Libaorui
Proposed endurance-nutrition positioning; actual identity, current label, official SKU and purchase URL require confirmation.
- Inspectable evidence
- 7 rapidly appraised human studies
- Purchase URL
- Not supplied; purchase link not activated
LIBAORUI · PROJECT CASE / UPDATED 6 OCTOBER 2026
Use evidence to inform purchases. Test the sales effect.
Sports-nutrition project: inspect actual evidence, planned purchase/partner materials and protocols for exposure, purchasing and long-term brand trust.
COMMERCIAL LOOP · EOS PROVIDES TRUSTWORTHY LIVING EVIDENCE
Make evidence a living asset for discovery, review and purchasing.
Platform strategy informs the questions and access routes; evidence still determines scientific findings. EOS produces, appraises, reviews, traces and updates evidence, collaborating with clients/platforms on publication, products and purchasing.
- EOS + client
Global agent strategy
Review markets, languages, buying contexts and platform mechanisms; deliver a matrix and fixed query set.
- Client + EOS
Product / claim definition
Client confirms SKU, label, formula, permissible wording and product URL; EOS registers gaps.
- CORE EOS DELIVERY
Trustworthy evidence
Appraisal or SR/Meta with sources, doses, populations, outcomes and limits, plus scoped professional review.
- Research team / client / EOS
Publication / public sources
Select a route for the scientific manuscript; separately publish confirmed HTML evidence and version records.
- Merchant / platform / EOS
Product / tool integration
Map eligible product feeds; APIs/MCP serve agents explicitly connected to them.
- EOS analysis + client data
Discovery / purchasing
Retain answers/citations and connect product clicks, enquiries, orders/refunds and concurrent tests.
- CORE EOS ONGOING DELIVERY
Feedback / living updates
New research, formulation changes or errors trigger reappraisal, material/version updates and retesting.
The loop returns to evidence: discovery, citations and feedback identify meaningful evidence gaps. Commercial changes do not alter medical findings; record platform and journal reviews separately.
DELIVERED S01 · OFFICIAL MECHANISM REVIEW / 6 OCTOBER 2026
Major global agents: retrieval, display and matching EOS assets.
Eight selected platforms, not every global agent. Developer APIs, consumer search and merchant integration are distinct. EOS assets are delivery recommendations inferred from public mechanisms, not platform endorsements.
| Platform / scope | Retrieval route | Display route | Matching EOS delivery | Conditions / unknowns |
|---|---|---|---|---|
| ChatGPTInternational · search / product feed | OAI-SearchBot supports search; GPTBot training controls are separate. Product feeds require partner approval. | Search sources and shopping data use distinct entry paths and specifications. | Public evidence page and accurate product data; prepare feeds within approved scope. | Medical-evidence weights are not public; do not copy other platforms’ PDF/FAQ fields. |
| Google AI SearchInternational · Search / Merchant Center | Retrieves from Search indexes, with related query expansion; verify AI-display eligibility. | Source links and product data; Merchant supports document_link and Q&A attributes. | Readable evidence pages; map verified product PDFs/FAQs to documented fields. | Documents must be public and product-specific; submission does not guarantee citation. |
| Bing / CopilotInternational · search / merchant catalogue | Copilot Search grounds answers in web sources; products also have merchant-catalogue paths. | Traceable source links; verify merchant conditions for products and transactions. | Public evidence sources, product records and merchant-catalogue mapping. | Assess retrieval, product discovery and checkout separately; no guaranteed placement. |
| PerplexityInternational · consumer product / Search API separately | Search API has its own retrieval architecture, distinct from full consumer ranking. | Check consumer shopping against current Instant Buy availability. | Public sourced evidence pages; separately test answer citations and product redirects. | Client eligibility unverified; evaluate consumer and API routes separately. |
| 百度 AI 搜索China · Qianfan developer capability | Live web retrieval returns site, title, URL and text excerpts. | Developer components summarise and link sources; test consumer behaviour separately. | Chinese evidence pages/source register; assess product-retrieval tool integration. | No Wenxin exposure guarantee or verified product catalogue inclusion. |
| 豆包 / 火山方舟China · official web-search plugin capability | Official plugins retrieve public web information, including product information. | Can expose source links; API capability is not Doubao app display policy. | Readable product evidence, source URLs and citation-test package. | General product-feed and native shopping access not confirmed here. |
| KimiChina · consumer search documentation | Documents web search and relevance, authority and freshness considerations. | Search answers include reference links. | Evidence pages with actual studies, applicability, dates and sources. | Weights are undisclosed; general feeds/payment integration not confirmed here. |
| DeepSeekChina · verifiable user-agreement scope | Terms confirm internet search of publicly available information. | Current citation and product-display formats need further verification. | Prepare public traceable pages, then test the consumer product. | Do not infer current features from old API announcements; merchant access remains unverified. |
How do official documents become auditable platform tests?
