Ecommerce ops · Production

Kesko Senukai Digital Improves CTR From Site Search by 47%

The problem

Kesko Senukai Digital's in-house Elasticsearch site search returned irrelevant results and could only be managed by engineers, excluding the merchandising team and leading to low search usage and poor conversion rates.

First attempt

KSD's self-built Elasticsearch search solution was overly simple and required a large, expensive dedicated engineering team, making it unsustainable and unable to fulfil sophisticated merchandising and personalization requirements.

Workflow diagram · grounded in source
1
Visitor submits search query
trigger
“understand the context of user queries better and return the most relevant results at any given time”
2
NLP multilingual query processing
ai_action
“custom stemming and natural language processing (NLP) algorithms, based on the specifics and grammar rules of these languages, to create more accurate and intuitive search”
3
Custom AI relevance ranking
ai_action
“create custom algorithms, explicitly designed for KSD on top of AI-based SearchNode search platform”
4
AI merchandising optimization
ai_action
“custom merchandising algorithms, enriched with AI. This provided the company with smart merchandising solutions running, analyzing, and optimizing search results for conversions, 24/7”
5
BitREC personalization integration
integration
“We integrated the SearchNode search platform with the leading personalization tool BitREC. This helped Kesko Senukai Digital in creating more intimate experiences based on a user's browsing history”
6
Monthly A/B improvement sprints
feedback_loop
“Each month search teams from SearchNode and Kesko Senukai Digital create a list of hypotheses for search improvements. KSD product owner manages and prioritizes this list to select what should be tested. Then SearchNode engineers create …”
Reported outcome

After deploying SearchNode, CTR from site search sessions rose 47% and add-to-cart YOY rate grew 22%.
In 2020, sessions with site search surged 86% YOY and conversion rates from site search grew 56% YOY.

Reported metrics
CTR from sessions with site search47%
add-to-cart YOY rate from site search22%
sessions with site search (2020 YOY)+86% YOY
conversion rates from sessions with site search (2020 YOY)+56% YOY
Reported stack
SearchNodeNLPBitRECElasticsearch
Source
https://www.nosto.com/case-studies/kesko-senukai-digital/
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

After deploying SearchNode, CTR from site search sessions rose 47% and add-to-cart YOY rate grew 22%.

What tools did this team use?

SearchNode, NLP, BitREC, Elasticsearch.

What results were reported?

CTR from sessions with site search: 47%; add-to-cart YOY rate from site search: 22%; sessions with site search (2020 YOY): +86% YOY; conversion rates from sessions with site search (2020 YOY): +56% YOY (source-reported, not independently verified).

What failed first in this deployment?

KSD's self-built Elasticsearch search solution was overly simple and required a large, expensive dedicated engineering team, making it unsustainable and unable to fulfil sophisticated merchandising and personalization…

How is this ecommerce ops AI workflow structured?

Visitor submits search query → NLP multilingual query processing → Custom AI relevance ranking → AI merchandising optimization → BitREC personalization integration → Monthly A/B improvement sprints.

Related ecommerce ops cases