Into the Heart of a UX-driven Knowledge Graph: IKEA’s RDF Way Forward

How is fitness related to a bench? What is suitable for small spaces and can fit by both a sofa and a bed, serving as table but also being flexible to function as a bedside table? And what is a relevant product to complement a bed? 

Imagine all these questions answered by a furniture website. In one form or another – either through straightforward answers (the Google’s AI Overview way) or by implicitly meeting the intent by recommending their answers as related products?

Wouldn’t that be the magic wand for e-commerce success – being able to match the intent of the user to your product base and not only, to the domain knowledge embedded in the content of the website.

It would. 

And this is exactly what IKEA’s knowledge graph does, along with all the other capabilities that IKEA’s knowledge, modified in RDF data, brought to the company.

The magic wand is a semantic knowledge graph.

IKEA’s Knowledge Graph

I first learned about Ikea’s knowledge graph at SEMANTiCS 2022, where Adam Keresztes presented his UX designer perspective on how the graph was built and most importantly why in his UX Design & Knowledge Graphs – The Perfect Match”. Thankfully the talk is recorded and available to watch in the conference’s channel:

Technically, the IKEA knowledge graph consists of “three layers, which are concepts, categories, and data. For example, in IKEA a concept would be a product, a category would be a bookcase, and one data point would be the BILLY bookcase in white 80x28x202 cm”[1].

What is more interesting through is the socio-technical aspect of IKEA’s knowledge graph – the strands that weave it, as is always the case for successful technical assemblages are tightly intertwined with human knowledge, procedures and practices. 

Think knowledge-graph aware domain experts, subject matter experts that are taught what ontology is and how it applies to their job, product data analytics coupled with UX design creativity.

Case in point, Katarina Kari shares how she and her team involved domain experts in the making of the IKEA knowledge graph (Watch What Over 7 Years of Building Enterprise Knowledge Graphs Has Taught Me About Theory and Practice)

Also recently, at the wonderful Connected Data London 2024,  Christelle Maignan, currently Lead Ontologist, Inter IKEA Group also presented the IKEA knowledge graph.

I

She talked about the graph not only as an enabler for ecommerce but also as a platform – a means towards improving data accessibility and also understanding and collaboration within the organization and across different teams. Watch the talk from Connected Data London’s Archive

A Knowledge Graph, but so what?

Creating digital customer experiences heavily relies on data. Yet knowledge graphs are not just about gathering and linking data. They are about practices that allow us to capture, codify, curate and ultimately share knowledge. 

And in IKEA’s knowledge graph case it is that domain knowledge of designers and experts that is stored in a knowledge graph to be further translate into better search, navigation and exploration experience on the company’s website.

Now let’s look at IKEA’s knowledge graph from the perspective of Tim, my imagined Webonaut who benefits from knowledge graphs couples with content and providing knowledge experiences.

From the perspective of Tim experiencing content about IKEA’s products, the knowledge graph enables relevant recommendations, search according to various criteria, get content that is highly relevant to him and also be able to serendipitously browse across products based on his topic of interest, without having to pay for that with his personal data [2].

Tim gets to browse complementary products, their benefits, and other useful information captured on a general level.

At a very high level, Tim actually is given the opportunity to navigate the design knowledge and experience IKEA have gathered throughout the years and embedded in a knowledge graph [3]

Improvements in online sales, knowledge transfer, De-siloing data – who could ask for anything more.

IKEA’s teams did see a three-digit increase in the sales of cushion pads, linked to the reasoning capabilities of the knowledge graph (see 11:54 in the mentioned video from CDN 2024) and the work across teams improved, making the vision of de-siloing data (and knowledge) reality.

Seeing this application of semantic web technologies and sustainable, user-centric user experience design dancing together for e-commerce is inspiring.

So much to learn from the project and rejoice about all the wins big and small among diverse stakeholders, systems and social practices [4].

And so much to apply in our content and marketing communications on the Web practises and rituals of relating!

Let’s do it.

References:

1. Katariina Kari, “IKEA’s Knowledge Graph and Why It Has Three Layers – Flat Pack Tech – Medium,” Medium, August 24, 2022.  See also: Leoncio, Pauline. “Building Explainable and Trustworthy Recommendation Systems: What We Learned From IKEA at KGC 2023,” October 29, 2023. 
2. ref. The Knowledge Graph Conference. “Knowledge Graph-Driven Recommendations — Katariina Kari, Inter IKEA Systems | KGC 2023,” March 11, 2024.
3. see also KGA. “KGA Inauguration, Tane Piper, IKEA,” November 19, 2023.
4. Parts of this text first appeared in my post. “A Little SEMANTiCS Goes a Long Way.” at Ontotext’s blog. September 23, 2022.

This knowledge graph story is part of the stories I told in Being Dialogic – my book about marketing communications on the Web and the way we weave enterprise content online, through dialogue and metadata.