Select markets, languages, SKU and channels. Freeze brand, ingredient, non-brand, comparison and negative-control queries. Separately log consumer/API answers, model, region, date, sources, product position and redirects. Use consistent baseline/retest criteria, retain failures/missing data, and do not infer undisclosed ranking weights.
How do we publish, get found and connect to products?
Publication, public readability, indexing, citations and purchasing are distinct outcomes. Choose the entry route and assess each step; uploading a PDF is not global agent discovery.
| Entry route | Implementation | Delivery / acceptance | Current Libaorui status |
|---|---|---|---|
| Scientific manuscript / preprint | Prepare complete methods/results and submit within journal/server scope; medRxiv screens content. | Deliver methods, findings, disclosures, manuscript and submission records; distinguish acceptance, review and indexing. | R01 is a rapid draft, not a completed publication-ready SR/Meta. |
| Research archive / version DOI | Consider Zenodo for authorised research reports/data/code, with authors, description, licence and versions. | Public record and DOI/version history enable citation; a DOI is an identifier, not efficacy proof. | Public scope/review unconfirmed; no deposit or DOI. |
| Public evidence page | Publish anonymously readable HTML on the client site or agreed public host: product, question, sources, applicability, authors/review status and update date. | Check stable URL, external readability, crawl permissions, sitemap and versions; test indexing/citation separately. | Current EOS site is a private showcase; no verified public evidence entry. |
| Product feed / documentation | Eligible merchants provide SKU, product URL, price/stock. Google documents product PDF/FAQ fields; ChatGPT uses its own feed specification. | Deliver mapping, submission/review records and product/redirect tests; documents must fit the actual product. | Identity, URL, merchant accounts and approvals pending. |
| Evidence API / MCP | Provide evidence, limits and versions to explicitly integrated agents, with authentication, tool enablement and answer tests. | Assess actual calls, citations and product redirects; non-integrated agents do not automatically use it. | No evidence endpoint or platform calls deployed in this case. |
Route references: medRxiv · Zenodo · Google product PDF · ChatGPT product feed
Libaorui next step: establish actual entry points.
Confirm product/public scope and professional review, deploy evidence and product links on the client/agreed public domain, apply/configure target-platform routes and retain initial retrieval/citation logs, then connect orders for a controlled pilot. R01 draft, S01 document review and P03 implementation protocol are prepared; public release, feed approval, endpoint calls and sales effects remain pending.
LIBAORUI · BEFORE AND AFTER
What did EOS do, and what changed for the client’s product?
Show prepared evidence assets first, followed by the evaluation of exposure, conversion and long-term trust. Map each outcome to EOS work, original records and an observation window.
| Measure | Before / baseline | EOS work | After / finding | Change / status |
|---|---|---|---|---|
| Global agent strategy | Markets, consumer behaviour and merchant baselines need confirmation. | Review official retrieval, citation and product routes for eight platforms, distinguishing APIs from consumer products. | S01 matrix available; actual consumer-product tests not executed. | Official mechanism review prepared |
| Project inputs | Five existing inputs: a formulation sheet and four project documents. | Review the submitted formulation and claims, recording versions and missing inputs. | An inspectable input list, formulation table and gap register. | Initial input review prepared |
| Studies and applicability | Study support in the supplied documents needed claim-level appraisal. | Appraise seven selected human studies for doses, populations, outcomes and contrary findings. | R01 contains seven study records; professional review and a full SR remain pending. | Seven studies rapidly appraised |
| Evidence assets | Existing documents needed a consistent, source-traceable delivery. | Generate the report, study table, structured dossier and sourced FAQ draft. | HTML/PDF/CSV/JSON are available with limits and a version record. | Report draft generated |
| Product and partner materials | Current product page, official SKU and purchase link not supplied; baseline needs archiving. | Plan to place reviewed evidence in comparisons, FAQs, product modules and partner briefs. | Route to the client’s official product page; purchase linking and deployment remain pending. | Deployment pending |
Assess completed assets separately from improved commercial outcomes. Missing baseline data are not zero, and expected gains are not findings. Before/after differences are observational; concurrent controls support incrementality.
