Identify conversion opportunity gaps and suggest ranking improvements

[emirbelkahia]

Context

Some search queries perform well in terms of engagement, users click on results (high CTR), showing initial interest, but fail to convert (low CVR). This disconnect often points to one of two issues:

1. The product ranking doesn't surface the right SKUs high enough, leading users to click on less relevant or underperforming products.
2. The product detail pages (PDPs) fail to convert, due to poor visuals, bad pricing, lack of reviews, or long delivery delays — issues outside of Algolia's scope.

In the 2nd case, the issue may lie beyond Algolia’s scope: for example, poor PDP content (e.g. missing images, weak descriptions, unclear pricing) may explain the drop-off. While this is owned by the merchant, MCP can still help identify the affected queries so CSMs or merchandisers can investigate further. It can also be an opportunity to position Recommend, to lower overall bounce rate by proposing related products in PDPs.

Today, these insights are often missed because they require active analytics deep dives. MCP could automate this analysis and turn raw engagement metrics into concrete, scoped, and actionable recommendations.

Opportunity

MCP can deliver strong value by:

1. Detecting queries with high CTR but low CVR (i.e., conversion opportunity gaps).
2. Highlighting which product attributes or ranking factors might be contributing.
3. Suggesting actions to improve relevance and conversion, such as boosting certain products, improving descriptions, or adjusting custom ranking.
4. Offering a preview of potential ranking changes or actions.

This allows clients to address high-impact queries that are leaking potential revenue.

Proposed behavior

The MCP node should:

• Retrieve analytics data for recent queries, including CTR and CVR metrics.
• Identify queries with high engagement but low conversion.
• For each flagged query (or as general recommendations for all of them, depending on what's technically doable):
  • Suggest potential reasons (e.g irrelevant products, poor product data).
  • Recommend actions (e.g boosting specific products, improving images or pricing).
  • Optionally generate a snippet to modify ranking or pin key products.

Example prompt

Can you find queries with high CTR but low CVR in my analytics, and tell me how to improve them?

Expected output

{
  "query": "running shoes",
  "ctr": 45.2,
  "cvr": 2.1,
  "insight": "This query has high interest but low conversions. The query returns 5640 results. Such a large set of results can lead to choice paralysis, and / or to suboptimal products shown on the first lines / pages.",
  "suggestedActions": [
    "🏷️ Boost in-stock products with better visuals.",
    "📦 De-prioritize items with long shipping delays.",
    "📝 Improve product descriptions for top 5 results."
  ],
  "optionalConfigSnippet": {
    "customRanking": ["desc(stockStatus)", "desc(imageQualityScore)"]
  }
}

Value

• Surfaces high-leverage opportunities that are often hidden in analytics noise.
• Helps clients act on real shopper intent data to improve business outcomes.
• Saves time for CSMs and provides tangible, data-backed recommendations.
• Promotes smarter use of custom ranking, pinning, and merchandising controls.

Notes

• Could be extended with segmentation (e.g. mobile vs desktop, by market).
• Confidence level per insight could help prioritize actions.
• Future version might include A/B testing suggestions.