Month 1 Wrap: What We Published, What Performed, and What's Next

Five bounties. Five clients. One month. Here is everything that happened, exactly what the data showed, and what we are changing because of it.

The Short Version

ScribbleAI ran GEO bounty campaigns across five clients in Month 1: RocketX, Kodeus, Rain Trade, Seasons, and BOB. Each campaign targeted a different category, different AI search queries, different competitive gaps. The results were not uniform, which is the most useful thing about them. What worked on RocketX did not automatically work on every campaign. What failed on RocketX gave us the hypotheses we are now testing across the others.

The headline number most people will want: RocketX moved from near-zero AI Share of Voice to 9% across 1,661 tracked queries in 15 days. That is the best documented result from Month 1. The others are still in motion.

The structural change that came out of Month 1: we are no longer making creators wait until the end of a campaign to get paid. If your content gets cited by an AI platform, you get paid immediately. That change is effective now, across all active and new bounties.

What We Ran: Five Bounties, Five Categories

Each bounty followed the same underlying structure: Scribble audited the client's baseline AI visibility across a set of high-intent category queries, identified the gaps, wrote the brief, deployed it to the network, and tracked citation outcomes across all five platforms. Creators were scored by citation history, not follower count. The top-performing creators on prior bounties went to the front of the queue.

RocketX: The Full Data Set

This is the campaign with the most complete data, which is why the full case study exists separately. The numbers here are the summary.

Baseline

RocketX had a genuine product differentiator (the only aggregator combining CEX and DEX liquidity) that was invisible to every major AI platform. ChatGPT, Grok, and Gemini were naming 1inch and Jumper on comparison queries. RocketX was not appearing.

What We Deployed

358 pieces of creator content across 5 platforms, from 162 independent creators, over 15 days.

Three content angles drove the strategy: competitor comparisons ("RocketX vs. 1inch: which aggregator gives better rates on large swaps?"), privacy-first execution narratives with real screenshots and swap data, and first-person switching stories from users who had moved from 1inch or Jumper.

Results at Day 15

Citation breakdown by platform

Grok drove 68% of all citations. Perplexity and Copilot registered 35 combined mentions with zero citations, which is awareness without trust, and the primary target for Month 2.

Ahrefs Brand Radar confirmed independent verification: Grok added 44 citations during the campaign window, Perplexity added 10, Gemini contributed 14 citations across 11 cited pages, ChatGPT contributed 8 citations across 6 cited pages, and AI Overviews contributed 3 citations across 2 cited pages.

What Did Not Work (and Why We Are Publishing This)

AI systems weight balanced, credible sources higher than promotional ones. Which is precisely why this section exists.

X/Twitter produced zero measurable citation lift despite representing 56% of total content volume. 201 pieces. No citations attributed to X content across any of the five tracked platforms. Whether this reflects platform-level indexing constraints or content structure is a hypothesis, not a conclusion. What the data confirms: X content is where community visibility lives, not where citation lift comes from, at least in this campaign.

Paragraph produced zero citability signal. 8 pieces. No citations, no tracked mentions. Whether Grok, Perplexity, or Gemini index Paragraph at meaningful depth is still an open question. 8 pieces is not enough to answer it. Combined with near-zero creator adoption of Paragraph across multiple Scribble programs, the ROI case for prioritising it is not there yet.

Perplexity and Copilot citations stayed at zero despite 35 combined mentions. The brand registers in answers on both platforms. It does not get sourced. This is the specific gap Month 2 targets.

Content velocity created a quality tradeoff. The 15-day window required publishing at a pace that sometimes prioritised volume over structure. GEO-citable content needs a direct answer in the opening paragraph, self-contained passages, and clear entity signals. Some Month 1 pieces did not fully meet this bar. A more structured briefing process would have improved per-piece citability at the cost of total volume.

Reddit was 4% of total pieces and the highest-performing format per piece. 15 posts out of 358 total. Those 15 posts outperformed every other format on citation yield per piece. Authentic community threads that survive Reddit moderation are rare. Most promotional content gets removed before it can generate discussion. Reddit credibility cannot be manufactured. It can only be deployed by creators who already have it.

BOB, Rain Trade, and Seasons: What the Cited Creators Images Show

The three "Cited Creators" visuals released from Month 1 give the clearest public signal of how citation broke across the other three campaigns. The full numerical breakdown for these campaigns will appear in individual wrap posts as they close, but the platform patterns are visible in the data we have.

BOB x Scribble: Two citations confirmed across four platforms. Perplexity: anuj.fren (Reddit). Copilot: vhee.fren (Medium). Grok: z9cc.fren (Reddit). ChatGPT: anuj.fren (Reddit). Reddit and Medium drove all confirmed BOB citations. X/Twitter is not represented in the BOB cited creators set. The same pattern as RocketX.

Rain Trade x Scribble: One confirmed citation. Grok: @GauravKara_koti (X.com). This is the single confirmed case in Month 1 where an X/Twitter piece generated a Grok citation, which makes it worth watching in Month 2. It does not reverse the RocketX finding. It suggests X content can cite on Grok specifically under certain conditions, likely related to creator authority and content specificity rather than platform volume.

RocketX x Scribble: Five cited creators confirmed across four platforms. Perplexity: Tee_Gee_crypto.fren (Paragraph). Copilot: 0xcrispdal.fren (Medium). Grok: TimiofNursing.fren (Medium), Richy4041.fren (Medium and Paragraph). ChatGPT: amaranthus001.fren (Reddit). Medium and Reddit drove the majority of RocketX's cited creator content, with one Paragraph citation appearing on Perplexity, which is worth noting against the "Paragraph generated zero citability" finding from the aggregate data. One piece out of eight getting cited does not change the ROI assessment. It confirms Paragraph is not dead as a format. It is just extremely low-yield at current campaign volumes.