Auditable record of prepared work
- Review five project inputs
Identify the submitted formulation version, daily taurine 300mg and missing glutathione amount.
- Rapidly appraise seven human studies
Record populations, doses, findings, access depth and product fit, including null and inconsistent results.
- Prepare inspectable assets
Report, evidence table, JSON dossier and FAQ draft are inspectable; professional review is pending.
- Prepare implementation protocol P03
Add strategy, publication/discovery, product routing, trust and 90/180-day cohorts; implementation and follow-up have not run.
DESTINATION · THE CLIENT PRODUCT
From “why choose it?” to the client’s own product.
Evidence pages and agent answers explain sources and applicability, buying/comparison pages address purchase questions, then link to the official client product page. Record product clicks, payments, refunds and follow-up separately.
Evidence / agent entrySelection / comparisonOfficial client productPayment / follow-up
Observed impressions Impressions
Recorded impressions within the agreed period, reported per platform rather than as cross-platform unique people.
Platform impression reports. Unavailable without these reports; agent test runs are not impressions.
Before — · after — · change unmeasuredProduct click-through rate Product CTR
Unique eligible evidence-page visitors clicking the official product link ÷ unique visitors in that same cohort.
Evidence-page and outbound-click records, confirmed SKU and product URL. A click is not a paid order.
Before — · after — · change unmeasuredLONG-TERM BRAND ASSET · A VALUE HYPOTHESIS
Test whether trustworthy information accumulates into reputation.
The hypothesis: evidence and transparent explanations improve understanding; actual product experience produces feedback; continuing appraisal and correction create a trust record; then test referrals and repeat purchases. Stages may improve, stagnate or decline, and experience, pricing and channels also affect results.
Transparent evidenceClaim traceability
Information trustFixed scale / response rate
Experience / feedbackNPS, complaints / refunds
Referrals / repeat purchases90-day net-order cohort
Accumulated customer value180-day realised contribution
“Compounding trust” is a business hypothesis to test.
Do not assume exponential trust or conversion growth. Both scores and conversion rates are bounded. Track new referred customers, net repeat purchases, trust and contribution each period to distinguish accumulation, acceleration, saturation or decline. Brand trust and medical efficacy are separate questions.
Long-term measures with denominators and observation windows.
| Measure / term | Definition | Records |
|---|---|---|
| Claim traceability rateClaim Traceability | Public health claims with sources, applicability, limits and a version date ÷ all preregistered public health claims. | Frozen product/partner materials and claim register. Measures transparency, not efficacy or consumer trust. |
| Brand-information trust scoreTrust Score · 1–5 | Use the fixed statement ‘I find this brand’s product information trustworthy’ on a 1–5 scale. Report group means and score differences for valid unique respondents. | Concurrent treatment/control surveys with invited/responding counts, response rates and uncertainty. A project-defined measure. |
| Net Promoter ScoreNPS · −100 to 100 | In a 0–10 recommendation survey, subtract the share of valid responses scoring 0–6 from the share scoring 9–10, in percentage points. | Fixed customer cohort and survey records; scores 7–8 remain in the denominator. Separate recommendation intent from actual referral purchases. |
| Organic paid-referral rateVerified Referrals | Within the fixed cohort with complete observation and refund settlement for 90 days after each customer’s first purchase, customers bringing a verified new net-paying customer during those 90 days ÷ customers in that same mature cohort. | Referral tags, deduplicated customers, orders and refunds. Report incentivised referrals separately; unknown origins are not organic referrals. |
| 90-day net repeat-purchase rateRepeat Purchase · Cohort | Within the fixed cohort with complete observation and refund settlement for 90 days after each customer’s first purchase, customers making a second paid order not fully refunded during those 90 days ÷ customers in that same mature cohort. | Deduplicated customer, first/subsequent order and refund records. Freeze cohort membership and settlement rules before analysis. |
| 180-day realised contribution per customerRealised Customer Value | Starting from each customer’s first purchase, fixed-cohort net revenue less order-variable costs over 180 days ÷ initial cohort customer count. Report only cohorts with complete 180-day observation and refund settlement. | Orders, refunds and mutually exclusive cost categories. Keep service/promotion investment separate; not an unlimited lifetime-value forecast. |
