# Scan bestsellers for product gaps (/use-cases/industry-specialized/ecommerce-bestsellers)



Copy this prompt to start your first task:

<Prompt>
  Set up a daily bestseller and sourcing gap scan for [category] on Amazon. Survey the top 10 bestsellers, compare against our catalog, and flag 3 sourcing opportunities worth pursuing with rationale and risk for each.
</Prompt>

<LivePreview component="NewThread" scenario="{ draft: &#x22;Set up a daily bestseller and sourcing gap scan for [category] on Amazon. Survey the top 10 bestsellers, compare against our catalog, and flag 3 sourcing opportunities worth pursuing with rationale and risk for each.&#x22; }" />

## What is this workflow? [#what-is-this-workflow]

"Scan bestsellers for product gaps" is a recurring research job: pull a category's real bestseller or trending list straight off Amazon (or another target platform) through the Browser, tag every entry with its source URL and collection timestamp, and compare what's trending against your own catalog or inventory when it's connected. Once set up, it becomes a scheduled scan that stays quiet until a tracked category's top rankings shift meaningfully, with every run archived — screenshots included — so a ranking claim from weeks ago still traces back to a real page.

<Callout type="warn">
  **Collection pacing, not optional:** a fixed cadence with randomized delays between category and product page reads, the same discipline any real Browser-driven Amazon collection needs to avoid getting an account flagged. If a CAPTCHA or login wall shows up mid-scan, it stops and tells a human rather than pushing through. It never writes a silently empty field: a failed extraction gets a retry, then a flag, because a blank field usually means Amazon changed its page structure, not that there's nothing to report. And this teammate reads bestseller pages, it does not buy, list, or reprice anything on your own storefront; any action that touches your own listings or pricing needs a human to actually do it.
</Callout>

## How to use this workflow [#how-to-use-this-workflow]

<Steps>
  <Step>
    ### Hire a Researcher: write the bio around category discovery, not tracking your own store [#hire-a-researcher-write-the-bio-around-category-discovery-not-tracking-your-own-store]

    Open **AI Teammates** in the sidebar, click **Hire AI Teammate**, and pick the **Researcher** template. The default bio is generic web research. Replace it on the Profile step so the teammate starts out already knowing this job is scanning the market for what's trending, then checking it against your own catalog, not summarizing what you already stock:

    <Prompt>
      You scan Amazon category bestseller and trending pages via the browser skill to find what's actually selling in the market, not just our own sales history. Cross-reference against our catalog when it's connected and flag anything trending we don't carry. Collect with randomized delays between pages; never write an empty field silently, retry and tell me. Hit a CAPTCHA or login wall: stop, don't force it. You read listings only; you never buy, list, or reprice anything yourself.
    </Prompt>

    Confirm the name and click **Create teammate**. They land in their own channel, which is where you'll send everything below. [More on hiring →](/ai-teammates/hire-an-ai-teammate)

    <LivePreview component="CreateAssistantPage" scenario="{ focus: &#x22;profile&#x22;, initialState: { start: &#x22;template&#x22;, templateRef: &#x22;@helio/researcher&#x22;, name: &#x22;Researcher&#x22;, bio: &#x22;You scan Amazon category bestseller and trending pages via the browser skill to find what's actually selling in the market, not just our own sales history. Cross-reference against our catalog when it's connected and flag anything trending we don't carry. Collect with randomized delays between pages; never write an empty field silently, retry and tell me. Hit a CAPTCHA or login wall: stop, don't force it. You read listings only; you never buy, list, or reprice anything yourself.&#x22; } }" />
  </Step>

  <Step>
    ### Connect the Browser skill: Amazon has no official monitoring API [#connect-the-browser-skill-amazon-has-no-official-monitoring-api]

    Amazon's storefront has no API Helio can connect to natively, and neither do the third-party tools sellers use to track it (Keepa, Helium10, Jungle Scout). What this teammate actually does is install the **Stagehand Browser CLI** skill (Skills & Plugins → Productivity) and drive a real browser: open the category's Best Sellers or Movers & Shakers page, read the ranked list, and screenshot it as an archive. [Skills & Plugins →](/connect/skills-and-plugins)

    <LivePreview component="AssistantSkills" />
  </Step>

  <Step>
    ### Give it a category, and your own catalog if you want gap analysis [#give-it-a-category-and-your-own-catalog-if-you-want-gap-analysis]

    The scan itself only needs a category and a platform. Connect your inventory sheet or storefront export as well if you want the trending list checked against what you already carry:

    <Prompt>
      Scan the Best Sellers page for [category] on Amazon. Give me the top 20 with current price, rating, review count, and BSR. Cross-reference against our own catalog [attached] and call out anything in the top 20 we don't carry.
    </Prompt>

