Unexpected operation usage when running multiple low-traffic Algolia indices

Answering Your Questions

1. Does merging indices bring the total record count below the 10 000‑record Free tier? Only if the indices contain duplicate objects that can be removed when you combine them. If each index holds unique records, the sum of all records stays the same, so you’ll still hit the limit.

2. Will the added relevance‑tuning complexity cancel out any savings? The complexity itself doesn’t increase operation counts, but it can raise maintenance effort and risk of cross‑type relevance interference. If you can keep a clean, shared ranking strategy, the cost of that extra work is usually lower than the administrative overhead of managing multiple indices.

3. Are there hidden operation costs when using filterable attributes? No additional search operations are charged for filters. A single search request counts as one operation regardless of how many filterable attributes you use. Indexing operations remain per object save/update; you won’t pay more for adding a new filterable field.

Why Merging May Not Reduce Storage

Algolia counts records per index, summed across the entire application. Moving objects from several indices into one does not change that total unless you actually delete duplicates. If your datasets are disjoint, the combined index will still contain every record, so the record quota stays unchanged.

When Merging Can Help

• There is overlap: the same object appears in two or more indices.
• Duplicates can be removed by adding a type or category filterable attribute and storing a single copy per object.
• You can then filter on that attribute in queries, keeping relevance intact.

Practical Steps to Verify and Merge

1. Check total records. In the Algolia dashboard, open Application Overview and note the Total records figure. Compare it to the 10 000 limit.
2. Export object IDs. Use the Algolia REST API or the dashboard export feature to pull objectID lists from each index.
3. Detect overlap. Run a quick script (Python, Node, etc.) to find shared IDs between lists. If the intersection is >0, you have deduplication potential.
4. Design a shared schema. Add a new filterable attribute, e.g., category, and populate it for every object. Ensure each object still satisfies the search use‑cases.
5. Re‑index into a single index. Use the batch API to push the deduplicated objects into the new index. Verify that the new index’s record count equals the unique object count.
6. Update relevance rules. Transfer any custom ranking, typo tolerance, or query rules to the new index. If you had per‑index relevance settings, consolidate them into shared rules that respect the category filter.
7. Test search behavior. Compare search results before and after merging to ensure no loss in relevance or precision.
8. Monitor operations. Check the Operations tab for a few weeks to confirm that indexing and search operation counts remain within the 10 000/month ceiling.

What to Ask Next

Do any of your current indices contain overlapping objectID values? If not, merging will not reduce the total record count below the Free tier threshold.

Conclusion

Consolidating indices saves storage only when duplicates exist. It does not add extra search or indexing costs, but it does require careful relevance tuning. Verify overlap first; if there is none, you’ll need to consider upgrading your Algolia plan or reducing the number of records per index.