| Customer complaint rateQuality Guardrail | Unique customers complaining within the agreed window ÷ customers in the same fully observed delivery cohort; report refund rates separately. | Support, delivery, complaint and refund records, retaining negative feedback and response times rather than selecting favourable reviews. |
| 180-day contribution per randomised visitorITT · Intention-to-Treat | From random assignment, each group’s 180-day net revenue less order-variable costs ÷ all originally randomised eligible visitors, retaining non-purchasers; compare group means. | Original random assignment, visitors, orders/refunds and costs. Analyse original groups and report missing follow-up and referral spillovers into the control. |
Archive claims, pages, SKU, query set and first-purchase cohort rules.
Planned milestone · not executedRetest agents; report product clicks, enquiries, CVR and information-trust surveys.
Planned milestone · not executedFollow mature, refund-settled first-purchase cohorts and retain negative feedback.
Planned milestone · not executedAssess realised customer contribution and maintenance costs to scale, revise or stop.
Planned milestone · not executedD30/D90/D180 are proposed post-deployment reporting dates. Referral, repurchase and customer-value windows start at each first purchase; include the same mature, refund-settled cohort in numerator and denominator. Continue follow-up for immature cohorts. Report survey response rates and loss to follow-up.
Distinguish customer-cohort performance from long-term incrementality.
Referrals, repurchases and per-customer contribution describe customer cohorts. Evidence exposure can change who buys, so buyer-only group comparisons do not establish an EOS effect. Prefer original randomised eligible visitors for long-term incrementality (ITT), including non-buyers, and record referral spillovers into the control.
Measurement references: NPS · Bain · Cohort · Google Analytics. Deduplicate repeat purchases using customer/order identity; device-based site revisits are not customer repurchases.
EVIDENCE TO PURCHASE · AUDITABLE MEASUREMENT
Libaorui: what is delivered, and what is measured?
Here, “the chance of being found” means discovery in a predefined agent query test. Conversion means net purchases by actual visitors. Measure both to locate value and remaining gaps.
Initial draft generated; seven human studies, professional review pending.
Pending: an externally searchable entry point is not established or verified.
Pending: visitor, payment and refund records are not connected.
Pending: no concurrent control or cost records.
Available now
Actual evidence report R01Download implementation protocol P03 · JSONDownload measurement template · CSV
P03 adds global agent strategy, publication/discovery routes and before/after, product-routing and long-term reputation measures; commercial outcomes remain unmeasured. Libaorui remains a proposed sports-nutrition project; its identity is unverified and is not assumed to be an approved health food.
Six core measures with denominators and data sources.
| Measure / term | Definition | Verification data |
|---|---|---|
| Agent discovery rateDiscovery Rate | Answers naming the target product with an accessible evidence link ÷ all completed test answers. | Fixed queries, platform/model, language, date, original answers and links. |
| Accurate citation rateCitation Accuracy | Answers correctly citing the target product’s designated evidence page or registered sources, with accurate relevant findings and limits ÷ all completed test answers. | Original answers, cited sources and review records; a product mention alone does not qualify. |
| Qualified enquiry rateQualified Enquiry Rate | Unique visitors with a qualified enquiry in the agreed window ÷ unique visitors in that same traceable, eligible cohort. | Source tags/dedicated links, site records and CRM; define eligible customers and genuine enquiries in advance. |
| Net purchase conversion rateCVR · Conversion Rate | Unique visitors retaining at least one paid order not fully refunded within the window ÷ all eligible visitors. | Visitor assignment, orders, payments and refunds; partial refunds reduce net revenue. |
| Conversion liftUplift · Incrementality | Evidence-group CVR − concurrent control-group CVR, in percentage points with uncertainty. | Concurrent random A/B assignment and order records; report relative lift separately and treat before/after changes as observational. |
| Incremental contribution marginContribution Margin | Evidence-group net revenue less variable costs, minus the control contribution margin adjusted for visitor count. | Orders, refunds and product/fulfilment/payment/channel-commission variable costs; deduct service and pilot promotion spend once, separately in ROI. |
Discovery and citation rates use project-defined test criteria, not an industry certification. Freeze platform, queries, language, region and date; log repetitions and failures and report completed/failed counts. Results do not estimate discovery across all agents or consumers. Enquiries and payments require actual records; button clicks indicate intent only.
