Every day dozens of people in a fashion chain's product category ask on X which brand's pyjamas are comfortable or whether anyone can recommend a sports bra. Most of them never tag the brand, so the marketing team never sees the demand. Capturing leads from social media is the flow that finds those posts without waiting for a brand mention, scores them by purchase intent and turns them into a list the sales team can work. This post explains how BranchVoice's lead pass works, how the score is built and how the lead center is used day to day.
Review monitoring versus lead capture
BranchVoice scans X in two separate passes. The review pass collects posts written about your brand; they are classified like reviews from any other platform and are covered in the X customer feedback post. The lead pass looks for posts that do not mention your brand but express a need in your product category. The two use separate keyword lists, separate sync frequencies and separate analysis groups; enabling the lead pass is a single switch in the settings.
The distinction matters because the keyword logic of the two passes is opposite. The review pass searches for your brand name and handles. In the lead pass the brand name produces the wrong results: adverts, price lists and campaign announcements. Lead keywords are written in the language of need and category, such as "lingerie recommendation" or "tights suggestions".
Which posts count as leads
Every collected post passes through an AI gate before it is stored. The gate asks three questions: is the author genuinely a prospective buyer, is the product they mention one of your categories, and is the post promotional? Posts that look like adverts, priced product lists, giveaways, affiliate links or brand announcements are dropped. Posts from the brand's own accounts are skipped outright. A post that fails any of the three is never stored, which keeps the lead list free of noise and limits analysis cost to posts that passed.
A post that clears the gate is stored as a review, flagged as a lead and sent for full analysis. Until the analysis completes the lead is a candidate: the card shows an "awaiting AI assessment" badge and the score stays blank.
How the lead score is built
The analysis asks lead-specific questions. The default lead analysis group has three: purchase intent, brand affinity and urgency. Brand affinity is a multiple-choice question: loyal customer, considering, unaware of the brand, loyal to a competitor or negative. You can add your own questions to the group, such as recommendation seeking, and set how much weight each carries in the score.
The score runs from 0 to 100. The weighted average of the AI answers forms its core; on top comes a small bonus for the author's reach, based on follower count, likes and whether the account is verified. The score places the lead in one of three tiers: 70 and above is hot, 40 and above warm, below that cold. Each lead carries a list of signals explaining the score, such as high purchase intent, time sensitive, seeking recommendations, considering the brand or micro-influencer.
The live lead test on the settings page runs any post text through the same pipeline without storing it. Before changing the keyword list you can answer "what would this post score" in seconds.
The lead center: funnel and status tracking
Captured leads gather in the Lead Center in the sidebar, where the number of new hot leads shows as a badge. The top of the page holds the tier distribution, a weekly new-leads chart and the status funnel. The demand trend card compares the last three days' rate per category and keyword with the previous two weeks, so marketing sees which need suddenly picked up. The number of untouched hot leads and how many hours the oldest has waited sit in the same area as a warning.
Every lead carries a status that the sales team updates on the card: new → contacted → qualified → converted, with irrelevant ones closed as dismissed. The lead detail shows the post text, author information, score signals, the answers to the lead questions and the product category. Team members leave comments on the lead, so a note like "called, asked about sizing, waiting on delivery time" is there at the next touch. The list filters by tier, status, period and free text.
The AI suggestions card above the Lead Center reads the lead statistics, demand trend and waiting hot leads and writes recommendations for the marketing team: which keyword to add, which category is due a campaign, how many leads are waiting for a reply. It refreshes on its own at most once a day.
The marketing team's daily routine
In the morning the team checks the hot-lead badge and opens the waiting ones. Hot leads with a time-sensitive signal go first; the author gets a reply from the brand account or a direct message, the status moves to "contacted" and a short comment is left. Once a week the demand trend card is reviewed: a rising category means the keyword list grows, a fading keyword gets removed. At month end the funnel is read: how many leads were captured, how many were contacted, how many converted.
Leads also appear in scheduled reports. Selecting the leads section in the report builder adds the period's tier and funnel breakdown, categories and sample hot leads; the details are in automated PDF and Excel reports. The lead analysis group follows the same question logic as every other group, which the analysis groups and topic taxonomy post explains.
Next steps
Capturing leads from social media is the mirror image of review monitoring: it finds the person stating a need rather than the one naming your brand. Define your product categories and a handful of need keywords and the first leads arrive with the first sync. To see how it works in your own category, request a demo.