Making Each Search Rewarding: How Ibotta Remodeled Supply Discovery With Databricks

At Ibotta, our mission is to Make Each Buy Rewarding. Serving to our customers (whom we name Savers) discover and activate related provides by our direct-to-consumer (D2C) app, browser extension, and web site is a crucial a part of this mission. Our D2C platform helps hundreds of thousands of consumers earn cashback from their on a regular basis purchases—whether or not they’re unlocking grocery offers, incomes bonus rewards, or planning their subsequent journey. By the Ibotta Efficiency Community (IPN), we additionally energy white-label cashback applications for a few of the greatest names in retail, together with Walmart and Greenback Normal, serving to over 2,600 manufacturers attain greater than 200 million shoppers with digital provides throughout companion ecosystems.

Behind the scenes, our Knowledge and Machine Studying groups energy crucial experiences like fraud detection, supply advice engines, and search relevance to make the Saver journey customized and safe. As we proceed to scale, we want data-driven, clever programs that assist each interplay at each touchpoint.

Throughout D2C and the IPN, search performs a pivotal function in engagement and must hold tempo with our enterprise scale, evolving supply content material, and altering Saver expectations.

On this put up we’ll stroll by how we considerably refined our D2C search expertise: from an bold hackathon venture to a sturdy manufacturing function now benefiting hundreds of thousands of Savers.

We believed our search might higher sustain with our Savers

Person search habits has advanced from easy key phrases to incorporating pure language, misspellings, and conversational phrases. Fashionable search programs should bridge the hole between what customers kind and what they really imply, decoding context and relationships to ship related outcomes even when question phrases don’t precisely match the content material.

At Ibotta, our authentic homegrown search system, at instances, struggled to maintain tempo with the evolving expectations of our Savers and we acknowledged a possibility to refine it.

The important thing areas for alternative we noticed included:

• Bettering semantic relevance: Specializing in understanding Saver intent over actual key phrase matches to attach them with the suitable provides.
• Enhancing understanding: Decoding the complete nuance and context of consumer queries to offer extra complete and actually related outcomes.
• Rising flexibility: Extra quickly integrating new supply sorts and adapting to altering Saver search patterns to maintain our discovery expertise rewarding.
• Boosting discoverability: We wished extra sturdy instruments to make sure particular forms of provides or key promotions have been constantly seen throughout a big selection of related search queries.
• Accelerating iteration and optimization: Enabling quicker, impactful enhancements to the search expertise by real-time changes and efficiency tuning.

We believed the system might higher hold tempo with altering supply content material, search behaviors, and evolving Saver expectations. We noticed alternatives to extend the worth for each our Savers and our model companions.

From hackathon to manufacturing: reimagining search with Databricks

Addressing the constraints of our legacy search system required a targeted effort. This initiative gained important momentum throughout an inner hackathon the place a cross-functional workforce, together with members from Knowledge, Engineering, Advertising Analytics, and Machine Studying, got here along with the concept to construct a contemporary, various search system utilizing Databricks Vector Search, which some members had discovered about on the Databricks Knowledge + AI Summit.

In simply three days, our workforce developed a working proof-of-concept that delivered semantically related search outcomes. Right here’s how we did it:

1. Collected supply content material from a number of sources in our Databricks catalog
2. Created a Vector Search endpoint and index with the Python SDK
3. Used pay-per-token embedding endpoints with 4 totally different fashions (BGE massive, GTE massive, GTE small, a multilingual open-source mannequin, and a Spanish-language-specific mannequin)
4. Linked every little thing to our web site for a stay demo

The hackathon venture received first place, generated robust inner buy-in and momentum to transition the prototype right into a manufacturing system. Over the course of some months, and with shut collaboration from the Databricks workforce, we remodeled our prototype into a sturdy full-fledged manufacturing search system.

From proof of idea to manufacturing

Transferring the hackathon proof-of-concept to a production-ready system required cautious iteration and testing. This part was crucial not just for technical integration and efficiency tuning, but in addition for evaluating whether or not our anticipated system enhancements would translate into constructive adjustments in Saver habits and engagement. Given search’s important function and deep integration throughout inner programs, we opted for the next method: we modified a key inner service that known as our authentic search system, changing these calls with requests directed to the Databricks Vector Search endpoint, whereas constructing in sturdy, swish fallbacks to the legacy system.

