Upzel – Audience Management on an Advertising Dashboard

2016 – UI-UX Design

Design Brief / Problem Statement

Although content reigns supreme in the media, data reigns supreme in advertising. Important information comes in a variety of shapes and sizes, so be sure you don’t leave any cookies untouched!What may this data be used for if you want to advertise to people based on their behaviour? How can you find out who is talking positively or negatively about your business and use that information to develop actionable insights and create ads for a specific audience? The Adfox Data Science team, in collaboration with NID, is working on a product to solve this problem by classifying data into a taxonomy that makes sense for business and advertising activities.

Secondary Research


The form of advertising that uses the internet to distribute promotions to consumers is known as online advertising. Companies can track people’s online habits and behaviours in order to target adverts to individuals who are most likely to buy a specific product or service. This provides many advantages to digital advertising over traditional types of advertising such as print, radio, and television. For starters, because digital ads can be measured, they can deliver a better return on investment. This helps you to focus on what works and less on what doesn’t before investing a lot of money on an ineffective ad. Because you may target regionally, by product interest, or even by online habits, digital ads can be more effective. It can also be less expensive than traditional means of advertising, allowing anybody to participate, from the smallest corner shop to Fortune 500 companies.

As a first step existing articles and books available were referred and prepared notes to know more about the domain.

Its method of placing the ads within search results. Search is a must for any small business loIts strategy for displaying advertisements in search results. Any small business wishing to offer a product or service locally should conduct a search. It’s cost-effective, and you won’t have to compete with large corporations with large advertising expenditures.

You may be more imaginative and surely more graphic with display ads. They’re the modern-day skyscraper, square, or button advertisements you find on some of your favourite websites.

Rich Media
Ads with video, audio, or other interactive features are known as rich media. Because the advertising encourage viewers to connect and engage with the content, this can be particularly beneficial for prominent businesses looking to boost exposure.

Ad Retargeting

You may have noticed adverts for bicycles cropping up everywhere you go online if you’ve ever gone to a website wanting to buy something, such as that new bicycle you’ve had your eye on, but opted not to make the purchase. It’s known as remarketing or ad retargeting. Websites include code that informs internet ad networks that you were interested in purchasing a specific product or service. The ad network will display advertisements for that goods or service all across the internet. These commercials are usually display ads, and they are effective for both creating awareness and selling products.

Native Advertising

Sponsored content graphically designed to have the same look and feel as the website where they show up are known as native ads. You can think of it as the digital version of the old print advertorial. There are two types of native ads, display and social, that can both serve to raise awareness.

Email advertising

Advertising by sending emails to potential target audience.

Evolution of online advertising

1.Site based targeting
  • Display ads based on the content of the website.
2.Standard Behavioural Targeting
  • Display ads based on users previous online behaviour and browsing history.
  • Good for specificity
3.Advanced Behavioural Targeting
  • Display ads based on insights from multiple data sources.
  • Good for specificity and scale
4.NLP + AI Behavioural Targeting
  • Display ads based on behavioural data collected based on NLP algorithms.
  • Good for specificity scale and accuracy.
  • Better ROI

Knowing your customer

User Actions


The user is merely interested in your product / service, such as reading an article or watching a video related to your product / service.


The user has an intent to buy the product/service. He might be viewing a product page, conducting a search, or customizing a product / service.


The user has purchased the product / service or submitted their contact information in the website for further enquiry.

The user is about to complete a purchase. The user has added the product / service to their cart, began to fill out the lead gen form, etc

Data in advertising

First party data is the data collected from customers directly. Its the most valuable form of data. Owned by the brand. CRM, Social media data, Subscriptions are examples of these.

Second party data is just the First party data collected from other source.

Third party data is the data aggregated from other websites and platforms.

An in-depth analysis of first-party and third-party demographic, contextual, and behavioural data about consumers and campaigns will provide answers to questions about customer knowledge. Marketers, agencies, and publishers can utilise this information to send tailored messages to customers at the precise point in the purchasing process when they are ready to buy.

Targeting the audience

In the case of standard targeting, the entire segment is targeted regardless of intelligence data. Meanwhile, while using big data to create user segments helps to target appropriate consumers to some extent, certain target users are left out. When compared to traditional targeting, this gives a higher return on investment. By studying individual visitors and their unique behavioural histories, the NLP-based system builds patterns relevant to a campaign. Then, using Artificial Intelligence, overlay that behavioural data with keywords retrieved using an NLP engine. This results in pinpoint accuracy as well as complete, auditable openness.

Demographic targeting is about reaching people based on their personal characteristics

  • Age
  • Gender
  • Language
  • Education
  • Ethnic Affinity
  • Generation
  • Home
  • Life events
  • Parents
  • Relationship
  • Work

Interest is information on people’s interests, hobbies, websites visited and Pages they like on social media. Companies like Facebook uses this information to build targeting options for advertisers to use.

  • Business and industry
  • Entertainment
  • Family and relationships
  • Fitness and wellness
  • Food and drink
  • Hobbies and activities
  • Shopping and fashion
  • Sports and outdoors
  • Technology

Behaviours are activities that people do on or off social media They are constructed from both of someone’s activity online or offline activity provided by data from trusted third-party partners.

Lookalikes are audiences that are similar to people who visited your website, given your their contact or personal information, visited your app or taken an event action.

