HadarBot
An experimental chatbot that reflects well...me…
  in order to help managers, potential clients
and recruiters to get to know me
better before making a decision.


Launch HadarBot



Why am I doing this?

Recruiters, managers, and potential clients see and browse a lot of portfolios on a daily basis.
Sometimes the process can get extremely monotonous & repetitive, and they may be quick to judge.
When a person enters my portfolio, I would like them to get a better sense of who I am & to better evaluate my abilities.

Who is it for (Roles)?

Design managers / Recruiters, potential clients

KPIs

- Get interviews for at least 3 interesting workplaces in the next month

- Increase portfolio views ‘conversion’ x3 in the next year




Research


First, I needed to frame a problem that is specific for me: My portfolio does not necessarily present
full case studies for certain projects, and yes, it may have been easier just to fix that - but when you can’t
expose a big chunk of your work (NDAs and such) - it’s an issue:

I need to find a way to introduce myself both professionally
and personally that would be clear and express my abilities.




In this experimental project, I’m going to try and explore a way to open myself
for roles and projects I’m interested in - currently those that revolve around
conversation design and AI powered products.


User Interviews - Understanding pain points

I started by interviewing some design managers that recruit designers for their teams,
asking about their processes and their approach; what do they ask during an interview,
what interests them, what’s repetitive and boring etc. Above all,
I tried to understand what hurts them the most in the hiring process,
here are few quotes that represent their main pain points that I will be focusing on creating this experiment:

Design Managers
Participants: 7
Age: 30-57
All recruited at least once in that past year.


    • “....(the recruiting process)... takes a lot of mental resources, time and attention...”

    • “...It’s hard, it’s f*ing hard to find someone who fits in,
      and there are so many disappointments along the way…”

    • “...sometimes it feels like ‘oh this is it!’ and then..no.
      And you have to start all over again…”

    • “...saying ‘No’ is never pleasant, but it’s for sure easier before you invest more time…”

    • “...Did they communicate their understanding of the problem they were solving?
      Very few people do this and it sucks. It sends the message that they were either
      lazy or that they value the wrong things about design…”

    • “...Different products need different skills. |
      It’s easy to go with the wrong set of skills,
      and easy to be swayed by the wrong things…”

It seems it's all about gaps between communication and understanding:
The candidate wants to communicate clearly who they are both personally and professionally
and design recruiters want to have an easy way to understand quickly and effortlessly who the candidate is.


Competitive analysis

What are the solutions offered for recruiters to evaluate candidates today?
How do candidates present themselves clearly nowadays?

Seems like most recruitment tools today focus on evaluating candidates before they get to the interview.
In addition, there are some AI tools that analyze body language and facial expressions during interviews.
These tools are based on tests that the recruiter can build and then compare results between different
candidates in order to decide who to move forward with.

There are also some checklists for the interviewer that can be used during an interview.
However, I did not encounter any automated solutions that can review a design portfolio - 
There are services that will do that for you, but it seems like they are mostly used
by the candidates themselves to evaluate themselves.

Harver (another hiring solution company) made a nice funnel diagram that shows the hiring software landscape:



As for candidates, the main tool is the portfolio itself.
It can be a more gamified-quirky one or a very tidy-structured one.
Most product design portfolios that are perceived as ‘good’ by recruiters are
those that present a case study that includes different stages of a design process
(e.g: empathizing, defining, ideating, prototyping and testing).

As for tools, as mentioned before there are people and services
that can provide the candidate feedback, but other than that it’s
based on experimenting and improving on the go (MVP style).
Personality-wise, it’s mostly the ‘About’ section where the candidates
write about themselves and the intention behind the design of the portfolio itself.

Journey Mapping

Understanding that the recruitment process differs between companies,
I created a ‘simplified’ recruitment process journey map to better understand the needs of a design recruiter.
The most meaningful step for this experiment is the portfolio examination.



