Algolia Recommend Alternative: Ranking by What the Shopper Says
Algolia Recommend answers "what goes with this?" from events your store has already recorded. TasteRay answers "which of these suits me?" from what one person just said.
TasteRay is an alternative to Algolia Recommend when a shopper or viewer states a need and you want each result ranked against it with a written reason. Algolia Recommend suits stores already on Algolia, whose models learn from click and conversion events and retrain daily beside search and merchandising.
Algolia sells recommendations inside a search and discovery platform, and its product page speaks to conversion on storefronts. If search is already on Algolia, adding its models is the short path. TasteRay does a narrower job that sits next to those models, and you can add it without moving anything.
Two different questions on a store page
- "What else goes with this?"
- Frequently bought together, related items, trending and looking similar answer from what other people did or from how products look. This is what Algolia's models are built for, and they need no words from the shopper.
- "Which of these suits me?"
- A shopper says they fly long-haul and want to sleep. TasteRay ranks up to ten products you send against that sentence and writes down why, which also works on the first day a product exists.
What each Algolia model needs first
Algolia documents a minimum for each model, so you can check your store against it in a few minutes. Looking similar is image-based and needs no events. After initial training, models retrain once a day, and Algolia says the frequency cannot be changed.
| Model | Algolia minimum | Nearest TasteRay call |
|---|---|---|
| Frequently bought together | 1,000 conversion events with two or more items | No equivalent. TasteRay has no co-purchase data, so keep a behavioural model for this |
| Related items | 10,000 click and conversion events | Send the current product as context and ten candidates in items |
| Related content | 10 items with values in their content-based attributes | Send the same attributes in description or metadata |
| Trending items and facet values | 250 conversion events | No equivalent. Trending needs your own event counts |
| Looking similar | No events, works from images | No equivalent. The API reads text and does not take images |
Algolia Recommend and TasteRay side by side
| Question | Algolia Recommend | TasteRay | Source |
|---|---|---|---|
| Input | Events and product attributes, or images for looking similar | What a person says, up to 50 rated items, and your list of candidates | Algolia docs, TasteRay docs |
| Keeping current | Models retrain once a day | Each request starts fresh, with no training on self-serve | Algolia docs, TasteRay docs |
| A reason per result | Not listed in their docs (as of Nov 2026) | why_match and key_factors on every result | Algolia docs, TasteRay docs |
| Beside it | Search, merchandising rules, A/B testing and Recommend Analytics for clicks, conversions and revenue | The ForMe widget for one product page. Usage analytics on Pro | Algolia product page, Stores |
| Free allowance | 5,000 recommendation requests a month | 1,000 requests a month | Algolia pricing, TasteRay prices |
| Paid | Grow and Grow Plus include 10,000 requests, then $0.60 per 1,000, with no contract | $49 a month for 10,000, $199 for 100,000 | Algolia pricing, TasteRay prices |
| Contract tier | Elevate, on an annual contract | Enterprise, custom | Algolia pricing, Enterprise |
The price at 100,000 requests
On recommendation requests alone, Algolia comes out lower. On a Grow plan, 100,000 requests means 90,000 over the allowance, or $54 at $0.60 per 1,000. TasteRay Pro is $199 for the same count. Any other Algolia usage on your plan is billed on top of that $54.
The units differ in what comes back. An Algolia request returns recommended items from a trained model. A TasteRay request generates a written reason for every result, which takes a few seconds, so it fits a request someone typed or a cached block better than every product tile on every page view.
Adding TasteRay to an Algolia store
Keep search and rows where they are
Nothing about Algolia has to change, and the two systems never have to talk to each other.
Pick one moment of doubt
A gift finder, or a text box above a category. Turn the shopper's words into
context.profile.Use Algolia to find the candidates
Take the top ten hits for the category or query and send them in
items, with your own price inmetadata.Show your facts and their reason
Read price, stock and specification from your database. Treat
why_matchas an opinion, and read a sample for wrong product claims before launch.
When to choose Algolia Recommend
- You already use Algolia search and want recommendations from one vendor beside it.
- Your catalog is commerce-like, and bought-together or similar-looking is the main job.
- You have the event volume in the table above, or you will soon.
- You want merchandising rules, A/B testing and revenue analytics in the same product.
- You want recommendation tiles on most pages at a low price per request.
When to choose TasteRay
- A shopper can describe a need or a person they are buying for, and you want that sentence to decide the order.
- You have new products or a small store with too few events for a behavioural model.
- The written reason is the point, on a gift finder or in a comparison.
- Your catalog is media, such as books, games or movies and TV series, where taste decides more than co-purchase.
Algolia Recommend is a separate company; see its own site. TasteRay is not affiliated with it.
Sources and further reading
- AI recommendations (Algolia, opened 5 November 2026)
- Algolia Recommend overview (Algolia, opened 5 November 2026)
- Algolia pricing (Algolia, opened 5 November 2026)
FAQs about Algolia Recommend Alternative
How many events does Algolia Recommend need?
It varies by model. Algolia documents 1,000 conversion events for frequently bought together, 10,000 click and conversion events for related items, and 250 conversion events for trending items, as of November 2026. Related content needs 10 items with attributes, and looking similar works from images without any events.
Does Algolia Recommend explain why it picked a product?
A per-result reason is not listed in Algolia's docs as of November 2026. We checked its product page and its Recommend overview, which describe the models, their minimums and analytics for clicks, conversions and revenue. If a written reason matters to you, ask Algolia directly, since documentation can leave things out.
Do I have to leave Algolia to use TasteRay?
No. You can keep Algolia for search and rows, take its top ten hits for a query, and send them to TasteRay with what the shopper said. The API ranks those products and writes a reason for each. Your server joins the two, and nothing on the Algolia side has to change.