How is the pilot assessed, and who supplies the data?
Define goals and baseline
Freeze one SKU, population and channel. Agree the primary measure, eligible visitors, attribution/refund windows, minimum meaningful effect and sample size before the pilot; targets remain to be agreed.
Fixed-query agent testing
Propose about 50 purchase queries, repeated three times per platform. Separate brand and non-brand queries and retain answers. Record mentions, sources and recommendations separately; repetitions assess stability.
Concurrent sales A/B test
Randomly show current materials or materials enhanced with reviewed evidence in one channel. Keep price/promotions consistent and record availability changes. Use orders/refunds to assess lift and uncertainty; insufficient samples remain inconclusive.
EOS handles appraisal, evidence-to-content mapping, the evaluation protocol and analysis. Clients supply and verify product records, channel permissions, visitors, orders/refunds and variable costs. Report traceable agent traffic separately from unknown origins. Connect data and run the pilot before updating this board; it is not connected to live commercial data.
GEO, CVR, Uplift, ROI: what do these terms mean here?
- GEOGenerative Engine Optimization
- Improve the visibility of genuine, verifiable content in generated answers. Deliver evidence pages and query tests without guaranteed indexing or rank.
- CVRConversion Rate
- Conversion rate: net purchases by unique eligible visitors here, supported by reconciled visitor, payment and refund records.
- UpliftIncrementality
- Incremental effect: compare concurrent treatment and control groups and report percentage-point differences, sample sizes and uncertainty.
- ROIReturn on Investment
- Return: (incremental contribution margin − pilot investment) ÷ pilot investment × 100%. Investment includes service, maintenance and promotion, excludes order-variable costs already deducted, and uses the same period.
Terminology and measurement references: GEO · KDD 2024 · Google Analytics · Google AI Search
Google says existing SEO fundamentals apply to its AI search and indexing is not guaranteed. Net purchases and incremental margin here need actual orders and costs; generic analytics key events are insufficient.
DELIVERED · RAPID EVIDENCE APPRAISAL R01
Inspect the evidence before investing in deeper research.
Five project inputs and seven selected human studies inform this focused appraisal of taurine, magnesium and glutathione. This is not an exhaustive review; other nutrients were not fully appraised for exercise outcomes.
Open the complete evidence reportDownload PDF reportDownload study table CSVDownload evidence dossier JSON
7original human studies
5existing project inputs
300 mgdaily taurine recorded in the input sheet
Ingredient leads do not validate the finished product.
Taurine results differ between running and cycling. Dose, timing and formulation limit extrapolation. No human endurance trial matching this product and dose was verified in this appraisal.
Personalised evidence starts with applicability.
Training, sport, age, nutrient status, dose and outcome need separate appraisal. Older-adult functional findings do not imply athletic endurance gains. Unsupported populations retain an evidence gap.
Assess feasibility before commissioning SR/Meta or a product trial.
Predefine a taurine/endurance PICO and search comprehensively before deciding on pooling. Establishing effects of the actual Libaorui formulation may require a product human study; literature synthesis cannot replace it.