Most of our early work targeted on understanding:

Within the first month, we ran a check with a small proportion of our Savers which didn’t obtain the engagement outcomes we had hoped for. Engagement decreased, significantly amongst our most energetic Savers, indicated by a drop in clicks, unlocks (when Savers specific curiosity in a suggestion), and activations.

Nonetheless, the Vector Search answer supplied important advantages together with:

• Quicker response instances
• A less complicated psychological mannequin
• Better flexibility in how we listed information
• New talents to regulate thresholds and alter embedding textual content

Happy with the system’s underlying technical efficiency, we noticed its better flexibility as the important thing benefit wanted to iteratively enhance search end result high quality and overcome the disappointing engagement outcomes.

Constructing a semantic analysis framework

Following our preliminary check outcomes, relying solely on A/B testing for search iterations was clearly inefficient and impractical. The variety of variables influencing search high quality was immense—together with embedding fashions, textual content combos, hybrid search settings, Approximate Nearest Neighbors (ANN) thresholds, reranking choices, and lots of extra.

To navigate this complexity and speed up our progress, we determined to ascertain a sturdy analysis framework. This framework wanted to be uniquely tailor-made to our particular enterprise wants and able to predicting real-world consumer engagement from offline efficiency metrics.

Our framework was designed round an artificial analysis surroundings that tracked over 50 on-line and offline metrics. Offline, we monitored commonplace info retrieval metrics like Imply Reciprocal Rank (MRR) and precision@okay to measure relevance. Crucially, this was paired with on-line real-world engagement indicators corresponding to supply unlocks and click-through charges. A key determination was implementing an LLM-as-a-judge. This allowed us to label information and assign high quality scores to each on-line query-result pairs and offline outputs. This method proved to be crucial for fast iteration primarily based on dependable metrics and accumulating the labeled information needed for future mannequin fine-tuning.

Alongside the way in which, we leaned into a number of elements of the Databricks Knowledge Intelligence Platform, together with:

• Mosaic AI Vector Search: Used to energy high-precision, semantically wealthy search outcomes for analysis exams.
• MLflow patterns and LLM-as-a-judge: Offered the patterns to guage mannequin outputs and implement our information labeling course of.
• Mannequin Serving Endpoints: Environment friendly deployment of fashions immediately from our catalog.
• AI Gateway: To safe and govern our entry to 3rd occasion fashions through API.
• Unity Catalog: Ensured the group, administration, and governance of all datasets used throughout the analysis framework.

This sturdy framework dramatically elevated our iteration velocity and confidence. We carried out over 30 distinct iterations, systematically testing main variable adjustments in our Vector Search answer, together with:

• Completely different embedding fashions (foundational, open-weights, and third occasion through API)
• Varied textual content combos to feed into the fashions
• Completely different question modes (ANN vs Hybrid)
• Testing totally different columns for hybrid textual content search
• Adjusting thresholds for vector similarity
• Experimenting with separate indexes for various supply sorts

The analysis framework remodeled our growth course of, permitting us to make data-driven selections quickly and validate potential enhancements with excessive confidence earlier than exposing them to customers.

The seek for the most effective off-the-shelf mannequin

Following the preliminary broad check that confirmed disappointing engagement outcomes, we shifted our focus to exploring the efficiency of particular fashions recognized as promising throughout our offline analysis. We chosen two third-party embedding fashions for manufacturing testing, accessed securely by AI Gateway. We carried out short-term, iterative exams in manufacturing (lasting a number of days) with these fashions.

Happy with the preliminary outcomes, we proceeded to run an extended, extra complete manufacturing check evaluating our main third-party mannequin and its optimized configuration in opposition to the legacy system. This check yielded combined outcomes. Whereas we noticed total enhancements in engagement metrics and efficiently eradicated the detrimental impacts seen beforehand, these positive factors have been modest—largely single-digit proportion will increase. These incremental advantages weren’t compelling sufficient to totally justify an entire substitute of our present search expertise.

