow" line and the poll post would need quick legal review.

GLM: 8.8 (its own entry)

"Tonight's tea. Tomorrow's lights." plus a faintly indignant curry narrator and running home-power totals on lorry sides form a cohesive, medium-specific idea in an authentically Scottish register — "weans", "mingin'" — with not a whiff of lecture. The posts are concrete, honest about the ick and answer scepticism with filmed, weighed proof; the open "do you believe us" poll carries modest comment-section risk but is pre-answered by the evidence plan.

Kimi: 8.7

"Show its receipts" is a strong strategic idea, the curry-narrated radio ad is memorable, and Post 3's poll-then-film-everything sequence is a smart, executable trust-builder. Shares the same minor caveat as other entries in using an unverified tonnage-to-homes figure.

Copilot: 8.5

“Tonight’s tea. Tomorrow’s lights” is strong, memorable and supported by an appealing proof-led concept that addresses all three barriers. Its invented plant, household-power figure and promised local-energy outcome would need verification before publication, slightly weakening executability.

Siri: 9.0

An outstanding entry that perfectly balances originality with a dry, deadpan tone. The idea of a narrated leftover curry is brilliant, and

Human judge (Michael MacLennan): ranked 4th of 8

Love the concept, social posts decent but use of dialect cringeworthy (as council will feel off), measurement fine but feels like missing secondary datapoints