Studies, doses, populations and findings: what was actually checked?
| Source and access depth | Population / dose | Study finding | Applicability to this product |
|---|---|---|---|
| Balshaw et al., 2013 | 8 trained male runners; randomised double-blind crossover. | Positive in this setting 3-km time: 646.6±52.8 vs 658.5±58.2 s; P=0.013. | Small sample; dose and formulation differ from 300 mg/day. No direct product-effect inference. |
| Ward et al., 2016 | 11 trained male cyclists; randomised double-blind crossover. | No time-trial benefit 4-km time: 390±27 vs 388±21 s; P=0.731. | No benefit for this outcome at the same dose; a universal endurance claim is unsupported. |
| Rutherford et al., 2010 | 11 endurance-trained male cyclists; randomised blinded crossover. | Performance unchanged No post-90-min time-trial difference; greater fat oxidation did not mean better performance. | Dose, exercise and measures differ; metabolic markers cannot substitute for endurance outcomes. |
| Finstad et al., 2001 | 32 active young women; double-blind placebo crossover. | Performance/recovery unchanged Ionised Mg increased; performance and recovery did not significantly improve. | Biochemical change is not functional benefit; no ratio to the product's 150 mg elemental Mg is assigned. |
| Veronese et al., 2014 | 139 older women randomised, 124 analysed; no placebo, researchers blinded. | Conditional physical-function benefit SPPB change difference: 0.41 points, P=0.03; signal stronger with lower intake. | Older-adult function, not endurance racing; not directly transferable to young athletes or this formulation. |
| Bomar et al., 2025 | 15 regular exercisers with normal baseline Mg; randomised double-blind crossover. | No TT benefit; some adverse signals 10-km TT unchanged (P=0.89); VO₂max and mean sprint power decreased. | Supports population-specific appraisal; different dose/form prevents a product-harm conclusion. |
| Aoi et al., 2015 | 8 men without regular exercise; double-blind crossover. | Exploratory fatigue signal Some subjective fatigue measures improved; no TT or time-to-exhaustion outcome. | Product content unknown; mitochondrial findings are in mice, not proof of human endurance or product efficacy. |
Expand the formulation sheet and evidence gaps
Amounts come from the submitted sheet, not a verified label or batch assay. Glutathione content is absent from that sheet; its mention in the dossier does not resolve the gap. Four tablets is a recorded input, not a dosing recommendation.
| Ingredient in submitted sheet | Recorded daily four-tablet amount |
|---|---|
| Taurine | 300 mg |
| Fibre (resistant dextrin + inulin) | 1 g |
| Elemental magnesium (from oxide) | 150 mg |
| Elemental zinc | 10.4 mg |
| Selenium | 24 µg |
| Vitamin C | 60 mg |
| Vitamin E (α-TE) | 10 mg |
| Vitamin D3 | 4 µg |
| Vitamin B1 | 1.2 mg |
| Vitamin B2 | 1.2 mg |
| Niacinamide | 6 mg |
| Calcium D-pantothenate | 2 mg |
| Biotin | 10 µg |
- Current label, applicable licence/category documents, SKU version and batch assays.
- Glutathione source, daily quantity and assay; reconciliation of the sheet with the current formulation.
- Finished-product human studies, exercise/diet controls, endurance outcomes, tolerability and conflicts.
- Further full-text review of all seven studies, paired crossover statistics, bias and certainty assessment.
How was this edition retrieved and appraised?
Retrieved 6 October 2026 using web-assisted discovery of original PubMed abstracts and available journal full text. Access depth is recorded by study. No duplicate independent screening, formal risk-of-bias/GRADE assessment or meta-analysis was performed. Professional review is pending.
OpenEvidence access was blocked; no account output was used. Original sources are linked directly. Future OpenEvidence queries can supplement retrieval and cross-checking.
The report records representative executed queries and source access. A full SR needs a frozen search strategy and screening log.
Inspect the draft FAQ for consumers / agents
Do taurine studies prove Libaorui improves endurance?
Not directly. T1 suggests a running benefit, while T2/T3 show no cycling-performance benefit. Dose, formulation and exercise differ, and a matching product trial was not verified.
Source IDs: T1 · T2 · T3
Does magnesium help every exerciser equally?
The selected studies do not support a blanket claim. M1/M3 show no benefit for their measured performance outcomes; M2 concerns older-adult physical function. Nutrient status and goals need appraisal.
Source IDs: M1 · M2 · M3
Why commission SR/Meta or a product trial?
