Product Properties — SKU, Size, Color, Room Type
The most powerful prompts in the library. They lean on SealMetrics' conversion_items table (one row per product, denormalized properties) and on Map(String, String) properties on micro-conversions like add_to_cart, view_product, view_room.
MCPs required: SealMetrics MCP
Pre-flight check: before running these, confirm which properties your site actually captures. Use list_property_keys and get_property_values to validate.
Ecommerce property prompts
SEAL-061 — SKUs viewed vs purchased
Using SealMetrics MCP for site {site_id}, for the last 30 days:
For every SKU that received at least 100 view_product events:
- view_product count, add_to_cart count, purchases (from conversion_items), view-to-purchase ratio.
Rank by view-to-purchase ratio ascending (worst at the top). Take the worst 50 with high view volume.
For each, classify the leak hypothesis: price too high, out of stock, weak product description, missing reviews, slow load.
Output: 1 sortable table + a "investigate first" top-10 list.
SEAL-062 — Add-to-cart without purchase by SKU
For site {site_id}, query SealMetrics MCP for the last 30 days.
Per SKU with more than 30 add_to_cart events:
- Add-to-cart count, purchases, cart-to-purchase ratio.
Rank by cart-to-purchase ratio ascending. Take the bottom 30.
For each, identify whether the leak is shipping cost (compare AOV vs free-shipping threshold), payment method friction, or pricing. End with a 3-bullet "fixes likely to recover X% revenue" summary.
SEAL-063 — Worst-converting sizes (per top product)
Using SealMetrics MCP for site {site_id}, for the last 60 days:
For the top 20 products by add_to_cart, break down view-to-purchase ratio by `size` variant.
Highlight sizes that are frequently added to cart but rarely purchased — likely stock-out or sizing issues. Highlight sizes purchased at high rate but rarely viewed — likely under-merchandised.
Output: a nested table per product. End with a "stock or size guide" recommendation per flagged product.
SEAL-064 — Color winners and losers per category
For site {site_id}, query SealMetrics MCP for the last 90 days.
Per product category: distribution of view_product events by `color`, distribution of purchases by `color`.
Compute exposure share vs purchase share per color. Flag overexposed colors (high view share, low purchase share) and underexposed colors (low view share, high purchase share).
Output: 1 nested table by category × color. End with a 5-bullet merchandising brief: which colors to push to homepage, which to retire from hero placements.
SEAL-065 — Price sensitivity by category
Using SealMetrics MCP for site {site_id}, for the last 90 days:
Per product category, distribute purchases by price bucket (0-25, 25-50, 50-100, 100-200, 200-500, 500+ in store currency).
For each category, identify which bucket leads in revenue. Compare to which bucket leads in add_to_cart (intent). Mismatches indicate price-objection risk.
End with a 3-bullet pricing recommendation per top-3 categories.
SEAL-066 — Hidden cross-sell opportunities
For site {site_id}, query SealMetrics MCP for the last 90 days.
For every conversion that includes more than one item in conversion_items, identify SKU pairs that co-occur. Compute pair frequency and pair lift (observed pair frequency / expected if independent).
Output: top 20 pairs by lift with at least 30 co-occurrences. Mark pairs that already form an official bundle in the catalog (if I provide {existing_bundles}) and pairs that are organic only.
End with 5 cross-sell pairs to promote on PDP and at checkout.
SEAL-067 — SKUs in cart but never bought
Using SealMetrics MCP for site {site_id}, for the last 30 days:
List every SKU with more than 50 add_to_cart events and zero purchases.
For each: SKU, category, add_to_cart count, average price, top traffic source, % of carts containing only this SKU vs mixed carts.
Recommendation per SKU: pause exposure / lower price / improve PDP / check stock. Sort by add_to_cart volume desc.
SEAL-068 — Orphan size-color variants
For site {site_id}, query SealMetrics MCP for the last 60 days.
For every (size, color) variant combination across the catalog: view events, purchases.
List combinations with views but zero purchases. Flag those likely to be stock-out (low view count and abrupt drop) vs likely failed (high view count and sustained zero purchases).
Output: 1 ranking table. End with a "discontinue / restock / reprice" verdict per combination.
SEAL-069 — Discount lift per SKU
Using SealMetrics MCP for site {site_id}, for the last 90 days:
Per SKU, compare conversions with `discount_code` present vs without. Compute lift in CR and lift in revenue per session.
Identify SKUs where discount usage does NOT increase conversion meaningfully (lift below 5%). Those are margin-bleeders.
Output: ranked list of SKUs by lift, marking the bleeders. End with a 3-bullet recommendation on which SKUs should be excluded from discount campaigns.
SEAL-070 — AOV by landing page and source
For site {site_id}, query SealMetrics MCP for the last 60 days.