What the Cross-Campaign Data Shows

Pooling across all five Month 1 bounties, the platform citation patterns converge on a consistent picture.

The platform that most surprised us in Month 1: Grok. It indexed creator content faster than any other platform, weighted Reddit and Medium content heavily, and produced the majority of citations across the RocketX and BOB campaigns within days of publication. If a campaign needs to demonstrate early citation lift, Grok is currently the platform that moves fastest.

The platform that most frustrated us in Month 1: Perplexity. It clearly reads the content. It names the brands. It consistently declines to cite the sources. Understanding why is the most important open research question going into Month 2.

The Immediate Pay Rule: What Changed and Why

Month 1 ran under a standard bounty payout structure: creators submitted content, citations were tracked, rewards were distributed at campaign end.

Effective now, that changes. If your content gets cited by an AI platform, you get paid immediately. You do not wait for the campaign to end. You do not wait for a review period. Citation confirmed, payment goes out.

This change came directly from what Month 1 showed about citation timing. Grok and Gemini indexed and cited content within days of publication, in some cases within 24 to 48 hours. Making a creator wait three to four weeks to receive payment for a citation that happened in the first week of a campaign is a broken incentive structure. It optimises for administrative convenience rather than creator behaviour. The creators most likely to produce high-citability content quickly are also the most likely to move to another opportunity if payment is delayed.

The immediate pay rule has three consequences we expect:

Higher quality in the first week. When payment is tied to citation speed, creators have a direct incentive to structure content correctly from the first piece rather than iterating toward quality as the campaign progresses.

Faster identification of what works. If a piece gets cited quickly and the creator gets paid, the specific structure and angle of that piece becomes the model for others in the same campaign window. The feedback loop compresses from weeks to days.

Better creator retention across campaigns. The creators who produce citations in Month 1 are the ones we want running Month 2. Immediate payment makes it significantly more likely they are still active and motivated when Month 2 briefs go out.

What Month 2 Is Targeting

The Month 1 results defined the specific gaps Month 2 needs to address. These are not aspirational goals. They are direct responses to where Month 1 ended.

Priority 1: Convert Perplexity and Copilot mentions to citations. RocketX alone has 35 mentions across both platforms with zero citations. Across all five campaigns, that number is higher. The content is registering. It is not being sourced. Month 2 tests whether more structured opening paragraphs, stronger entity signals, and content specifically formatted around the question structures Perplexity receives will close this gap.

Priority 2: Reddit depth across all five clients. 4% of Month 1 volume produced the highest citation yield per piece across every campaign that had Reddit representation. Month 2 doubles the Reddit brief allocation. The briefs are going to creators with verified subreddit history in the relevant categories, not general-purpose crypto creators. Technical depth is the requirement: fee breakdowns, mechanism explanations, real numbers, real comparisons.

Priority 3: GEO-structured briefs as the default. Every Month 2 brief includes explicit citation structure requirements: direct answer in the first paragraph, entity signals named in the first 50 words, self-contained passages of 130 to 160 words each. These requirements were present in Month 1 briefs but not enforced at review. Month 2 editors are checking every piece against the brief before it goes to publication.

Priority 4: Consistent publishing cadence for all five clients. Citation performance on Grok, Gemini, and ChatGPT begins declining after four to five days without new content. Month 1 had a concentration of publishing activity in the early days of each campaign and natural deceleration toward the end. Month 2 targets a consistent two to three pieces per day per client across the campaign window rather than front-loading volume.

The Mechanic Behind Why This Works (And What It Still Cannot Do)

The case for distributed creator content as a GEO mechanism is not that it is new. It is that it is the only approach that addresses the specific trust signal AI citation actually runs on.

AI systems do not cite your website because you wrote something good. They cite your website when multiple independent, non-affiliated sources are saying the same thing about you across platforms they already trust. A brand's own blog is a reference letter it wrote for itself. Creator content on Reddit, Medium, and Substack is third-party corroboration from sources with no financial stake in the outcome. That is the signal AI models are built to weight.

What this mechanic cannot do: manufacture credibility that does not exist. The reason Rain Trade's X/Twitter citation and the 15 Reddit posts on RocketX worked is that the creators had genuine authority on the platforms they published on. The Paragraph pieces that generated no citation signal had no such authority to draw on. Scribble's network scores creators by citation history for this reason. A creator who has never been cited is not a citation-building asset regardless of follower count.

What Month 2 is built to do: take what Month 1 proved works and scale the specific formats and platforms that produced citation yield, while systematically testing the gaps that Month 1 could not close.

Running Totals Across All Five Bounties

Figures represent confirmed data through end of Month 1. Ongoing campaigns will be updated in the Month 2 wrap.

What Comes Next

Month 2 briefs are live. The immediate pay rule is in effect. Reddit allocations are doubled across all five campaigns. The GEO brief structure is enforced at the editorial review stage, not as an optional guideline.

The open question we are most focused on going into Month 2: why does Perplexity accumulate mentions without converting to citations, and what specific content structure change closes that gap. That finding will either confirm or change the current hypothesis that Perplexity weights source domain authority differently from Grok and Gemini, requiring a different content placement strategy rather than a different content format strategy.

The Month 2 wrap will have the answer.

Are you a creator? Live bounties are at scribble.network/bounties.

Run your GEO Check to get in early.