Focused appraisal cannot exclude missing studies or selection bias. SR/Meta can assess comparable evidence systematically; the current formulation's effects still require direct product evidence.
Source IDs: T1 · T2 · T3 · M1 · M2 · M3 · G1
Two evidence deliveries for different questions.
What can the evidence support, and what is missing?
One product, agreed populations and a small set of core claims. Convert retrieval into reviewable findings and applicability judgements.
- Source table, evidence cards, dose/population matching, limitations and gaps.
- Report, structured dossier, consumer FAQ and next-study recommendations.
- Usually 5–10 working days, depending on inputs and professional review.
- Fixed quote by SKU and agreed claim count; not an SR or product certification.
Acceptance: source-level traceability, explicit abstract/full-text depth, null findings and missing information.
Request a rapid-appraisal quoteSystematic review / Meta + conditional applicability
Assess feasibility before systematic searches, duplicate screening, full-text extraction, bias and certainty assessment. Address precise effects and population questions.
- Protocol, search/exclusion logs, data, statistical code, complete report and manuscript.
- Meta only when studies and data are sufficiently comparable; a non-poolable result is still a deliverable.
- Appraise training, sport, age and nutrient status; individual prediction needs additional data.
- Typically 6–12 weeks or longer; source access, study volume and question complexity determine workload.
Acceptance: reproducible protocol, screening and analysis. Preprint preparation, submission and publication are distinct; fees are not tied to positive findings or acceptance.
Assess SR / Meta feasibilityHow does personalised evidence add value?
When data allow, answer which people, exercise, doses and durations a finding applies to. Changing the population label is insufficient; inadequate data remain a gap. Testing the current formulation may need a separate finished-product human study.
Charge for visible deliverables and validation.
Propose separate appraisal, purchase/partner deployment and outcome-validation fees. Begin with the evidence and proceed to a commercial pilot when it is applicable. Scope and prices remain to be agreed.
| Service | Fees and acceptance | Client value |
|---|---|---|
| Global agent strategy | Quote by markets, languages, platforms and questions; deliver the mechanism matrix and test protocol. | Define discovery/display requirements before choosing evidence and product assets. |
| Evidence appraisal | Fixed fee by SKU and claims; assess sources, dose/population fit and limits. | Identify supported value, wording to revise and worthwhile research. |
| Purchase and partner deployment | Deployment fee for agreed evidence pages, FAQs, product-page modules and partner briefs. | Place evidence in comparisons and purchasing, address key questions and support partner diligence. |
| Commercial validation pilot | Pilot fee for query tests, data connection and concurrent A/B analysis, with a complete outcome report. | Measure agent discovery, enquiries, net conversion and margin, with diagnostic findings even when effects do not improve. |
| SR / Meta and product studies | Staged feasibility, protocol, screening and analysis fees; separately budget product studies. | Address important positioning/applicability gaps; fees do not depend on positive findings or publication. |
| Annual maintenance and optional outcome bonus | Annual evidence/FAQ/retrieval maintenance with quarterly outcome boards and 90/180-day cohort reports; any agreed bonus uses verified incremental contribution and a cap. | Keep evidence current and follow trust, referrals, net repeat purchases and realised customer contribution; separate bonuses from findings and publication. |
RECOMMENDED FIRST ENGAGEMENT
One product · one core purchase question · one sales channel
Deliver the appraisal and usable materials, then connect order/cost records to assess scaling. EOS provides applicability, source checks, professional review and outcome analysis. Tools such as OpenEvidence can assist retrieval; outputs still need verification.
Optional assets include preprints, evidence pages, version history, partner briefs, FAQs and structured data for research review, discovery and purchasing. This site retains its existing private audience; external agent discovery, sales-data connection and commercial lift remain unverified.
What has the initial review established?
Existing project dossier
The ingredient sheet and four existing documents inform a rapid appraisal of seven human studies, including sources, dose, population, findings and product applicability.
Product evidence gaps
The submitted sheet records daily taurine 300 mg and elemental Mg 150 mg, with no glutathione amount. Current label, product version and finished-product human findings remain unverified.
Focused project plan
Propose one endurance context, one core claim and one SKU. Complete applicability appraisal before selecting a finished-product study.