    <LivePreview
      component="AssistantChat"
      scenario="{ assistantName: &#x22;Researcher&#x22;, messages: [
    { author: &#x22;user&#x22;, text: &#x22;Scan the Best Sellers page for [category] on Amazon. Give me the top 20 with current price, rating, review count, and BSR. Cross-reference against our own catalog [attached] and call out anything in the top 20 we don't carry.&#x22; },
    { author: &#x22;assistant&#x22;, text: &#x22;Opened the Best Sellers page for the category through the browser skill and pulled the top 20, source amazon.com/gp/bestsellers, collected today at 14:02 UTC. Price range $34.99 to $129.00, ratings 4.1 to 4.8 stars. Cross-referenced against our catalog: 3 of the top 20 aren't anything we carry, most notably #4, a style that's been climbing for the last two weeks based on review velocity. Full list and a screenshot of the page are in the thread.&#x22; },
  ] }"
    />
  </Step>

  <Step>
    ### Turn it into a recurring category scan [#turn-it-into-a-recurring-category-scan]

    The alert lands in `#merchandising`, and a teammate can only post in channels it's a member of — create that channel if it doesn't exist yet and add this teammate to it first. Then open **Automations** in the sidebar, click the **Create via Chat** dropdown, and pick this teammate. Once a day is enough for a category scan, staggered and not on the hour:

    <Prompt>
      Once a day, staggered timing with a random delay, rescan the Best Sellers page for [category] on Amazon. Diff the top 20 against yesterday's snapshot. If the top 10 shifts meaningfully (a new entrant, a top-5 reshuffle) or a product we don't carry breaks into the top 10, alert #merchandising the same day. Otherwise just log the snapshot; I don't need a message every day nothing moved.
    </Prompt>

    <LivePreview component="NewThread" scenario="{ populated: false }" />
  </Step>
</Steps>

**Nothing here needs your approval by default:** reading bestseller pages through the Browser is a read action, and posting a scan alert to a channel is an internal message — neither reaches outside Helio nor is irreversible, so this flow runs on its own. It's the moment this teammate would touch your own storefront (buying, listing, repricing) that would cross into something needing a human to actually do it. See [Control](/ai-teammates/control) for what pauses for sign-off and why.

## Tips for Better Results [#tips-for-better-results]

| Tip                                                           | Description                                                                                                                                                                            |
| ------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Name the exact category and platform                          | The scan only runs against what you give it — a specific category page on a named platform, not "check what's trending."                                                               |
| Connect your catalog for gap analysis                         | Without an attached inventory sheet or storefront export, the teammate can only report the bestseller list, not what you're missing from it.                                           |
| Ask for shifts only, not daily noise                          | The recurring scan can log a snapshot every day and only message you when the top 10 shifts meaningfully or an uncarried product breaks in — say so, or you'll get pinged for nothing. |
| Treat a CAPTCHA or login-wall stop as a signal, not a failure | The teammate is designed to stop and tell you rather than push through. That's the collection pacing working as intended, not a broken scan.                                           |

## Common Questions [#common-questions]

<Accordions type="single">
  <Accordion title="Does this touch my own storefront?">
    No. This teammate reads bestseller pages only — it never buys, lists, or reprices anything on your own storefront. Any action that touches your own listings or pricing still needs a human to do it.
  </Accordion>

  <Accordion title="Where does the data actually come from?">
    A real browser, driven by the Stagehand Browser CLI skill, opening the category's Best Sellers or Movers & Shakers page on Amazon (or your target platform) and reading the ranked list directly. Amazon has no official monitoring API, and neither do the third-party tools sellers normally use (Keepa, Helium10, Jungle Scout).
  </Accordion>

  <Accordion title="What happens if a CAPTCHA or login wall shows up mid-scan?">
    It stops and tells a human rather than pushing through. That pacing, along with randomized delays between page reads, is the same discipline any real Browser-driven Amazon collection needs to avoid getting an account flagged.
  </Accordion>

  <Accordion title="How often will the recurring scan actually message me?">
    Only when it matters: the top 10 shifts meaningfully — a new entrant, a top-5 reshuffle — or a product you don't carry breaks into the top 10. Otherwise it just logs the day's snapshot without pinging you.
  </Accordion>

  <Accordion title="Can I trace an old ranking claim back to a real page?">
    Yes. Every scan is archived with a screenshot, the source URL, and the collection timestamp, so a claim from weeks ago still traces back to something real, not just a number in a chat message.
  </Accordion>
</Accordions>

<Cards>
  <Card title="Industry & Specialized" href="/use-cases/industry-specialized" description="Back to the Industry & Specialized use case overview." />

  <Card title="Watch competitors weekly" href="/use-cases/research/competitive-analysis" description="Track a locked list of named competitors, not the whole category." />

  <Card title="Run a social media account" href="/use-cases/industry-specialized/social-media" description="Design a launch plan and publish the first week." />

  <Card title="Source and screen candidates" href="/use-cases/industry-specialized/recruiting-sourcing" description="Turn a job description into a scored candidate shortlist." />
</Cards>