Our Journey To Launch A Resume Checker Just For Product Manager Candidates
Have you ever wondered whether your resume is good enough?
You know you’ve had big accomplishments in your role. But what if you want to change your role to a Product Manager? You had accomplishments as an engineer, sales, marketer, or analyst. Do your accomplishments on paper highlight the Product Management skills you’ve gained?
After seeing many aspirants face this dilemma, our team at Spark Creative Technologies built an algorithmic resume-checking software. I worked with Suraj Shivshankar, data science intern, to build the product. This article is written jointly by Suraj and me.
Thumbnail includes Image by pch.vector on Freepik.
Your Life - Summed Up In A Page
Lebenslauf. That is the German word for a resume (also known as CV or Curriculum Vitae). It directly translates to “a run-through of your life”. This meaning is accurate because you need to summarize your life in a page or 2 for your resume. Using the wrong words could give an incomplete explanation of your skillset.
With companies receiving thousands of resumes for just a single job posting, applications want to stand out. These companies usually use automated checkers to filter out the applicants and that can only happen if they use certain keywords in their resumes. When a hiring manager or recruiter sees a resume, they are able to shortlist or reject candidates within a few seconds. Your resume is crucial in your job hunt process because of the low conversion rate from submitting a resume to getting an interview call. See more at: Why Is Your Resume Crucial when applying to PM roles?
What Are PM Aspirants Looking For?
Aspirants want to be seen by prospective employers. They want to highlight essential parts of their work on their resume. But how to define essential? What’s essential on a resume for project management may be unnecessary for Product Management.
Aspirants want to get an interview call. Should an engineer highlight being a great engineer interested in Product Management, or should they highlight mediocre Product Management experiences?
What Do Users Try Now?
Conversations with dozens of aspirants helped me understand that they have three ways to improve their resume. Thumbnail includes results from Craiyon.
Inbuilt spelling and grammar check in Word Applications like Google Sheets and Microsoft Word.
Get feedback from acquaintances who are Product Managers or from friends.
Using one of the dozens of resume checkers on the market.
Some resume checkers have a business model that identifies issues in the user’s resume for free and charges for the editing service. Some others require payment for the diagnosis. However, they enter a rabbit hole because the services are generalized resume-checking services. They are not specific to any particular job role. On the other hand, some services let you review your resume vs. a job description. I see aspirants waste a lot of time editing their resume independently for each job they apply to.
What Is The Ideal Solution For A User?
If we had a magic wand, we would hear the user’s experience, ask their desired role, and write out a resume that shows their strengths for their desired role while being true to their experiences.
We don’t have a magic wand. Let’s take it a notch lower.
An ideal solution would be a service that analyzes the user’s resume and prescribes steps to improve it. Aspirants will use it to know their areas of improvements and how to address them. Some of these insights include
a score (a metric to show how your resume compares to a set value) - an example from Enhancv below,
what you can add to your resume to make it better, and
what you can remove from your resume to make it more succinct and less redundant.
For this to be a great experience from the user’s side, it should involve very little human interaction. The resume should get submitted after 1-2 clicks. The user should then get a personalized and customized output based on where they can improve to make their resume better. This output could be in the form of a dashboard.
We Defined A Product Goal
We need to develop a metric to rate Resumes for the Product Manager role. Suraj and I developed an algorithm in Python that returns a rating for an uploaded resume. We used data science to define the review criteria for a candidate’s resume.
We chose Python for this task. Although Java and Javascript are better for web-related services, Python is better for manipulating data and converting raw data, which is crucial to the resume checker.
We aspired to embed the Python script into an HTML webpage. This embedding will allow us to display the ranking score and results on the page immediately. We can deliver this solution at zero marginal cost. It’ll allow aspirants to edit their resume and re-run the analysis. They can interact with the strengths and areas of improvements output.
We store aggregate anonymized information of the submissions to know usage statistics and see where we need to make changes.
What is the Minimum Viable Product?
Our goal for the MVP was to build a prototype that shows the product can be built (technical feasibility) and test for customer needs (business viability). We built a landing page consisting of a PDF upload button and a Python script we manually ran to calculate the score.
We found early adopters who tested the product but also realized we had to change the code significantly to focus on areas of improvement in textual form instead of a numerical score.
Where Did We Reach Till February 2023?
Naturally, we did not achieve the product goal by February 2023. We deployed a working model but the overall service required human involvement. Owing to Wix limitations on embedding Python, we were unable to deploy the Python dependencies with the HTML script.
Our workaround was to run the Python script manually through these steps:
We created a form on the company’s webpage. Users can upload their resume and enter their Email address through the form.
Wix emails the information to us.
We run the Python algorithm on the resume in our local system.
We email the results to the user.
Our product changed significantly from February 2023 to March 2023, which I will keep for another post.
How Would Users Use The PM Resume Checker?
The service is easy for the customer to use. They must fill in their email address and upload their resume in PDF format. Hit submit. Then, they wait for 24 hours while we process and email them the output.
What Challenges Did We Face?
We faced two significant challenges.
One was owing to the limitations of Wix. While the website is accessible through its drag-and-drop features, it was tough to add our code. Wix does not accept Python code (apart from just 2-3 lines). So we need to put the Python code in an HTML wrapper. Suraj tried doing this using Py-Script. Py-Script is a valuable package developed by Anaconda very recently. By calling this package in HTML, you can continue coding in Python, removing the need for Javascript. But the issue that arose was that we would need to upload both the HTML file along with the Python dependencies, which is not possible on Wix.
The other major challenge was more apparent when we moved from a scoring output to a textual output. We thought PDF is the standard format to share documents. And yes, it is. We thought if Python can parse the words in a PDF, it can also parse the PDF into a structured output. Nopes. We searched and tried various libraries but could not find a way to convert a PDF resume into a structured list of bullet points, companies, or accomplishments.
Technical Perspective On Changes We Can Make In The Future
Because we are far from the finished product, we have much more work to do in this field. Our main challenge remains to make the process completely automated. For this, we need to start incorporating the python dependencies with the HTML file. This will be difficult as we need to find a workaround for Wix. Another opportunity is to shift the website from Wix to another hosting service, but some of those alternatives turned out to be quite expensive for an unproven market need.
Apart from hosting issues, we would also like to improve the algorithm by providing more insights to the users. We want to use tools available on pandas and Python to manipulate data.
We changed the product significantly since writing this blog, although I published this blog several weeks later. You can check out the resume checker product here. I’ll publish the next blog here in a few weeks.