Job stories (JTBD & User stories hybrid)


This is an example of a JTBD I mapped.
I decided to transfer it into a few Job Stories so I will have more actionable items.
Here are the top 3:

As a design recruiter...

When I go over a portfolio, I want to understand who the candidate is so that I can determine if they're the right fit for my team.

When I decide if I should interview the candidate, I want to be able to evaluate their abilities faster in order to save time and resources.

When I’m reading the ‘about’ section / CV, I want to get to know the person faster and to understand if they are the right candidate for me / my team.


Solution


Conversation design - Chatbot

“Human language is the new UI layer”

Stay Nadella, Microsoft CEO



Since I’m aiming to get to a role in the field of conversation design (ahem… but also open to other interesting suggestions…),
and due to my experience in this field, the solution I chose is to create a chatbot that will present me, my character, and my work.

As a MVP, it’s going to include a limited amount of questions / answers and one case study.
It will grow and change after some testing, data collection and insights that will come along with time.

The bot is not going to use any AI modules at this point.
It’s going to be based on a ‘decision tree’ to keep it as simplified and fluent as possible.
It will allow free text and will know how to answer a few predefined questions based on keywords.
It will be trained to learn more as it will be monitored. If it fails to answer, it will fail gracefully.

I want recruiters to have a fluent & easy conversation with the Hadarbot,
an ability to understand who I am, review projects easily,
be able to go in depth if they want to, ask me some questions, and easily contact me.


Bot Persona

Conversation is an emotional experience. Studies show that when people connect with devices,
It’s sincere and real (Don’t we all name our Roomba vacuum cleaner?).
Even if it's “only” a device, the emotion is still very authentic.
Therefore, when designing a conversation we have to be ethical and responsible for what we say and how we say it.



Since the bot should represent me, the persona is gonna be based on well, me.
I will use memoji’s as it’s a simple way to visually represent my digital self
and will try to keep it friendly and informal yet professional. 

It’s also pretty safe to assume that I will screw up a bit at first.
After all, I am but one writer for this bot and with my own assumptions and biases.
I will do my best to represent myself and be as authentic as possible without being hurtful to anyone.

With time, The bot will go through iterations and hopefully I will be able to improve it more and more.

I also intend to use the conversation as a modality and not as a replacement to my portfolio.
I will consider which parts fit in a conversation form and which don’t.


Conversation Flow

As part of my planning and thinking process, and similar to what I do in other visual interface projects,
I created a flow that represents the conversation to keep myself honest and make sure I don’t have any gaps.



Script and “Wizard of OZ” testing

I wrote the script according to the flow, adding some personality and extra options.
Once I was done, I did a ‘Wizard of OZ’ exercise with a few potential users to make sure
the text makes sense and the main flows work.



As this conversation’s interaction is (mostly) driven by buttons,
it’s easier to prevent mistakes as long as the text and CTA are clear.
In addition, I came to a conclusion that showing a full case study is not as sufficient in a chat form
since it’s supported by visual elements that are significant to the storytelling
and it becomes long and exhausting for some.


Prototype

I created a full working bot prototype using Landbot




Usability Testing

Objective: Identify any obstacles and or ease of use the user encounters
when trying to navigate the bot and provide opportunities for improvement.

Goals: 
  • Observe how the user interacts with the bot interface.
  • See how the user reacts to the text, tone and conversation.
  • Test how the user navigates.
  • Note frustration, satisfaction & duration for all.




Conclusions


Findings

Overall reactions were positive to the bot persona. Navigation was clear and users enjoyed the experience.
Some problems that can be fixed includes typos and message delay to allow clear reading.
I decided not to use the memoji excessively since it created an illustion of ‘not a bot’ for some users.
In addition, there are some technical issues that are inevitable at this point due to the platform I am using.


Future Improvments & Iterations

With time, I plan to improve and enrich the bot.
I will follow the analysis on the platform and act accordingly.

All issues mentioned will be fixed in the near future.


 

Launch HadarBot

Back   |   Home