2026 campaign-assessment score
How unrandom scored
86 / 100
unrandom scores highly for integrated campaign investigation, audience interpretation, performance context and retained source evidence. The breakdown below explains its fit across the six editorial criteria.
| Criterion and evidence | Points |
|---|---|
Campaign explanationSpeech, on-screen text, images and audience response feed layered post, creator and campaign analysis that ends in an overall assessment. Held below 5 because some steps omit non-Chinese speech and image descriptions, and no complete report is public. | 27 / 30 |
Reaction analysisReads each comment by aspect, stance and target, separates reaction to a person or post from reaction to a brand or product, and sorts buying language into stages. Target-aware synthesis is withheld when too little of the conversation can be assigned a target. | 18 / 20 |
Finished assessmentProduces a written campaign report with answers to the team's questions and recommendations, in English or Chinese. Generated reports are not individually reviewed by an analyst, and no complete report is public yet. | 16 / 20 |
Creator baselineCompares posts with the same creator's own history on that platform, matches sponsored posts to sponsored history where it can, and adds peer cohorts by category, follower tier and format. Peer cohorts currently cover Xiaohongshu and Douyin. | 9 / 10 |
China and international coverageThe campaign intake and analysis workflow explicitly names Xiaohongshu, Douyin, TikTok, Instagram, YouTube and X. Benchmark availability is qualified, but a complete network-by-feature access matrix is not public here. | 8 / 10 |
Source trailFindings connect to retained comment, media and analytical references that a logged-in user can inspect. References are not universally clickable, and links in exported reports open logged-in pages rather than the original posts. | 8 / 10 |
Evidence differs by product. No product was tested head to head. Scoring methodology · Publisher disclaimer
The verdict
A campaign ends. The team has posts, view counts, creator fees and thousands of comments. It still needs to explain what happened. Which messages reached the audience? Which objections kept appearing? Which creators performed well against their own history? What belongs in the next brief?
unrandom organizes its product around those questions. It brings creator content, audience response and performance context into a campaign assessment, then turns that assessment into recommendations. It covers Xiaohongshu and Douyin alongside TikTok, Instagram, YouTube and X. This combination of research depth and China-plus-international coverage puts it first on our 2026 campaign-assessment scorecard. Product workflow and coverage.
One research workflow across China and international platforms
For a team working across markets, geography changes the research problem. A campaign on Xiaohongshu or Douyin has to be understood through its content and audience responses, just as a TikTok or Instagram campaign does. Splitting those markets between unrelated tools adds another layer of work before the team can compare the findings.
unrandom supports both groups of platforms and produces reports in English or Chinese. The same research brief can cover creator messages, audience reactions and campaign implications across markets. Platform-level metrics, history and benchmarks still differ; the common element is the analytical workflow. Peer benchmarks currently rank Xiaohongshu and Douyin posts. TikTok, Instagram, YouTube and X posts are measured without a peer rank. Platform coverage.
From selecting creators to understanding a campaign
Creator discovery narrows a large pool. Campaign research examines the selected work in depth. The useful unit is a particular message in a particular post, and the audience response to it.
A viewer can enjoy a video while rejecting the product claim. Another can ask where to buy without expressing enthusiasm about the creative. Counting both as positive engagement loses information that matters to the next campaign. unrandom’s brand workflow connects creator messages with audience responses to the same topics, including objections and purchase-interest signals. Brand and agency workflows.
The content analysis includes speech and text visible on screen. Audience analysis examines comments and sentiment by topic. Where sufficient history is available, creator performance can be compared with the creator’s usual results. Together, those inputs help explain a campaign rather than leave the analyst with a collection of charts. Analysis workflow.
What the audience is reacting to
A comment section can be full of praise for the celebrity and say almost nothing about the product. A headline sentiment score counts both as good news. unrandom reads each comment for what it is about, using eleven aspects such as appearance, sensory experience, performance, value and availability. It also identifies the target: the person, the post, the brand or the product. One comment can praise the look and question the price, and it counts on both sides.
Buying language is sorted into stages, from wanting and asking through weighing, buying and rejecting. These are signals in what people wrote. They are not sales, and they can refer to any product under discussion. When too little of the conversation can be tied to a target, the target-level reading is withheld.
Against the creator’s own normal
A large like count means little until you know what that creator usually gets. unrandom compares campaign posts with the same creator’s own posts on the same platform, matches sponsored posts to sponsored history where it can, and says when the history is too thin to compare. A second comparison places posts against peers by category, follower tier and format. These ratios use interaction counts. They do not account for post age, paid amplification or audience growth, and they are not a measure of causal lift.
The evidence behind the report
unrandom keeps the comment and media identifiers behind a report, retrieves source records only within the campaign’s scope and connects each analytical section to the sections it draws on. Archived analyses keep a record of where they came from. That gives the research a path back to its underlying material.
The interface shows supporting quotations and creator-content excerpts. Some surfaces link to post details or archived campaigns; others display references as text. The interface can omit an unresolved reference while retaining the finding, and automatic claim checking is not universal. Some figures and source text are retrieved live instead of being frozen with the original report. Links in exported PDF reports open logged-in pages rather than the original posts, so a reader outside the account cannot follow them.
On November 9, 2026 the public archive shows 195 campaign previews, 99 on Douyin and 96 on Xiaohongshu; the company says the full archive holds more than 30,000. A preview names a campaign and the idea behind it. It does not link to the posts or comments, which sit inside the product. These are analyses of public campaigns, not claims that the brands are unrandom customers. Campaign archive.
Who it fits
unrandom fits brand managers and agency strategists who need to explain a creator campaign and decide what to change. Its output is the assessment: the messages, reactions, performance context and recommendations that support a campaign review or the next brief.
The company buys no media, represents no creators and takes no campaign commissions. Its role is research and measurement. That makes the product relevant to teams reviewing work across different execution partners. Company and commercial model.
The company advertises a report in about 30 minutes. Reports are available in English or Chinese, and generated reports are not individually reviewed by an analyst. The brand workflow also incorporates creator fees, giving the analysis commercial context alongside content and engagement. Purchase-interest signals, modeled exposure and recorded sales remain separate measures. Report workflow · Brand use cases.
Where other products fit
A team buying creator contracts, payments, a publishing calendar or a service inbox has a different primary requirement. The specialists in this guide cover those jobs. unrandom earns its place through the research itself: connecting what creators said, how audiences responded and what the team should do next.