J Chaparro | Brandi.AI

Overview

The revolution will be personalized.

Problem Statement

How might we help marketing and sales teams to effectively personalize video assets at scale?

Background

Brandi.ai, founded by two ex-Googlers, is a startup that utilizes Natural Language Processing (NLP) models to automate ads and marketing asset creation. The founders wanted to explore the possibilities of expanding the product to create hyper-personalized video marketing assets.

Why now?

Though the use generative AI has mostly been used in the creation of static images, Brandi’s founders are thinking ahead by considering how generative AI can be utilized to personalize video marketing assets. As AGI improves and the trend towards video marketing assets continues to grow, it will be a major means for vendors to increase sales and engagement.

A quick note:

We had a mid-way focus shift that required additional research and a different approach to the MVP. This case presents the final results but will address any significant findings and decisions made before the focus-shift.

The road ahead

The main challenges, solutions, and anticipated results.

These are some of the core challenges we encountered and how we addressed them through research, design, and working closely with the stakeholders:

Video content is tough

23% of non-video marketers don't use video because they lack time and don't know how to start.¹

Match user know-how

Create a product that is accessible to users of all-levels, not just marketing specialists.

An easy-to-adopt product

AI access for everyone! Specialists, content creators, sales persons, and small business owners, and more.

Marketing is a huge space

Awareness, consideration, purchase... and more. An MVP can't tackle it all, so where do we start?

Set a clear target

By working closely with our stakeholder, we focused in on the cold-outreach stage for B2B and B2C use-cases.

A laser-focused MVP

With clear purpose, value and direction that can be used to demo functionality to potential investors.

No 1:1 competitors exist

How can we determine our requirements and features in an exploratory area of a rapidly growing space?

Leverage existing models

We don't need to reinvent the wheel. We can succeed by utilizing models and patterns that already work.

Familiar from the start

Users will have little to no learning curve, allowing them to utilize Brandi's full-power from the get-go.

A rapidly changing space

It feels like there's something new in the world of AI every day. How do we keep up and plan ahead?

Think big, but stay viable

Our challenge is to push the limits but always work closely with stakeholders to ensure viability.

Revolution + room to grow

Brandi will change cold-outreach, but that's just the beginning. There's an entire marketing journey to tackle!

Research Constraints & Objectives

There’s a lot we don’t know... how do we craft the first prompt?

Preliminary research and Stakeholder’s experience helped us identify our knowledge gaps and establish our initial research direction. From there we set our project objectives and success metrics.

Our objectives and success metrics:

Brandi must...

Allow at-scale video personalization

Users should be able to capitalize on the power of AI to create personalized video content.

Provide simple, but powerful editing features

By providing essential editing features, we can minimize the time users have to switch between platforms to make basic changes.

Be technologically viable

AI's potential feels limitless, but our time to market is not. We must be forward-thinking but keep the MVP viable to have timely development.

We'll succeed if Brandi...

Is easy to integrate & adopt

The interface can't be too technical or complex. It should be learnable and easy to integrate into existing automations and workflows.

Brings value to users

By saving them time, increasing campaign engagement, or being the reason they finally use video marketing. It has to give, not take.

Allows for user control and freedom

Users should be able to create videos their way - by recording in-platform or uploading their own, while having simple editing features.

The Outcome

A streamlined AI video platform that allows users to record, edit, and deploy marketing videos at-scale through simple but powerful features and integrations.

Research & Insights

Identifying needs, existing process, and opportunities.

Research Plan and Methodologies

We set out to learn the existing marketing process, identify needs and determine opportunities.

Our research consisted of four main phases:

Video Marketing Research
Interviews & Synthesis
Competitor Analysis and Design Review
Focus Shift: Interviews & Synthesis Round 2

Scroll ahead to the deliverables and key takeaways for a quick overview, or read away to learn more!

Our homework was to understand:

The marketing journey

We needed to learn the stages, touchpoints, and challenges beyond just a journey diagram.

Current marketing trends

Specifically, what is the current video-marketing landscape like? What can we capitalize on?

Demand for AI-personalization

Identifying market needs could help us gauge potential interest in an AI-driven solution.

Our target users

Evaluate the needs and preferences of our intended audience and market.

Video Marketing Resarch

“Personalization is the norm in marketing and there’s a shift towards hyper-personalization as digital marketing becomes more competitive”

- Deloitte Study

We looked into current video marketing trends and reports to determine what areas AI could help with, how it could do so, and who could be the target users.

