All Case Studies

SEO Made Simple With AI

SEO Made Simple With AI

Designed an AI-powered SEO tool, increasing feature usage and retaining at-risk clients.

Designed an AI-powered SEO tool, increasing feature usage and retaining at-risk clients.

Designed an AI-powered SEO tool, increasing feature usage and retaining at-risk clients.

35%

Increase in SEO customization usage

80%

At-risk clients retained

7/10

Users report SEO ranking improvement

Role

Product Design

Team

1 designers, 1 PM, 5 developers

Timeline

3 months

Contents

01

Context

02

Problem

03

Solution

04

Impact

05

Process

06

Next Step

01

Context

This project set out to improve the SEO customization feature of a SaaS e-commerce operations platform. The platform gives store owners a single place to manage their online stores, from products to orders to customer relationships. SEO customization has long been one of its stronger selling points, helping online stores shape how their products show up in search.

02

Problem

The SEO customization feature was designed for SEO specialists, and it assumes a level of expertise that most store staff don't have. The current interface left non-expert users unsure how to work on SEO. And most stores couldn't afford to hire specialists.

On top of that, competitors were offering simpler SEO tools, which gave the business a real reason to worry about losing clients.

What users say

/3/

I knew SEO mattered, but I had no idea what my score even meant, let alone how to improve it.

Store Owner

Runs a small e-commerce shop

What stakeholders say

/3/

We were losing deals to competitors who made SEO simple. It was becoming a real reason people chose someone else.

Sales Manager

Sells the product to new clients

03

Solution

Below are the key design decisions that shaped this project's outcomes. For each one, I first focused on solving the underlying challenge, then weighed it against constraints.

Simplify SEO with AI assistance

Challenge

SEO scores meant little to users who weren't SEO experts. Raw scores and technical breakdowns left them stuck, unsure what any of it meant or where to start.

Constraint

AI-generated information isn't always accurate. And letting users rely on it completely to make SEO decisions was risky.

Design Decision

I turned the raw SEO score into a short list of actions, guiding users on what to do next instead of leaving them with a number they didn't understand. Alongside it, an AI-powered search let users ask "why" and get an answer based on their store's own context, turning the feature into something that taught SEO, not just scored it. To limit the risk of AI inaccuracy, I kept it in an assistant role rather than a decision-maker.

SEO became simple for non-experts to handle

Guide users with field-level AI

Challenge

Knowing which actions to take didn't mean knowing how to execute them. Users might know their product description needed work but still have no idea what to write.

Constraint

The feature wasn't monetized, so AI token usage had to stay minimal. Long, open-ended conversations weren't sustainable within that budget.

Design Decision

Since usage had to remain minimal, I attached a focused AI assistant to individual editable fields rather than an open conversation. Users make a single request, and the AI acts on it directly, providing targeted help without the cost of back-and-forth dialogue.

Users got targeted AI help where needed

04

Impact

35%

Increase in SEO customization usage

Users adopted the AI assistant quickly, engaging with SEO customization far more than before.

80%

At-risk clients retained

Sales used the new feature to win back clients who were ready to cancel over poor SEO performance.

7/10

Users report SEO ranking improvement

Users who adopted the AI assistant said their products moved up in search, landing on the first page of Google.

05

Process

User-centric approach to improve an existing feature

The problem started with the user experience, not the technology, so I kept users at the center of every stage. My process moved from discovering who was actually using the feature, to understanding how they used it, to co-designing the solution with them, to gathering their feedback on prototypes before anything was built.

User Discovery

Survey

Persona Identification

I surveyed 30 existing users to understand who was actually using it and how. The results confirmed the mismatch: most respondents were store owners or managers, not SEO specialists. Over half found the feature difficult to use, and more than a third had considered giving up on SEO.

User Discovery

Survey

Persona Identification

I surveyed 30 existing users to understand who was actually using it and how. The results confirmed the mismatch: most respondents were store owners or managers, not SEO specialists. Over half found the feature difficult to use, and more than a third had considered giving up on SEO.

User Discovery

Survey

Persona Identification

I surveyed 30 existing users to understand who was actually using it and how. The results confirmed the mismatch: most respondents were store owners or managers, not SEO specialists. Over half found the feature difficult to use, and more than a third had considered giving up on SEO.

