PTE Academic is almost entirely scored by AI. That is both its biggest challenge and its biggest opportunity. Unlike IELTS, where a human examiner's mood can affect your result, PTE's AI applies the same criteria to every single response. Once you understand what the AI is measuring, you can train specifically for it.

This guide breaks down every task type in every section — what gets scored, what gets ignored, and the specific tactics that move your mark up.

Speaking & Writing

Read Aloud

What is scored: Pronunciation (40%), Oral Fluency (35%), Completeness (25%).

  • Completeness carries 25% of your score. If you stop halfway through because you stumbled on a word, you lose those marks permanently. Keep reading — always finish the passage.
  • Oral Fluency measures your pace and smoothness, not perfection. A natural, slightly imperfect delivery scores higher than a slow, over-careful one full of unnatural pauses.
  • Use all 30–40 seconds of prep time. Scan for difficult words and plan how you'll handle them before the microphone opens.
  • Do not re-read a sentence you mispronounced. The AI is not tracking errors backwards — it moves on as you do.

Common mistake: Stopping at a hard word, going silent for several seconds, then re-starting. The pause kills your Fluency score more than the mispronunciation would.

Repeat Sentence

What is scored: Pronunciation (40%), Oral Fluency (35%), Completeness (35%).

  • You hear the sentence once. Your goal is to repeat as many words as possible, in order.
  • If you can only remember 70% of the sentence, say that 70% fluently — you will score far better than someone who hesitates and reconstructs a clunky approximation of the full sentence.
  • Focus hardest on the beginning. Primacy helps — starting confidently tends to lock in the rest.
  • There is no prep time. Start speaking immediately when the microphone opens.

Common mistake: Trying too hard to remember every word, going silent for 3+ seconds, then stumbling through. Silence costs Fluency marks; confident partial repetition does not.

Describe Image

What is scored: Content (33%), Pronunciation (33%), Oral Fluency (34%). A human expert also reviews your Content score before it is finalised.

  • The human reviewer penalises memorised template responses. Phrases like "As we can see from the image, this graph shows..." followed by a formulaic structure are specifically flagged as lower quality.
  • Aim to cover: the main trend or relationship, the highest and lowest values (for charts), what each element represents, and a brief conclusion or implication.
  • Use 30–35 seconds of your 40-second window. Finishing in 12 seconds signals shallow content coverage.

For bar/line/pie charts: Name the axes, state the highest and lowest, describe the trend, and make a concluding observation.

For process diagrams: Describe the flow — start, middle, end — and note what changes at each step.

Re-tell Lecture

What is scored: Content (33%), Pronunciation (33%), Oral Fluency (34%). Your Content score also contributes to your Listening pillar.

  • Take notes during the audio. You have 10 seconds after the audio ends before recording begins.
  • Do not let silence exceed 3 seconds. PTE's system automatically ends the recording if you pause for more than 3 seconds.
  • Cover: the main topic, 2–3 key supporting points, and a closing sentence.

Answer Short Question

What is scored: Content only — binary. Correct answer = full marks, wrong answer = zero.

  • Give a short, direct answer. Usually 1–3 words. Extra words don't help.
  • If you did not hear the question, give your best guess. Silence scores zero; a wrong answer also scores zero.

Writing

Summarize Written Text

What is scored: Content (max 2), Form (max 1), Grammar (max 2), Vocabulary (max 2), Spelling (max 1) = 8 points total.

Critical rule — the Zero-Rule: If your Content score is 0, your entire task is scored 0.

  • Your response must be exactly one sentence, between 5 and 75 words. Violating this is an automatic Form score of 0.
  • Do not copy phrases directly from the passage — the AI penalises unparaphrased text in the Vocabulary criterion.

Template that works: [Main idea] which/because/although [supporting detail], suggesting/demonstrating that [conclusion].

Write Essay

What is scored: Content (max 3), Written Discourse (max 2), Form (max 2), Grammar (max 2), Vocabulary (max 2), Spelling (max 1) = 12 points total.

  • Word count is strict: 200–300 words. Below 200 or above 300 loses Form marks. Count as you write.
  • Vocabulary is partially measured by Type-Token Ratio. Repeating the same words throughout the essay actively lowers your score.
  • Spelling is only 1 point but it is free marks — proofread the last 2 minutes of your 20-minute window.

Reading

Re-order Paragraphs

What is scored: 1 point per correct adjacent pair.

  • Find the introduction paragraph first — it introduces the topic without reference to anything prior, and will not start with a pronoun or conjunction.
  • Look for reference chains: a pronoun in one paragraph ("this approach", "they", "however") links it to a specific previous paragraph.

MCQ Multiple Answer (Reading)

What is scored: +1 for each correct option, −1 for each incorrect option. Minimum 0.

Negative marking is in effect. Only select an option if you are confident it is correct.

Listening

Write from Dictation

What is scored: 1 point per correctly spelled word in the correct position. No negative marking.

  • Spelling matters. Every word you spell correctly in the right position earns a point.
  • Getting the beginning of the sentence right is critical — the AI checks words positionally.
  • These sentences are typically 8–14 words and played once.

Highlight Incorrect Words

What is scored: +1 for each correctly highlighted word, −1 for each incorrectly highlighted word.

  • Only highlight a word if you heard clearly that the speaker said something different from what is written.
  • This is one of the highest-risk tasks for negative marking. Be conservative.

General Strategies

Time traps to avoid:

  • Spending more than 2 minutes on any single Reading MCQ
  • Finishing Describe Image in under 20 seconds
  • Writing fewer than 200 words in the essay
  • Going silent for 3+ seconds in any speaking task

On templates: PTE introduced human expert review of the Content dimension specifically for Describe Image and Re-tell Lecture because formulaic template responses had become widespread. Practise generating real descriptions from real images instead.

ScoreFluent provides AI-powered PTE practice with criterion-level feedback — the same dimensions the real exam scores, explained in plain language after every attempt.

Share𝕏 / TwitterLinkedIn