These were our key takeaways:

Video is popular and effective

Videos can be 1-to-1 or 1-to-many, either way there are many surefire ways to personalize videos and increase engagement.

Shift towards Hyper-personalization

Using data, analytics, AI, and automation to send highly customized communication to specific customers at the right place and time.

Opportunity across the journey

The marketing journey only begins with outreach/awareness, there are many applications and opportunities to leverage.

Video is costly

Both in time and money. Making consistent and personalized content is hard when you don't have the time or don't know where to start.

A marketer's definition of success:

To over 60% of marketers, the success of a video is measured by engagement, number of views, and resulting leads/clicks.²

Avoid greed

To use AI to its full potential, we must think beyond reaping immediate rewards and consider factors like scalability and adaptability.

Interviews & Affinity Mapping

Our Interviews confirmed that time and money were the first big barriers to videos and personalization.

We conducted 5 initial interviews, with B2B, B2C business owners, and marketing professionals to gain insights that would inform our ideation process.

We asked questions about:

Their business/job and day-to-day
Campaign ideas and channels, and goals
Specific products or tools they use
Do they create personalized content?
Challenges they face

We held an affinity mapping session with stakeholders to organize and identify trends and determine which areas we’d tackle.

What did we learn from our interviews?

Maintaing a personal connection

Interviewees understood the value of personalized connection, but takes lots of time and effort to build and maintain.

Cost and knowledge barriers

Benefits of personalization are clear, but it’s time consuming and not cost-effective for everyone.

Technical knowledge and control

Non-creative users may not have the necessary knowledge to create more than a basic video.

Work smart, not hard

Video is expensive and may have a limited shelf-life. It’d be helpful to be able to reuse created assets.

Competitive Analysis & Design Review

We identified core and specialized features by comparing a wide-array of competitors.

Many competitors were locked behind a trail/demo or their available promotional materials left a lot of blank spaces, so we aimed for quantity! We looked at over 15 products an services that utilize AI to create and personalize video content, and platforms that allow users to create and manage video assets.

We created a competitive feature matrix and mapped out key flows for every competitor. Synthesizing this information visually helped us to identify core features, strengths, and opportunities in the video marketing space.

What made a platform stand out... for better or worse?

The flow & UI

A straightforward & uncomplicated video creation flow made it easier to pick up a new platform.

Editing capabilities

All competitors provided basic in-platform video editing features. Simple & intuitive > power features.

(Good) Templates!

can be a great starting point. Better to have a few high-quality than an excess of low-quality templates.

No analytics

Important to our interviewees, but many competitors did not support basic analytics in their platform.

“AI” Features

Some platforms use, or claimed to use AI, but it did not always perform well enough to actually be useful.

Additional research & Interviews

We pivoted from marketing asset creation to hyper-personalized videos for cold outreach. This required more research and interviews.

After presenting our research findings, our stakeholders took some time to process what we had to share, then decided they wanted to shift focus.

Initially we were working on creating and managing video assets for B2B and B2C users and marketing professionals. Now we were headed towards creating hyper-personalized videos for the cold-outreach stage with a specific MVP focus on B2B.

We conducted two additional B2B interviews and looked into 6 additional competitors with a focus on AI features that would make the creation of cold-marketing videos easier.

Easy wins and opportunities we identified:

Script & video templates

Some interviewees write highly personalized cold-emails - scripts and even video templates could save them a lot of time.

Clear instructions and tutorials

Using variables to personalize may be a novel experience for most users. It'll be important to really nail the experience.

Streamlined from creation to deployment

Most direct competitors we looked at are limited to video personalization and generation - can we provide a start-to-finish solution?

What existing weaknesses could turn into opportunities?

Avoid extra work

Some competitors required recording additional sound clips to dynamically personalize videos... it felt less like AI and more like extra work.

No middle ground in variables

Identifying and populating variables felt either oversimplified and limited, or overly technical and time-consuming. Could we do better?

Synthesis & Strategy

Establishing Brandi's roadmap.

Target-Users & Journeys

B2B users would be our priority user, but it was still important to consider B2C and Marketing Creatives as well.

We brought together all of our research to create 3 distinct user personas and their user journeys. Together, they represented a B2B Sales Lead, a B2B marketing creative, and a B2C small-business owner. Journeys helped us empathize with how they’d use Brandi and consider what features would be important for specific users.

