Revolutionising
Ad Management

Launched in 2015, smec reaps the first-mover benefits of using data to optimise bidding strategies for a while.

In the meantime, Google keeps releasing more and more data-driven automation. The biggest hit of it all was when Google announced their new solution, Performance Max in October 2020.

I joined smec in 2021 in its journey of searching for its new USP.  

Experiment Builder - fully set
icon-computer
Product Type
B2B
WebApp
icon-blush
Work Type
Innovation
Product Vision
icon-tool
Tools
Figma
icon-hourglass
Timeframe
Jun'22 - Sep'22

Becoming Irrelevant

Performance Max comes with a simple promise.
"We use AI to optimise everything for you. You just give us the money.

Previously, smec's USPs were about being smarter than your competitors by using data to fine-grain your ads and being efficient by automation.    
With Performance Max's promise, is there room for another software?  

My Role

As a user-focused stakeholder, I supported the VP of Product by identifying opportunities and shaping the direction of our future offerings.

Insights from User Research

I start off with assessing the situation, to understand what I was dealing with, by conducting various research.

Value proposition canvas

User Profile

From interviews and acquisition funnel analysis, I noticed that there are three user roles; Owner, Manager, and Operator. The roles are clustered by their needs.

CSM's workflow and how they interact with clients

Workflow

From a shadowing session, I mapped out the workflow of how the client and CSM interact.
The main tasks revolve around daily monitoring–making sure everything is as expected, and optimising the campaigns to meet the goals.

Mapping solutions to clients' resources

Solutions to Needs

There is a correlation between clients' resources–man-power, expertise, technology maturity–and how we make our offers more attractive to them.

Blue Ocean Strategy Canvas

Blue Ocean Strategy Canvas*

Looking at the competitive landscape.
(*This is not a comprehensive canvas. I created this mini-version only for demonstration. Without giving away too much.)

Understanding
E-Commerce

Additionally, to get a good understanding of the business, I also conducted industry expert interviews and analysed Google, Microsoft and Meta Ads. As well as reading up on industry news.

I arrived at a conclusion what is the core of e-commerce.

The Core of E-Commerce

Selling the right products(Inventory),
to the right people(Audience),
at the right place(Channel),
at the right time(Customer Journey),
and at the right price(Inventory).

basic e-commerce components

Workflow of Running a Business

On an abstract level, we can see most workflows as assessing the situation, deciding what we want to achieve, make a plan for how to get there. Act on those plans, take measures, reevaluate and update the plans. And repeat.

Plan > Act > Measure > Evaluate workflow

Choosing a Problem

Looking at the same workflow but from a data perspective, how do we go from data to action? I identified three main challenges:

Identified problems
  1. How might we enable hassle-free data integration?
  2. How might we best enable users to form hypotheses?
  3. How might we enable users to act on the hypotheses they formed?

The rest of this case study is only focusing on point 3.

How might we enable users to act on the hypotheses they formed?

Choosing a Target User

Once a problem area is identified, the next crucial question to address is: "Who are we solving this problem for?"

From user research, with our current offering, I identified three user roles: Owner, Manager, and Operator. These three roles, on detailed level, they have different needs and tasks. But on a high level they one thing in common.

It's not about managing ad campaigns. It's about having a business impact. It's about optimising things that are under their control to get the most out of it, optimising for highest return on investment. And looking competence doing it.

How does that look like in practice?

Introducing

Experiment Builder

Experiment Builder - continue range number selection

Design Principles

  1. understandability / transparency
  2. reassurance / confidence
  3. suggestive / supportive / guidance

Bringing Data Structure to UI

Experiment Builder - Customer Journey stageExperiment Builder - Data category selectionExperiment Builder - One criteria selectedExperiment Builder - and/or criteriaExperiment Builder - low-fi frameExperiment Builder - low-fi all data typesExperiment Builder - low-fi continuous number selectionExperiment Builder - low-fi multi select

Data Data Data

All kinds of favours and textures.

E-commerce data model behinds experiment builder

The End of the Beginning

Unfortunately, we had to stop the project.
Due to macroeconomic headwinds, smec decided to stop long-term-oriented product development.

If We were to Continue

A question that we have to ask ourselves, looking at the market, is why this solution has not existed yet.

The biggest risk that we would have had to investigate was its feasibility. Is it feasible and sustainable? As we would still be dependent on ad channels, and that means if they change something in their APIs or their data structure, we would have to adapt based on their timeline, which is not a good position to be.

And a big challenge with this design is usability. Interaction design in itself would be such an amazing design challenge to tackle. As well as the mindset shift required and how we can design a holistic experience to accommodate that shift.

other showcases