Basis: five existing project inputs and seven selected original human studies. The initial rapid report is generated; professional review, a full SR/Meta and finished-product trials are not complete.
Claim opportunity map: how far does the evidence reach?
| Direction | Current status | Next deliverable |
|---|---|---|
| Endurance nutrition | The project uses endurance positioning; category, label and standards matching require verification. | Category, formulation, dose and label appraisal. |
| Ingredients and exercise performance | Seven human studies have been appraised rapidly; findings and doses differ. | Professional review, comprehensive retrieval, bias appraisal and pooling feasibility. |
| Finished-product endurance benefit | No human efficacy results for this finished product have been supplied. | Predefine a core outcome and design a controlled study; findings may be supportive, negative or uncertain. |
| Additional ageing, hair and metabolic claims | Discussed in the dossier; product efficacy and permissible marketing wording are not established. | Create an R&D opportunity and gap register, then assess evidence and product pathways. |
Evidence adds credibility; it does not automatically expand marketing permissions.
Appraise ingredient research, finished-product research and communication permissions separately. Sales pages, brand answers and shopping agents use the same reviewed claims without turning scientific possibilities into established benefits.
Milestones and expected completion
An initial rapid report is generated. Below is the proposed scope for the remaining full project, starting with complete inputs; product trials require separate commissioning.
M01 weeks
Product identity and positioning
Freeze one SKU, formulation, daily intake, label, licence and endurance context.
M11–2 weeks
Claim opportunity map
Separate supported findings, evidence gaps and wording requiring further review.
M23–5 weeks
Core evidence assets
Source register, approximately 10–20 evidence cards and formulation/dose applicability appraisal.
M35–6 weeks
Human- and machine-readable delivery
Evidence page, consumer FAQ, channel summary and structured dossier.
M46–8 weeks
Agent integration and evaluation
About 50 purchase queries, three repeats per platform, with retained answers/sources to measure discovery and accurate citation; integration is separately scoped.
M59–12 weeks
Live channel pilot
Run a concurrent A/B pilot in one channel, measuring origins, qualified enquiries, net purchases, refunds and incremental margin.
How can agents support discovery and channel development?
Consumer queries
Answer questions about exercise, budget, ingredients and use, citing evidence and identifying effects unverified for this product.
Channel prospecting
Propose screening running groups, cycling clubs, event providers and sports retailers, with fit-based prospect lists and partner materials.
Outcome evaluation
Track citation accuracy, qualified visits, qualified leads and orders. No platform integration, channel orders or conversion gains are established yet.
EXPECTED ANSWER EXAMPLE · NOT A LIVE AGENT
“Does ingredient research prove Libaorui will improve my endurance?”
Not yet. The actual dose, formulation, study population and exercise outcome must match; endurance effects of this finished product remain unverified.
Business value: test the investment against actual traffic.
Editable assumptions, not Libaorui operating data or a sales forecast. Contribution margin should deduct incremental product, fulfilment and channel costs; the project fee is deducted separately.
ASSUMES STABLE TRAFFIC AND EFFECT FOR 12 MONTHS
- Additional monthly orders
- 300
- Additional monthly sales
- ¥90,000
- Annual additional contribution
- ¥432,000
- After service fees
- ¥282,000
Additional orders = visits × conversion increase. A zero increase generates no additional sales benefit. Validate effects with a concurrent controlled pilot.
EOS model: separately scoped appraisal, purchase/partner deployment and commercial validation, with annual maintenance and deeper research. Any optional bonus uses verified increments; scientific findings remain independent.
Standards and platform sources
Sports nutrition consumer guidance
Hosted by SAMR with content from the Chinese Institute of Food Science and Technology; covers categories and label-based selection.
Open original sourceTY/T 5002—2025
Recommended sports-nutrition efficacy evaluation standard, effective on this date.
Open original sourceAdvertising Law of China
Review advertising restrictions, including disease-treatment claims, against the actual product.
Open original sourceAgent product discovery
Structured product data supports discovery, not health-claim certification or guaranteed recommendations.
Open original sourceStart with one product and one verifiable value.
Confirm the actual formulation and exercise context, then scope evidence, presentation and the channel pilot.