Certificates vs portfolio: where your next 40 hours should go
In this article (6 sections)
Related programme: the Data Analytics with Generative AI course (3–4 months; ₹90,000 (inclusive of GST) · EMI from ₹4,000/month).
If you already have one relevant certificate, your next 40 hours almost always do more as a portfolio project than as a second certificate. A certificate tells a recruiter you passed a vendor's test; a portfolio lets an interviewer see how you handle messy data, choose a method and explain a result. The main exception is when the jobs you want name a specific certification as a filter.
What each can and cannot prove
Think about what a reviewer is trying to check at each point, and which evidence answers it.
| What the reviewer wants to know | Certificate answers it? | Portfolio answers it? |
|---|---|---|
| Have you covered the standard topics for this tool? | Yes | Partly |
| Can you clean data that arrives broken? | No | Yes, if the project shows the cleaning |
| Can you pick the right question to ask of data? | No | Yes |
| Can you explain a finding to a non-technical manager? | No | Yes, through a README or walkthrough |
| Do you know this vendor's product well enough for day one? | Yes, for that product | Partly |
| Is this person's claim verifiable quickly? | Yes, via credential ID | Yes, via a public repository or report |
The right column covers more of what an analyst actually does, which is why interviews for data roles tend to spend their time on projects. The left column still matters at the screening stage, when a recruiter is matching keywords before anyone opens your GitHub.
Allocate the 40 hours by your situation
| Your current position | Suggested split | Why |
|---|---|---|
| No certificate, no projects | 10 h fundamentals course or certificate prep, 30 h first project | You need both a keyword and something to show; the project matters more |
| One certificate, no projects | 40 h project | A second certificate adds little the first did not |
| Projects, but postings name PL-300 or a cloud exam | 25 h exam prep, 15 h improving one project | The credential is a stated filter |
| Two or more certificates, one weak project | 40 h rebuild the weak project properly | More certificates will not fix a thin portfolio |
| Strong projects, no certificate, postings don't name one | 40 h second project in a different domain | Breadth of evidence beats a badge nobody asked for |
"Postings name it" means you have actually counted. Collect ten real postings for the role you want and tally how many mention a certificate by name. If it is two or fewer, it is not a filter.
A 40-hour project, hour by hour
A portfolio project is worth the time only if it ends with something a reviewer can open and check. Here is one way to spend 40 hours on a retail-sales analysis using a public or synthetic dataset.
| Hours | Work | Output |
|---|---|---|
| 1–3 | Write the business question and what decision it informs | A one-paragraph brief |
| 4–10 | Load data into SQLite or a database; profile nulls, duplicates, ranges | A data-quality note listing every issue found |
| 11–18 | Clean, with each rule written down and row counts before/after | A cleaning log and a clean table |
| 19–26 | Analysis: 3–5 SQL queries or pandas steps that answer the brief | Queries saved as files, each with a comment on what it answers |
| 27–33 | Build one dashboard page or three charts, titled with findings not labels | "Weekend orders are 18% larger" rather than "Orders by day" |
| 34–37 | Write the README: question, data, cleaning, method, findings, limits | A document a stranger can follow |
| 38–40 | Ask someone to reproduce it from your README; fix what they got stuck on | A project that actually runs on another machine |
The last three hours are the ones people skip, and they are what separate a project an interviewer trusts from one they suspect was copied.
When the certificate should come first
Put certification ahead of a project in three situations:
- 1The role names it as required. Some service-company and partner roles list a specific Microsoft or cloud certification.
- 2You need the structure. If you have never studied the tool at all, a certification syllabus gives an ordered route through it — though a good course or official documentation can do the same.
- 3Your employer pays for it. If your company funds the exam and time, take it; then use the knowledge in a project.
Our guide to PL-300, Google Data Analytics and the AWS and Azure AI exams covers what each one tests.
Presenting both on a resume
Put them where each earns its place. Certificates go in a short section with the credential ID so they can be verified. Projects go higher, each with a one-line result:
Retail basket analysis (SQL, Power BI) — found weekend baskets were larger in two of five categories; documented 7 cleaning rules affecting 3% of rows. Repository and report linked.
A line like that tells a reviewer what you did, with which tools, and that you can be questioned on it. Keep figures honest: if the data was synthetic, say so.
A quick self-check
Ask yourself: if an interviewer said "walk me through something you built", what would you open? If the answer is a certificate PDF, the next 40 hours belong to a project.
To see how your current evidence reads to a screener, run your resume through the free skill check. The Data Analyst programme is built around graded projects of this kind, with mentor feedback in live online sessions and optional in-person labs at our Noida centre.
Pankit Kumar has 10 years in Data Science & AI, building and shipping production systems in regulated pharma and clinical environments. He is a freelance trainer at Boston Institute of Analytics, AnalytixLabs and Scaler, and has taught this material to thousands of working professionals.
This article is part of our Data Analytics with Generative AI programme — 3–4 months. The full analyst stack — Excel, SQL, Power BI and Python pipelines — then a generative-AI layer you can prove is right.
Explore Data Analytics with Generative AI