◈

How to analyze Xiaohongshu content in bulk for ecommerce

Export Xiaohongshu notes, comments, orders to tables, then run Sales Extractor, Sales Lens and Customer Insight locally for leads, cohorts, cross-sell pairs.

Skills in this bundle

1/6

B2B ROI Pulse

Calculate B2B sales ROI with detailed financial analysis including payback period, NPV, and profit projections

Skill3 Nex
2/6

Sales Lens

Sales analyze copilot across 5 objects (mode x top_n; LLM-orchestrated structured result). (Action Co-pilot: generates next-best-actions (actions.json), follow-up tasks and calendar reminders on top of the analysis — what to act on today.)

Skill1 Nex
3/6

Sales Extractor

Sales extract copilot across 1 objects (mode x top_n; LLM-orchestrated structured result). (Action Co-pilot: generates next-best-actions (actions.json), follow-up tasks and calendar reminders on top of the analysis — what to act on today.)

Skill3 Nex
4/6

Value Canvas

Generate structured value proposition canvas for B2B sales based on customer profile and pain points. Maps customer jobs, pains, gains against products/services, pain relievers, and gain creators.

Skill3 Nex
5/6

Cross-sell Recommender

Deterministic cross-sell via product co-occurrence. (Action Co-pilot: generates next-best-actions (actions.json), follow-up tasks and calendar reminders on top of the analysis — what to act on today.)

Skill3 Nex
6/6

Customer Insight

Unified customer-data analyzer: churn risk, cohort retention, buying journey, health score, sentiment (Voice of Customer), user story map.

Skill10 Nex
Quick answer

Export your Xiaohongshu note data, comments, DM leads and store orders as Excel or CSV, then run six sales skills locally in Nexus Runner: Sales Extractor, Sales Lens, Customer Insight, Cross-sell Recommender, Value Canvas and B2B ROI Pulse. You get structured lead tables, content findings, customer cohorts and sentiment, co-purchase pairs and a payback model, all produced as files on your own machine.

  • Input is the Xiaohongshu content and order tables you export yourself (Excel/CSV); skills run in Nexus Runner 1.0.10 on your PC and the files never leave it.
  • Sales Extractor turns comments and DMs into structured contact and intent fields (3 Nex per run); Sales Lens returns structured findings by mode and top_n (1 Nex per run).
  • Customer Insight covers churn risk, cohort retention, buying journey, health score, sentiment and RFM segmentation in one run (10 Nex per run).
  • Cross-sell Recommender mines co-purchase pairs from order lines deterministically, with no model call (3 Nex per run).
  • Free tier is 200 Nex per month, reset at the start of each month; Pro is CNY 49/month with 3000 Nex; Enterprise is CNY 299/month with 50000 Nex.

What this actually is

It is not a bot that browses Xiaohongshu for you. It is a local pipeline that turns data you already exported into structured conclusions. You export notes, comments, DM leads and order lines from the creator and merchant dashboards, Nexus Runner reads those files, and the six sales skills run in sequence: each one takes fixed input fields and returns a fixed result shape, so one skill's output can feed the next. Skills come in two kinds only, skill and agent, and every skill on this page is already in the library and callable on your machine.