Though Adam would be our main focus for the MVP, providing all three would bring value to the stakeholders in post-MVP development.

Feature Prioritization & Architecture

Our high-priority features were: in-platform video recording and editing, variables and data mapping, and email/landing page templates.

We used insights from competitive research and persona+journey building to identify the most relevant features. We held a session with our stakeholder to prioritize for the MVP. The resulting features were then used to draft a sitemap of the platform to help us conceptualize where features would show up.

Flows

We used prioritized features + sitemap to map our flows for clearer ideation.

Mapping flows out this way gave me an idea of what screens I’d have to design. The key tasks for a climber using the MVP would be: 

Record and edit a video
Define variables and connect to CRM data
Create a new campaign from existing campaign
Insert video into marketing email / landing page and deploy

Concept

Turning our synthesis into a first-draft.

Touchpoints

Brandi's main touchpoints will be on desktop and inboxes!

Ideation & Wireframes

All designers sketched out ideas. We brought them together and drafted rapid lo-fi versions to share with stakeholders.

Our main tasks were to consider how relevant campaign and asset info would be displayed, how and when users would define variables, and how the results would be deployed.

We iterated on with a focus on interactions patterns like drag-and-drop templatized building, highlight to open context menus, and ease of use features like video teleprompter and script editing.

Image of Brandi video recorder and editor sketches

An image of low-fidelity wireframes showing a dashboard and video recording screens.

An image of low-fidelity wireframes showing video editing, variable mapping, and email template screens.

The cutting room floor

After a second round of prioritizing, my email & landing page builders were left for a later stage. I focused my efforts on video editing and variable screens.

After sharing our initial lo-fi wireframes, we consulted with the stakeholder and decided the deployment flow and screens was too much for the MVP stage... I had to say goodbye!

We instead focused on the video asset creation with a simplified deployment that would leverage existing integrations. I’ve included some of the wireframes so you can have a peek at where Brandi is likely to go in the future. ;)

Visual System

We created a design system to aid in rapid development.

We created simple branding and component kit for Brandi. We stuck to essential components and styles to minimize fluff and allow for a more rapid development stage.

Design

Personalized outreach at-scale

Hi-Fidelity Design

How would a user create a video using Brandi?

We focused our design around the key tasks and flows to create a straight-forward and intuitive approach to creating a marketing video completely within Brandi.

Recording a Video

Easy in-platform recording

Users can upload their own video file, or record within the platform. Helpful features like template scripts, a teleprompter, and multiple takes make it a breeze to record within Brandi.

[data-wf-bgvideo-fallback-img] { display: none; } @media (prefers-reduced-motion: reduce) { [data-wf-bgvideo-fallback-img] { position: absolute; z-index: -100; display: inline-block; height: 100%; width: 100%; object-fit: cover; } }

A high fidelity wireframe image of Brandi.ai's video editing screens.

Editing a Video

No fuss, intuitive editing.

Users can make simple, but effective edits to their video within Brandi. They can trim, add subtitles, and choose their thumbnail, saving them time spent between platforms and having to download/upload to make edits.

Adding Variables & Data

Magic: The Variables!

Brandi automatically generates an editable transcript for every video, this transcript is then used to define variables to receive data. Users can keep track of their variables using the panel on the right.

At scale in 3 steps:

Select a word or phrase in the transcript, give it an @variable name.
Upload a dataset and map it to @variables that were defined.
Hit preview ~ if it looks good, generate away!

A high fidelity wireframe image of Brandi.ai's video preview screen.

Preview & Generate

Sit back and let Brandi do the heavy lifting!

Brandi generates an example personalized video using the variables and data the user provided. This acts as a quality check and confidence boost that ensures users that the video quality will up to par. And if it’s not, they can still make edits or try again.

From there, users don’t have long to wait. They are provided with quick-options or will receive a notification once their videos are generated if they move to a different page.

Project Dashboard

Everything in its right place

Users have access to engagement metrics, recipients and variables used. From here they can also re-upload data to amend errors, deploy via third-party integrations, or use the project as a template to create a new campaign!

Usability & Iteration

It was time to put our prototype to the test.

We conducted moderated testing with 5 users ranging from marketing designers, B2B SDR’s, and Marketing VPs.

Flows we tested:

Recording, editing, and mapping data to a video.
Exporting video CSV links.
Creating a follow-up video and replacing the dataset.