Cold Audience

A cold audience is any audience that has not interacted with your company. They know nothing about you, and are unlikely to make a purchase right away. Advertisers spend the majority of their money targeting cold audiences.

Warm Audience

A warm audience is a group that has some sort of interaction with the brand. They have taken an action related to your online or offline properties. That action is measurable and trackable online.


Anyone who has purchased your product. Knows your brand well. There is a chance of moving away from your brand in course of time

With millions on data points the system identifies the gender using regression modelling.

How NLP works now?

One of the most natural methods to convey your query is as a natural language question or request to another human being. Search engines employs a combination of text analytics and human moderation to provide a question-answering search experience, popularised this type of engagement. The killer search feature has been natural language processing (NLP), but many have failed owing to the inherent difficulty of creating solid algorithms for natural language processing. It also reflects the characteristics of the search experience itself: we need to promote an open, scalable, and interactive discourse to successfully assist human information-seeking across the largest range of work situations.

How an NLP API interprets a sentence.
I tried chatting with a chatbot to understand how it worked.

NLP in chatbots

Most chatbots are powered by NLP and offer great customer experience.

  • Immediate assistance: Chatbots never sleep and never put you on hold. They are available 24 hours a day, seven days a week for real time interaction.
  • More efficient service: Chatbots can find the precise answer customers need in any connected knowledge base.
  • Human-like engagement: Like a human assistant, the chatbot offers a personalized, one-to-one experience, in a conversational style..
  • Cost and time savings: Chatbots can handle many questions without human intervention, and allows operators to focus on more complex activities.

Pixel Audience

Primary Research

The primary research began with determining which research methodologies would best suit the knowledge-gathering objectives. The goal of the Generative Research phase was to define the problem, and the Evaluative Research phase was to test the design later. The initial idea was to interview potential Upzel users as part of a stakeholder interview process. Upzel’s User Experience was the subject of a questionnaire. The users’ contributions were narrowed down using Generative Analysis. After gathering user feedback, sticky notes were used to organise it rationally.The main research objective was to understand how the users classify the third party data. To gather general information about end-users and to know about the difficulties which they currently face while using the data. To collect their answers based on which the overall user experience of Upzel can be enhanced.

  • Can you please describe your typical day at work?
  • What are the typical activities you do?
  • Which tasks or activities take up most of your time?
  • What types of customer data do you handle daily?
  • How do you organize your customer data?
  • What are your pain points while handling customer data?
  • While creating a user segment for advertising what’s most important thing you look for?
  • What functionalities would you like to have to solve your problems better?
  • Have you used any other product that solves your problem in a better way?
  • How would you like your data results to be displayed?
  • What do you like and don’t like about handling advertising data?
  • How long usually a customer campaign in run?
  • What social media platforms do you generally advertise in?
  • Would you like to search for the data or filter the data to get results?
  • How do you find results in your existing system?
  • How do you run social media campaigns?
The answers were initially noted down on a paper and later grouped using sticky notes.


Persona 1
Persona 2

User needs and insights

How might we questions

  • How might we use a scalable system to display all of the audience data in one place?
  • How might we manage related terms such that the right people see them?
  • How might we make the process of creating target audiences go more smoothly?
  • How might we estimate the size of the target audience at the time of its creation?
  • How might we could build all of our social media efforts in one place?
  • How might we quickly target a user or a group of users?
  • How might we easily target an audience based on their location?
  • How might we switch brands and campaigns quickly?
  • How might we reuse the target audience data we’ve already created to save time and effort?

Information Architecture & Sitemap

An open card sorting exercise was conducted in parallel with the users that helped to derive the information architecture and sitemap. Initial options were created on paper, and the final versions were digitised.

Task flows

In-scope task flows were identified and created.


Low fidelity wireframes were created on paper. and finalized options were digitized.

Low fidelity wireframes on paper
Users – Landing UI – Low fidelity wireframe
Wireframes – UI Navigation structure
Wireframe – Upload Data
Wireframe – Ad Manager – Create New Ad – Target Audience
Wireframe – Ad Manager – Create New Ad – Target Audience – Intelligent keywords
Wireframe – Ad Manager – Campaign Details

Visual Design

Logo Design

The original logo explorations were done on paper. The goal was to explore around keywords up, sell, hot & cold users. A handful of them were finished and transferred to digital, and one was chosen that was also acceptable to the client.

Logo designs – Explorations on paper
A few explorations that were digitized
Final Logo

The logo depicts how a cold user becomes a hot user while also incorporating characteristics of ‘up’

Logo colors
Vertical lockup
Horizontal lockup

User Interface Design – Key Screens

(Limited data displayed due to NDA)

Login UI
Upload UI
New Campaign: Create Target Audience
Target audience – All users UI
Target audience – Search
Target audience – Hot users
Target audience – Cold users
User 360 – Profile
User-360 Actionable insights


A click through prototype of the design was created and tested while wireframing as well as visual design stage.

Currently Upzel is designed as a minimum viable product featuring Users and initial version of Ad manager sections. The designs and flows were presented internally in various stages for continuous feedback and evaluation which helped in the improvement of the product. The engineering team was very talented enough to implement the design without major changes. The development is in progress and the plan is to roll out the first build soon. The future scope includes many more modules coming into the dashboard and making it a full fledge solution for medium scale businesses. Overall it was a fulfilling experience working on this project.

Previous Draco
Next Canoeing away