Extra troubling, nevertheless, was the perception gained from our granular evaluation: whereas efficiency considerably improved for sure search queries, others noticed worse outcomes in comparison with our legacy answer. This inconsistency offered a major architectural dilemma. We confronted the unappealing selection of implementing a posh traffic-splitting system to route queries primarily based on predicted efficiency—an method that will require sustaining two distinct search experiences and introduce a brand new, advanced layer of rule-based routing administration—or accepting the constraints.

This was a crucial juncture. Whereas we had seen sufficient promise to maintain going, we would have liked extra important enhancements to justify totally changing our homegrown search system. This led us to start fine-tuning.

Advantageous-tuning: customizing mannequin habits

Whereas the third-party embedding fashions explored beforehand confirmed technical promise and modest enhancements in engagement, additionally they offered crucial limitations that have been unacceptable for a long-term answer at Ibotta. These included:

1. Lack of ability to coach embedding fashions on our proprietary supply catalog
2. Problem evolving fashions alongside enterprise and content material adjustments
3. Uncertainty concerning long-term API availability from exterior suppliers
4. The necessity to set up and handle new exterior enterprise relationships
5. Community calls to those suppliers weren’t as performant as self-hosted fashions

The clear path ahead was to fine-tune a mannequin particularly tailor-made to Ibotta’s information and the wants of our Savers. This was made attainable due to the hundreds of thousands of labeled search interactions we had gathered from actual customers through our LLM-as-a-judge course of inside our customized analysis framework. This high-quality manufacturing information grew to become our coaching gold.

We then launched into a methodical fine-tuning course of, leveraging our offline analysis framework extensively.

Key components have been:

• Infrastructure: We used AI Runtime with A10s in a serverless surroundings, and Databricks ML Runtime for classy hyperparameter sweeping.
• Mannequin choice: We chosen a BGE household mannequin over GTE, which demonstrated stronger efficiency in our offline evaluations and proved extra environment friendly to coach.
• Dataset engineering: We constructed quite a few coaching datasets, together with producing artificial coaching information, finally deciding on:
  • One constructive end result (a verified good match from actual searches)
  • ~10 detrimental examples per constructive, combining:
    • 3-4 “exhausting negatives” (LLM labeled, human-verified inappropriate matches)
    • “In-batch negatives” (sampling of outcomes from unrelated search phrases)
• Hyperparameter optimization: We systematically swept issues like studying fee, batch dimension, period, and detrimental sampling methods to seek out optimum configurations.

After quite a few iterations and evaluations throughout the framework, our top-performing fine-tuned mannequin beat our greatest third-party baseline by 20% in artificial analysis. These compelling offline outcomes offered the boldness wanted to speed up our subsequent manufacturing check.

Search that drives outcomes—and income

The technical rigor and iterative course of paid off. We engineered a search answer particularly optimized for Ibotta’s distinctive supply catalog and consumer habits patterns, delivering outcomes that exceeded our expectations and supplied the pliability wanted to evolve alongside our enterprise. Based mostly on these robust outcomes, we accelerated migration onto Databricks Vector Search as the inspiration for our manufacturing search system.

In our last manufacturing check, utilizing our personal fine-tuned embedding mannequin, we noticed the next enhancements:

• 14.8% extra supply unlocks in search.
  This measures customers choosing provides from search outcomes, indicating improved end result high quality and relevance. Extra unlocks are a number one indicator of downstream redemptions and income.
• 6% improve in engaged customers.
  This exhibits a better share of customers discovering worth and taking significant motion throughout the search expertise, contributing to improved conversion, retention and lifelong worth.
• 15% improve in engagement on bonuses.
  This displays improved surfacing of high-value, brand-sponsored content material, translating immediately to raised efficiency and ROI for our model and retail companions.
• 72.6% lower in searches with zero outcomes.
  The numerous discount means fewer irritating experiences and a significant enchancment in semantic search protection.
• 60.9% fewer customers encountering searches returning no outcomes.
  This highlights the breadth of impression, displaying that a big portion of our consumer base is now constantly discovering outcomes, enhancing the expertise throughout the board.