Compute AOV per landing page (top 50 by entrances) and per utm_source (top 20).
Identify the landing pages and sources that systematically deliver above-average AOV — those deserve more investment.
Identify those that systematically deliver below-average AOV — those need cross-sell prompts at checkout.
Output: two tables (landing × AOV, source × AOV) + a 3-bullet brief.
SEAL-071 — Stock-out detection (proxy)
Using SealMetrics MCP for site {site_id}, scan the last 14 days for SKUs with a sudden drop in view_product events without a corresponding drop in upstream traffic (search, listing, ads).
A drop is sudden if 7-day average drops more than 60% vs prior 14-day average.
Output: candidate SKUs likely hidden by stock-zero rules in the storefront. Cross-check with last purchase date.
SEAL-072 — Returning category buyers
For site {site_id}, query SealMetrics MCP for the last 90 days.
Per category, % of purchases made by returning customers (sessions with login micro-conversion before purchase) vs new customers.
Rank categories by retention strength. Top categories deserve loyalty / email investment. Bottom categories need acquisition.
Output: 1 ranking + 3 retention bullets.
SEAL-073 — Cart abandonment by payment method
Using SealMetrics MCP for site {site_id}, for the last 30 days:
If `payment_method` is captured on the reach_payment micro-conversion or on the conversion macro:
Per payment method: reach_payment events, conversions, completion ratio, average AOV.
Identify payment methods with abnormally low completion ratio (>20% below site average) — likely friction or gateway issue.
Output: 1 table + a 3-bullet recommendation.
SEAL-074 — Currency mix vs country
For site {site_id}, query SealMetrics MCP for the last 60 days.
Cross `currency` from conversions with the country of origin of the session. Identify users buying in a currency that is NOT their country's default — opportunity for currency localization.
Output: top 10 country-currency mismatches by lost revenue potential. End with the top 3 countries to add native currency to.
SEAL-075 — Pareto of revenue by SKU
Using SealMetrics MCP for site {site_id}, for the last 180 days:
Build a Pareto curve of revenue by SKU. Compute how many SKUs produce 80% of total revenue and how many produce the long tail.
For the bottom 40% of SKUs by revenue (catalog drag), output the count, % of total catalog, and a 3-bullet recommendation on whether to deindex, deprioritize, or merge.
Hotels property prompts
SEAL-076 — Room type oversold vs undersold
For site {site_id} (a hotel), query SealMetrics MCP for the last 90 days.
Per `room_type`: view_room count, bookings, view-to-book ratio, revenue, average ADR.
Identify room types with high view share but low booking share (oversold in marketing) and low view share but high booking share (undersold in marketing).
End with a 3-bullet brief: which room type to feature on the homepage, which to push in remarketing, which to repackage.
SEAL-077 — Pickup curve by days-to-check-in
Using SealMetrics MCP for site {site_id}, for the last 90 days:
For every booking, compute days-to-check-in = check_in - booking date (using the property `check_in` on conversions). Bucket: same-day, 1-3, 4-7, 8-14, 15-30, 31-60, 60+.
Per bucket: bookings, revenue, ADR, share of total bookings.
Compare to same-period last year if data exists. Flag any bucket whose share moved more than 5 percentage points.
End with a 3-bullet revenue management recommendation.
SEAL-078 — Length-of-stay opportunities
For site {site_id} (a hotel), query SealMetrics MCP for the last 90 days.
Distribution of bookings by `nights` (length of stay), per month and per country of origin.
Identify country-month segments where average length of stay is below site median and total volume is meaningful — those are candidates for "minimum 3 nights" promos or stay-longer-save bundles.
Output: 1 heatmap-style table country × month. End with 3 specific package ideas (target country, ideal LOS, hero room type).
SEAL-079 — Rate plan winners
Using SealMetrics MCP for site {site_id}, if `rate_plan` or `board` (BB, HB, FB, AI) is captured:
For the last 90 days, per rate plan: view-to-book ratio, bookings, revenue, ADR, average length of stay.
Identify rate plans frequently viewed but rarely booked (priced wrong or competing with cheaper plan in same flow) and rate plans rarely viewed but frequently booked (under-merchandised).
Output: 1 ranking + 3-bullet packaging recommendation.
SEAL-080 — Channel × room type matrix
For site {site_id}, query SealMetrics MCP for the last 90 days.
Cross utm_source / channel with `room_type` purchased. Per channel: distribution of bookings by room type, ADR per channel, revenue contribution.
Identify channels that systematically bring low-margin room types (low ADR) — they may not deserve more spend even if booking count is high.
Identify channels that bring premium room types — they deserve more investment.
Output: 1 channel × room type matrix + a 3-bullet bidding recommendation.