AI certification study plan template, driven by weights
· 7 min read
A good study plan for an AI certification is built from two numbers per domain: how much the exam guide says it weighs, and how well you score on it today. Split your hours by weight, shift some of them towards your weakest domains after a baseline test, run the same weekly cycle of learn, practise and review, and reserve the final week for timed mocks. The method below works for any exam that publishes domain weights, with a worked example for the AWS Certified AI Practitioner (AIF-C01).
Step 1: collect the facts from the official guide
Before planning anything, get the current exam guide from the provider's site and write down:
- The guide version. Your material must follow the same version.
- Each domain and its weight. Weights tell you how the scored questions are distributed.
- The task statements inside each domain. These are your checklist; each one needs at least one study session and some practice.
- The format. Question count, time limit, and question types such as multiple response, ordering or case studies.
Some providers publish weights as ranges. Microsoft's AI-901 study guide, for example, gives 40–45% and 55–60% for its two domains. Use the midpoint for planning.
Step 2: decide your total hours honestly
Count the hours you can actually protect each week, not the hours you hope to find, and multiply by the number of weeks until your exam date. Then set aside roughly one sixth of the total, or at least one full week, for final mocks and review. The rest is your domain budget.
There is no correct total. It depends on your experience with the platform, how much of the guide is new to you, and the level of the exam. The baseline in step 3 tells you more than any rule of thumb.
Step 3: take a baseline
Answer a short mixed set of questions for each domain, around 20 to 25 per domain, without studying first. Record accuracy per domain. The aim is not a good score; it is a map of where you stand.
A full-length mock works as a baseline too, but domain-level sets give cleaner per-domain numbers because each domain gets the same number of questions.
Step 4: turn weights and baseline into hours
Start with a weight-only split, then blend in your gaps.
- Weight share for each domain is simply its exam weight.
- Gap score for each domain is its weight multiplied by (1 − your baseline accuracy). A heavy domain you are weak in scores highest.
- Gap share is each domain's gap score divided by the total of all gap scores.
- Final share is the average of the weight share and the gap share.
- Hours per domain are the final share multiplied by your domain budget, rounded to half hours so that the total still matches the budget.
Blending the two keeps you honest in both directions. A weight-only plan ignores what you already know. A gap-only plan can starve a heavy domain you happen to be good at, and heavy domains are where most of the marks are.
Step 5: run a weekly cycle
Every study week follows the same four-part loop.
- Learn. Work through one or two task statements from the guide using official documentation or training. Keep sessions short.
- Practise. Answer questions on what you just learned, ideally in 25-question sets. Start with topic-level sets and move to mixed sets once a domain feels familiar.
- Review. Read every explanation, including those for questions you got right, and note each missed concept in an error log with the domain it belongs to.
- Revisit. At the start of each week, re-attempt missed questions from the week before. Spacing the retry is what makes it stick.
At the end of each week, update your per-domain accuracy. If one domain is falling behind, move half an hour or an hour into it from the domain you are strongest in. Our article on [how to use practice exams effectively](/blog/how-to-use-practice-exams-effectively) explains the research behind this loop.
Step 6: the final week
The last week is for rehearsal, not new content.
- Two full-length mocks under real conditions: the exam's question count and time limit, one sitting, no notes.
- A full review after each mock, feeding the error log.
- Targeted repair on the two or three concepts that keep recurring.
- A light final day. Skim your error log and rest.
If both mocks show a domain well below the others, it is reasonable to move the exam date rather than hope. Providers' rescheduling rules vary, so check yours early.
Worked example: AIF-C01
The AWS Certified AI Practitioner exam has 65 questions in 90 minutes, according to the AWS certification page. The exam guide (version 1.4) states that 15 of those questions are unscored and that the passing score is 700. It lists five domains:
| Domain | Weight |
|---|---|
| 1. Fundamentals of AI and ML | 20% |
| 2. Fundamentals of Generative AI | 24% |
| 3. Applications of Foundation Models | 28% |
| 4. Guidelines for Responsible AI | 14% |
| 5. Security, Compliance, and Governance for AI Solutions | 14% |
Details as of September 2026; confirm on the official AIF-C01 page before booking.
