← AI Graphic Design Pro
0/15 complete

Module 05 · Running a Design Practice

Building a Portfolio That Sells Design, Not Just Software Skill

20 minfocused lesson7practical steps4grounded questions4source links
Open lesson + course map

On this lesson

Course outline

A portfolio should help a specific buyer judge whether you can solve their kind of problem. Tool logos, AI galleries, and unexplained mockups show activity; they do not show that you can understand a brief, make defensible decisions, handle constraints, or deliver usable files.

In this lesson you will turn one authorized project into a concise case study and assemble a portfolio structure around the service you want to sell. Sample work will remain clearly labelled, and private client details will stay private.

// concept

Pick a Buyer and a Decision

Write one sentence:

// prompt — copy me2 lines
I help [specific buyer] make [specific design outcome] for [specific use],
with [relevant constraint or strength].

Examples, not promises:

  • “I build practical bilingual identity systems for Pakistani service businesses that need social, invoice, and low-cost print assets.”
  • “I create evidence-safe campaign templates for small teams publishing across Instagram and WhatsApp.”
  • “I turn early product briefs into accessible launch visuals and editable handoff systems.”

The sentence determines what stays. If you sell identity systems, six unrelated AI portraits dilute the evidence. Three relevant, deeply explained projects can be more useful than twenty disconnected thumbnails.

Design the page for the buyer’s questions: Have you handled a similar problem? What did you actually do? Can I see the decisions? What will I receive? Can I trust your claims and rights process?

// concept

Write Case Studies as Evidence

Use this structure:

  1. Context: client or labelled hypothetical, audience, and channel.
  2. Problem: the approved design problem, not a dramatic invented business crisis.
  3. Role and limits: what you owned, collaborators, timing, and constraints.
  4. Process: only the decisions that explain the result.
  5. Output: flat artwork plus real or clearly labelled sample applications.
  6. Validation: checks performed, client approval, or observable implementation evidence.
  7. Reflection: one limitation and what you would test next.

AI can help compress source notes without inventing outcomes:

// prompt — copy me8 lines
Turn the verified project notes below into a 250-word portfolio case study
using Context, Problem, Role, Decisions, Deliverables, Validation, Reflection.
Use only supplied facts. Do not add metrics, client praise, business results,
research participants, dates, or team members. Label the project hypothetical
if PROJECT STATUS says hypothetical. Return an unsupported-claim checklist.

PROJECT STATUS: [real authorized / redacted client / hypothetical]
VERIFIED NOTES: [paste]

Check every sentence against a source: signed brief, version history, approval message, published artifact, or your own dated observation. If you cannot support a performance result, describe the artifact and test instead.

// concept

Show Work Without Exposing Clients

Request permission that names the assets, channels, client name, and timing you may show. A public website, a private proposal deck, and a job interview are different audiences. If permission is absent, redact or rebuild a hypothetical exercise from scratch; do not merely blur a phone number while leaving customer names in screenshots.

Label roles precisely:

// prompt — copy me5 lines
Role: visual identity and template system
Client supplied: strategy, final copy, product photographs
Collaborator: Urdu translation reviewed by client-appointed editor
AI assistance: early image ideation and copy-length variants
Human work: brief, selection, vector construction, type, QA, and delivery

For AI-assisted images, retain the tool, date, prompt/version notes where permitted, source asset permissions, edits, and client approval. Do not imitate a living artist or present an AI mockup as a manufactured item.

// concept

Make the Portfolio Easy to Scan

Lead with the service statement and three strongest relevant projects. Each card needs problem, deliverable, role, and one representative flat image. On the case-study page, use short sections, captions that add evidence, and accessible alt text. Compress images, define dimensions to prevent layout shifts, and test on mobile data.

End with one action matching the page: request a scoped project conversation, download a capability sheet, or view the next related case. Avoid fake live counters, fabricated testimonials, and client-logo walls without permission.

// worked_example

Worked Example

The revised home page says the learner builds practical identity and social systems for local service businesses. The first case study is explicitly Self-initiated hypothetical project — not client work. It explains the mobile-recognition and one-color-print constraints, shows the three visual territories, documents why two were rejected, and presents the rebuilt vector, template tests, and file guide.

There is no invented “40% engagement increase.” Validation says: “Tested at 28 pixels, in one-color print, in English/Urdu lockups, and by a second editor using the templates.” The reflection notes that customer comprehension was not tested and would require real users with consent. The case is less flashy and far more credible.

// failure_cases

Failure Cases to Diagnose

7 cases to diagnose

  • The home page leads with software

    replace tools with buyer, problem, output, and evidence.

  • Every project gets equal space

    curate relevant work and archive experiments that weaken the signal.

  • A hypothetical is written like paid work

    label it at card, title, and case-study level.

  • Results are invented or implied by mockups

    state performed checks and known limitations instead.

  • Role is unclear

    separate your work, client inputs, collaborators, and AI assistance.

  • Confidential screens remain readable

    obtain permission or replace/redact the entire sensitive context.

  • Portfolio images are too heavy

    compress, size, and test the page on a real phone and modest connection.

// pakistan_angle

Pakistan Angle

Pakistani buyers may review a portfolio from WhatsApp on an affordable phone rather than a desktop. Create a lightweight web page or PDF capability sheet with readable captions and working links. Test it on mobile data and keep the first useful case above long biography copy. Include local production evidence—bilingual typesetting, Daraz or WhatsApp formats, flex-sign tests—only when it genuinely belongs to the project.

Permission culture matters even when work has already been published. A shop owner may not want pricing, staff faces, customer messages, or an unfinished campaign shown to another client. Obtain specific written permission and keep CNICs, addresses, analytics exports, private phone numbers, and raw WhatsApp chats out of AI prompts and portfolio media.

// hands_on

Hands-On Exercise

7 steps

  1. Write your buyer-and-outcome sentence and choose one primary service.

  2. Audit every current project as keep, rewrite, private-only, or archive.

  3. Select one real authorized or clearly labelled hypothetical project.

  4. Assemble evidence for context, role, decisions, deliverables, validation, and limitation.

  5. Write a 250–500 word case study and an unsupported-claim checklist.

  6. Create one card, one full case page, and a lightweight PDF summary.

  7. Ask a target buyer or peer to state what you sell, what you did, and what evidence they saw after a 60-second scan.

// completion_rubric

Completion Rubric

6 checks — tick as you verify

0/6

// sources

Sources

4 official sources — check every claim yourself

// check_yourself

Check yourself

4 questions · answers and options are taken word-for-word from this course

0/4

  1. 1 / 4 · diagnose

    Your work shows this failure mode: “The home page leads with software.” What does the lesson tell you to do about it?