How Bookbot powers 3.5 million saved searches and ~10% of revenue with Manticore Search

Описание и применение

Bookbot использует Manticore Search для поддержания актуальности 1,29 млн физических книг, точного отображения копий и обратного поиска новых поступлений по 3,5 млн сохранённых запросов, что приносит ~10 % дохода.

Задача

Second‑hand bookstores need search that reflects real‑time inventory, precise copy details, and reverse matching for future demand, while keeping performance high.

Результат

Bookbot now serves accurate, up‑to‑date search results, reduces inventory mismatch, and generates ~10% of revenue through Watchdog alerts, all while using a single search engine across multiple workloads.

Рабочий процесс

• 1. Capture order, reservation, price change, or catalog event via Kafka.
• 2. Update, replace, or delete the corresponding record in Manticore’s book index in real time.
• 3. For customer queries, filter copy‑level records, group by title, and select representative copy with WITHIN GROUP ORDER BY.
• 4. For new book arrivals, submit the book to Manticore’s percolate table to retrieve matching Watchdog rules.
• 5. Apply business logic to decide notification timing and send alerts to customers.
• 6. Use a dedicated typo‑correction dictionary for forgiving search while keeping inventory index fresh.
• 7. Maintain separate title‑level precomputed index for fast browsing when copy‑level precision is unnecessary.
• 8. Reuse Manticore for bibliographic lookup, internal search, and geo‑search across the business.

Доказательства

Manticore lets us update the searchable offer as the physical inventory changes. When one copy sells, we can replace or remove that record without rebuilding the whole index.

Источник

Manticore lets us match each newly available book against around 3.5 million saved searches. Watchdogs now account for roughly 10% of our revenue.

Источник