Past user-facing positive factors, the brand new system delivered on efficiency. We noticed 60% decrease latency to our search system, attributable to Vector Search question efficiency and the fine-tuned mannequin’s decrease overhead.

Leveraging the pliability of this new basis, we additionally constructed highly effective enhancements like Question Transformation (enriching imprecise queries) and Multi-Search (fanning out generic phrases). The mix of a extremely related core mannequin, improved system efficiency, and clever question enhancements has resulted in a search expertise that’s smarter, quicker, and finally extra rewarding

Question Transformation

One problem with embedding fashions is their restricted understanding of area of interest key phrases, corresponding to rising manufacturers. To deal with this we constructed a question transformation layer that dynamically enriches search phrases in-flight primarily based on predefined guidelines.

For instance, if a consumer searches for an rising yogurt model the embedding mannequin won’t acknowledge, we will remodel the question so as to add “Greek yogurt” alongside the model identify earlier than sending it to Vector Search. This supplies the embedding mannequin with needed product context whereas preserving the unique textual content for hybrid search.

This functionality additionally works hand-in-hand with our fine-tuning course of. Profitable transformations can be utilized to generate coaching information; for example, together with the unique model identify as a question and the related yogurt merchandise as constructive leads to a future coaching run helps the mannequin study these particular associations.

Multi-Search

For broad, generic searches like “child,” Vector Search may initially return a restricted variety of candidates, probably filtered down additional by focusing on and finances administration. To deal with this and improve end result variety, we constructed a multi-search functionality that followers out a single search time period into a number of associated searches.

As a substitute of simply looking for “child,” our system mechanically runs parallel searches for phrases like “child meals,” “child clothes,” “child drugs,” “child diapers,” and so forth. Due to the low latency of Vector Search, we will execute a number of searches in parallel with out rising the general response time to the consumer. This supplies a much wider and extra numerous set of related outcomes for wide-ranging class searches.

Classes Discovered

Following the profitable last manufacturing check and the complete rollout of Databricks Vector Search to our consumer base – delivering constructive engagement outcomes, elevated flexibility, and highly effective search instruments like Question Transformation and Multi-Search – this venture journey yielded a number of precious classes:

1. Begin with a proof of idea: The preliminary hackathon method allowed us to shortly validate the core idea with minimal upfront funding.
2. Measure what issues to you: Our tailor-made 50-metric analysis framework was essential; it gave us confidence that enhancements noticed offline would translate into enterprise impression, enabling us to keep away from repeated stay testing till options have been actually promising.
3. Do not leap straight to fine-tuning: We discovered the worth of totally evaluating off-the-shelf fashions and exhausting these choices earlier than investing within the better effort required for fine-tuning.
4. Accumulate information early: Beginning to label information from our second experiment ensured a wealthy, proprietary dataset was prepared when fine-tuning grew to become needed.
5. Collaboration accelerates progress: Shut partnership with Databricks engineers and researchers, sharing insights on Vector Search, embedding fashions, LLM-as-a-judge patterns, and fine-tuning approaches, considerably accelerated our progress.
6. Acknowledge cumulative impression: Every particular person optimization, even seemingly minor, contributed considerably to the general transformation of our search expertise.

What’s subsequent

With our fine-tuned embedding mannequin now stay throughout all direct-to-consumer (D2C) channels, we subsequent plan to discover scaling this answer to the Ibotta Efficiency Community (IPN). This is able to carry improved supply discovery to hundreds of thousands extra consumers throughout our writer community. As we proceed to gather labeled information and refine our fashions by Databricks, we imagine we’re nicely positioned to evolve the search expertise alongside the wants of our companions and the expectations of their clients.

This journey from a hackathon venture to a manufacturing system proved that reimagining a core product expertise quickly is achievable with the suitable instruments and assist. Databricks was instrumental in serving to us transfer quick, fine-tune successfully, and finally, make each search extra rewarding for our Savers.