How to do it

1Step 1
Step 1: export the raw tables
From the Xiaohongshu creator and merchant dashboards, export note data, comments, DM leads and order lines as Excel or CSV into one folder. The skills read the files you hold; they do not sign in to Xiaohongshu and do not operate your account.
2Step 2
Step 2: Sales Extractor turns free text into fields
Contacts, stated needs and intent signals buried in comments and DMs are extracted into a structured list. The object type is inferred from the input, and missing fields are left empty rather than invented. 3 Nex per run.
3Step 3
Step 3: Sales Lens shows which content produced conversations
It analyzes five sales objects (outreach, pitch, proposal, prospect and more) and returns structured findings by mode and top_n, such as which batch of notes produced the most inbound leads and which leads to work first. 1 Nex per run.
4Step 4
Step 4: Customer Insight profiles the buyers
Feed it the conversion and repeat-purchase data and it returns churn risk, cohort retention, buying journey, health score, Voice of Customer sentiment, a user story map, RFM segments, CRM de-duplication and a combined report. 10 Nex per run.
5Step 5
Step 5: Cross-sell Recommender finds co-purchase pairs
Give it order lines (order id plus product) and it mines cross-sell combinations from in-order co-occurrence. The calculation is deterministic with no model call, so the same order table gives the same result twice. 3 Nex per run.
6Step 6
Step 6: Value Canvas for the message, B2B ROI Pulse for the math
Value Canvas maps customer jobs, pains and gains against your products, pain relievers and gain creators, giving the message frame for the next batch of notes. B2B ROI Pulse computes payback period, NPV and profit projections from your spend and deal data. 3 Nex per run each.

Three ways to do it

MCP NexusManual workGeneric cloud AI
Where the data goesSkills run in Nexus Runner on your machine; exported tables and result files stay thereFiles stay local, but a person reads them row by rowThe raw table has to be uploaded before analysis can start
Bulk handlingReads a whole table at once; six skills chain in sequence, each output feeding the nextCopy and paste per row, so large batches get sampled insteadContext limits force you to split long tables and stitch results back yourself
Output shapeStructured tables and report files with fixed fields, ready for the next stepA hand-assembled document formatted differently every timeA block of chat text you still have to turn into a table
ReproducibilityCross-sell Recommender is deterministic; the other skills follow a fixed output contractDepends on which colleague is on shift that dayThe same data asked twice can return different conclusions
How it is billedPer run in Nex, 1 to 10 Nex for the skills here, free tier 200 Nex per monthBilled as staff hoursBilled per seat, same price whether you run it once or often

Real use cases

Collecting leads out of comments and DMs
Input is the exported comment and DM CSV. Sales Extractor pulls contacts, needs and intent into a structured lead table, then Sales Lens ranks follow-up order by mode and top_n, producing a list you can assign directly.
Two steps in one pass, 4 Nex total
Co-purchase pairs from store orders
Input is the order line table (order id plus product name). Cross-sell Recommender computes frequent in-order combinations and returns a pair list you can use for note selection and bundle pages.
Deterministic: the same order table returns the same pairs
Message and payback check before the next push
Input is a customer profile with pain points plus your spend and deal figures. Value Canvas maps jobs, pains and gains onto products and pain relievers; B2B ROI Pulse returns payback period, NPV and profit projections.
Two skills in one pass, 6 Nex total

FAQ

How do I get Xiaohongshu data into these skills?

Through Excel or CSV files you export yourself from the creator and merchant dashboards. The skills read local files only; they do not sign in to Xiaohongshu, do not operate third-party accounts, and do no screen-level automation.

These are sales skills, so can they really analyze Xiaohongshu content?

Yes, once the content data is in table form. Comments, DMs and inbound leads go to Sales Extractor and Sales Lens; conversion and repeat purchases go to Customer Insight and Cross-sell Recommender; messaging and payback go to Value Canvas and B2B ROI Pulse. They analyze the fields you exported, not the pixels of images or video.

How much Nex does one full pass cost?

Running all six once totals 23 Nex: Sales Lens 1, Sales Extractor 3, Value Canvas 3, Cross-sell Recommender 3, B2B ROI Pulse 3, Customer Insight 10. The free tier gives 200 Nex per month reset at the start of the month, Pro is CNY 49/month with 3000 Nex, Enterprise is CNY 299/month with 50000 Nex.

Is my data uploaded anywhere?

No. The skills execute inside Nexus Runner 1.0.10 on your own machine, and both the exported tables and the result files stay on that machine. Runner supports Windows 10/11 x64 only.

Are the results reproducible, and can the skills make things up?

Cross-sell Recommender is a deterministic calculation, so the same order table returns the same result twice. The other skills follow a fixed field contract and leave a field empty when the input does not contain it rather than inventing a value.

Related use cases