Successes

User's familiarity with other marketing platforms made Brandi intuitive.

Users had previous experience recording, editing, data upload and exporting in other platforms, conducting these tasks in Brandi was a breeze.

5/5 users

Found video recording and editing intuitive

It was easy to set up and use of the teleprompter and conduct simple video edits to the recorded video.

4/5 users

Uploaded and mapped user data easily

Marketing platforms and user datasets go hand-in-hand, and users felt right at home using them in Brandi.

Confidence

In the effectiveness of Brandi's AI

Using the first data entry to generate a preview video made users confident that the AI would do a good job.

5/5 users

Exported their video data without a hitch

Exporting the CSV of video links was straightforward and familiar task to all users, they had no trouble with this task in Brandi.

Improvements

There were a few interactions to tidy up, but most of the trouble was coming from my work... the variables.

Though some popular marketing platforms provide variable capabilities, they did not seem to be a common part of our tester's workflows. We had some work to do to make defining variables (and a few other features) more intuitive.

Adding Variables & Data

Adding data was easy, but defining variables was confusing.

For starters... the instructions were towards the bottom, making them easy to miss. Some users uploaded data first and then expected to drag and drop the variables from the panel into their transcript. But we did learn that the common mental model was define variables → add data, so we redesigned to reinforce that model!
‍
Here's what changed:

Added a first-time tutorial modal.
Revised the variable instructions copy.
Moved those instructions to the top!

Interaction & Action

The teleprompter and editing features were intuitive, but applying them was not.

This was more of a subtle observation, but users occasionally skipped the “apply” button. It was too similar to the “next” button, so we made some visual and proactive changes.
‍
What changed?

Changes made to the teleprompter script were automatically applied.
The video editing buttons were simplified to stand apart from the “next” buttons.

Project Dashboard

Duplicating a project was easy, but replacing data was a chore.

Users received confirmation that their video has been duplicated... but then they were left on their own to figure out how to change the dataset.
‍
To resolve this issue, we:

Added a prompt allowing users to use the same data or upload a new set.
Made the “Replace Data” button accessible in every project page, instead of burying it in a specific tab.

Additional Testing

Small changes, big impact!

We conducted a quick round of testing with two new participants to see if our changes had the desired impact.

100%

First-try success in defining variables .

The variable tutorial modal was quickly skimmed by the users and led to no issues when defining variables.

All users

Replaced data without an issue.

The prompt made their options clear, from there mapping data was as intuitive as before.

“I don’t have anything to compare it to, because we don’t use that kind of process/technology yet.
But it was fairly easy and very understandable.”

”I don’t feel there was a learning curve,
it was very on point.”

Outcome & future Roadmap

Big news... Brandi is now in development!

Impact

We delivered an MVP Prototype that was successfully used to demo functionality and secure investment!

But don’t be fooled! Our impact doesn’t stop with the MVP. Here's how our work continues to impact Brandi's current and future business:

Delivered an initial and future direction

Research and collaboration gave us the opportunity to shape the product's initial direction. Everything that didn't make the MVP is an opportunity to be explored.

The Prototype is a resource

Our prototype was instrumental in securing investment and will also aide development as a resource for functionality and interaction.

A Foundational system

Our brand and design system kept it simple to aide rapid development. It will be easy to build upon as Brandi adds and builds out new features.

Future Roadmap

What would we work on next?

This is only the beginning for Brandi, here's what would be next if we were to continue:

Design the deploy-via-integration flow

Third party marketing integrations are common across platforms. We included the capabilities but chose not to prototype them. It would be important to design and prototype these functions next.

Conduct more extensive testing of the dashboard

Most of our user-testing tasks were for the video flows. Future rounds of testing should focus on the dashboard. Especially after designing integration flows.

My Learnings & Challenges

What went well? What would I want to do better next time?

The beauty and challenge of this project is how open it was.

We played a key role in shaping its initial direction, which wasn’t always a cake-walk. It provided great opportunity to collaborate and defend our design decisions with insights and research as a team of designers.

Keep feature-creep and ambition in check

We wanted to go big from the start, and it almost spiraled into a feature avalanche. Though the features we cut will be very useful in later development, they were not high-priority for the MVP.

Regular stakeholder sessions are gold

Our regular meetings and brainstorming sessions with stakeholders really helped keep everything in check and on a realistic timeline for development.