Hours. Take a candidate with six hours a week for six weeks: 36 hours. The final week's six hours go to mocks, leaving a domain budget of 30 hours.
Baseline. Suppose this candidate, who has used generative AI tools but not built on AWS, scores the following on a baseline (illustrative figures): Domain 1 80%, Domain 2 60%, Domain 3 55%, Domain 4 70%, Domain 5 50%.
Calculation.
| Domain | Weight share | Gap score | Gap share | Final share | Hours |
|---|---|---|---|---|---|
| 1 | 20% | 0.20 × 0.20 = 0.040 | 10.7% | 15.3% | 4.5 |
| 2 | 24% | 0.24 × 0.40 = 0.096 | 25.7% | 24.8% | 7.5 |
| 3 | 28% | 0.28 × 0.45 = 0.126 | 33.7% | 30.8% | 9 |
| 4 | 14% | 0.14 × 0.30 = 0.042 | 11.2% | 12.6% | 4 |
| 5 | 14% | 0.14 × 0.50 = 0.070 | 18.7% | 16.4% | 5 |
| Total | 100% | 0.374 | 100% | 100% | 30 |
Domain 3 gets the most time because it is both the heaviest and one of the weakest. Domain 1 drops below its weight share because the candidate already knows it well, but it still gets four and a half hours so it does not decay.
Weekly schedule.
| Week | Focus | Hours |
|---|---|---|
| 1 | Domain 1 (4.5 h) and start Domain 2 (1.5 h) | 6 |
| 2 | Domain 2: generative AI concepts, capabilities and limits, AWS infrastructure for generative AI | 6 |
| 3 | Domain 3: design considerations and prompt engineering techniques | 6 |
| 4 | Domain 3: training, fine-tuning and evaluating foundation models (3 h); Domain 4: responsible AI (3 h) | 6 |
| 5 | Domain 4: transparent and explainable models (1 h); Domain 5: securing AI systems, governance and compliance (5 h) | 6 |
| 6 | Two full-length mocks (65 questions, 90 minutes each), full reviews, targeted repair | 6 |
Weeks 1 to 5 add up to the 30-hour domain budget: 4.5 + 7.5 + 9 + 4 + 5. Each week also starts with a short revisit of the previous week's missed questions, taken from within that week's hours.
Our [AIF-C01 study guide](/blog/aif-c01-study-guide) covers what each of these domains tests in more detail.
Adapting the template to other exams
The same steps apply to any exam with published weights. For exams with many small domains, such as the six-section [PMLE](/blog/pmle-study-guide), group adjacent domains into weeks. For exams with a heavily dominant domain, such as the 31% first domain of [AIP-C01](/blog/aip-c01-study-guide), expect that domain to span two or more weeks. For exams that include hands-on or console-based components, add practical lab time to the "learn" step, because question practice alone cannot rehearse it.
Browse the [certifications page](/certifications) for each exam's domains and weights, or see [how to choose an AI certification](/blog/how-to-choose-an-ai-certification) if you have not picked one yet.
Frequently asked questions
How many hours should I plan for an AI certification?
It depends on your experience with the platform and the exam level, so no single figure is reliable. Plan from the hours you can protect each week, take a baseline, and adjust after the first two weeks based on how quickly your domain accuracy moves.
What if my baseline is already high in every domain?
Keep the weight-only split, shorten the plan, and bring the final-week mocks forward. If your mocks confirm the baseline, you may be ready sooner than planned.
What if the exam guide gives weight ranges instead of fixed percentages?
Use the midpoint of each range. Microsoft, for example, publishes ranges such as 40–45%; planning with 42.5% is close enough for scheduling.
Should I study domains in the order the guide lists them?
Usually, yes. Guides tend to put foundations first, and later domains often build on them. Move a domain earlier only if your work needs it sooner or your baseline shows a serious gap.
What should I do if a mock score drops in the final week?
Read the review before reacting. A drop concentrated in one domain calls for targeted repair; a drop spread across every domain often reflects pacing or fatigue, so rehearse timing rather than cramming content.
CertifyAIx is independent and not affiliated with any certification provider. All CertifyAIx practice questions are original.