Contextual Inquiry

Observation

Interviews

I ran contextual inquiry sessions with 8 store owners and managers, observing and interviewing them as they went through their workflow. The findings split into two parts.

First, users felt lost and didn't know what to do next. The SEO score read as just a number to them, and they got lost in the jargon with no clear next step, so they resorted to guessing and testing.

Second, users leaned on tools like ChatGPT and Gemini for guidance, but the workflow was inefficient. They had to repeatedly re-explain their business and product context just to get a relevant, useful answer.

Contextual Inquiry

Observation

Interviews

I ran contextual inquiry sessions with 8 store owners and managers, observing and interviewing them as they went through their workflow. The findings split into two parts.

First, users felt lost and didn't know what to do next. The SEO score read as just a number to them, and they got lost in the jargon with no clear next step, so they resorted to guessing and testing.

Second, users leaned on tools like ChatGPT and Gemini for guidance, but the workflow was inefficient. They had to repeatedly re-explain their business and product context just to get a relevant, useful answer.

Contextual Inquiry

Observation

Interviews

I ran contextual inquiry sessions with 8 store owners and managers, observing and interviewing them as they went through their workflow. The findings split into two parts.

First, users felt lost and didn't know what to do next. The SEO score read as just a number to them, and they got lost in the jargon with no clear next step, so they resorted to guessing and testing.

Second, users leaned on tools like ChatGPT and Gemini for guidance, but the workflow was inefficient. They had to repeatedly re-explain their business and product context just to get a relevant, useful answer.

Co-Design Workshop

Affinity Mapping

Card Sorting

I brought the same 8 users into a co-design workshop, guiding them through affinity mapping to capture everything they wanted from an SEO tool onto notes. From there, card sorting grouped those notes into two clear areas: improving SEO and learning SEO. Users wanted clear, guided steps to improve their score, and practical knowledge specific to their own products, not generic SEO info.

Co-Design Workshop

Affinity Mapping

Card Sorting

I brought the same 8 users into a co-design workshop, guiding them through affinity mapping to capture everything they wanted from an SEO tool onto notes. From there, card sorting grouped those notes into two clear areas: improving SEO and learning SEO. Users wanted clear, guided steps to improve their score, and practical knowledge specific to their own products, not generic SEO info.

Co-Design Workshop

Affinity Mapping

Card Sorting

I brought the same 8 users into a co-design workshop, guiding them through affinity mapping to capture everything they wanted from an SEO tool onto notes. From there, card sorting grouped those notes into two clear areas: improving SEO and learning SEO. Users wanted clear, guided steps to improve their score, and practical knowledge specific to their own products, not generic SEO info.

User Feedback on Prototypes

Prototype

Usability Testing

After the first prototype was built, I designed a usability test where users were asked to use the new SEO tool to improve a product's SEO.

Users were mostly happy with the updates. The main feedback for improvement was that too much information and too many actions were presented on the UI at once, making it overwhelming and hard to focus. Occasionally, the AI also gave irrelevant responses.

I took that feedback and refined the final design before release.

User Feedback on Prototypes

Prototype

Usability Testing

After the first prototype was built, I designed a usability test where users were asked to use the new SEO tool to improve a product's SEO.

Users were mostly happy with the updates. The main feedback for improvement was that too much information and too many actions were presented on the UI at once, making it overwhelming and hard to focus. Occasionally, the AI also gave irrelevant responses.

I took that feedback and refined the final design before release.

User Feedback on Prototypes

Prototype

Usability Testing

After the first prototype was built, I designed a usability test where users were asked to use the new SEO tool to improve a product's SEO.

Users were mostly happy with the updates. The main feedback for improvement was that too much information and too many actions were presented on the UI at once, making it overwhelming and hard to focus. Occasionally, the AI also gave irrelevant responses.

I took that feedback and refined the final design before release.

06

Next Step

Right now, users still review and act on SEO suggestions one product at a time, which gets harder as a catalog grows. I want to explore letting the AI handle that at scale, scanning a store's full catalog and flagging what needs attention, with users staying in control of what actually gets changed. It's still early, but it feels like the natural next step for stores with larger catalogs.

Next

Centralize E-commerce Operations

Let’s start a real conversation

hello@dstudio.info

Vancouver, Canada

Let’s start a real conversation

hello@dstudio.info

Vancouver, Canada

Let’s start a real conversation

hello@dstudio.info

Vancouver, Canada