She opened Zoom and wiped a tear she didn’t want me to see. “This year my application will expire. I need to pass the TOEFL.”
She had studied pharmacy in French. English was her third language. She spoke better than many colleagues, yet the score report kept saying 24 or 25. Over and over.
She wasn’t failing knowledge. She was trapped by a system that measures performance in 17 minutes, inside a headset, under fluorescent lights, while your heart fights the clock. She had already tried tutors. She had tried styles that did not fit. What she wanted was a report that showed exactly where to focus.
When we looked at her data inside My Speaking Score, I saw the pattern I see in many repeat test takers. Hours and hours of AI practice. Almost no tests in Real Mode, where the test rules are enforced. That means no prep time, re-recording allowed, and transcripts visible during practice. Great for study, but not how you get a valid baseline.
I told her the truth: to reach 26 you need multiple fours across tasks, and that requires intelligible, fluent speech at test speed. You can simulate a 26 in practice if you push rate to about 150 words per minute and rehearse, but anxiety inside a center pulls performance down if the habits are fragile.
She nodded. Then she cried again. “I’m on antidepressants because of this test,” she said. We added credits, invited her to our Slack, and set a plan. Not more random practice. A measured plan.
Note: Even though I have her permission to discuss her case in public, I've changed a few details to protect the identity of this customer.
What the Data Said
SpeechRater™ does not give opinions. It gives evidence.
- Practice pattern: Heavy AI Assist usage, almost no Real Mode baselines.
- Intelligibility pressure: Strong accent interacting with listener comprehension. If the listener goes “huh,” fours are out of reach.
- Pronunciation flags: Vowel-related measures rarely in the green.
- Lexical weakness under stress: Vocabulary diversity often in the red. The quickest lever was modifiers such as descriptive adjectives and adverbs.
The problem was not theory. It was the absence of a system: no alternation between study and test, no formal baseline, and no targeted drills tied to the numbers.
The Plan We Built
1) Separate Study from Test
- Study = AI Assist Mode: Pause. Re-record. Read transcripts. Deliberately fix one variable at a time: structure, content, speed, then pronunciation features.
- Test = Real Mode: 17 minutes. One pass. Full rules. No shortcuts. Establish a weekly baseline that reflects true readiness.
2) Make Intelligibility Non-negotiable
- Target clarity first. If a listener struggles to decode your words, a four does not appear. Track vowel quality and consonant clarity inside your reports and pair with slow-to-fast shadowing.
3) Engineer Fluency at Test Speed
- Build to ~150 wpm only after clarity is stable. Use chunking and breath cues rather than pushing rate blindly. The goal is durable fluency that holds under anxiety.
4) Upgrade Lexical Variety Under Pressure
- Pre-build modifier packs. Practice “describe then defend” patterns to force adjectives and adverbs into spontaneous speech.
5) Choose Strategy Over Lottery
- Stop booking weekly tests as a hope strategy. Use a plan with targets, goals, milestones, and one focused test window.
What My Speaking Score Delivers
Below is how we convert tears into traction.
A Note to Pharmacists on Attempt 10, 20, or 30
You are not broken. You are missing a system. Use data to locate the real blockers, and then work one lever at a time. Alternate study and test. Add accountability. The work is hard, but the path is clear.
I have watched pharmacists break through after 8, 13, 40, even 57 attempts. The moment the number turns to 26 or 27, people cry in parking lots and hug family in kitchens. It becomes one of the happiest days of a career.
If you are ready to try a data-guided path, open My Speaking Score, run a Real Mode test today, and look your truth in the eye. Then use AI Assist to fix exactly what the numbers show.
Frequently Asked Questions
Q1: Why do I score higher in practice than on test day?
Because practice often happens in AI Assist where rules are relaxed. You can re-record, peek at transcripts, or start whenever you want. That builds knowledge but not game-day reliability. Establish a Real Mode baseline under full rules.
Q2: Do I need to talk faster to reach 26?
Speed helps only after clarity is secure. Many reach 26 by getting to about 150 wpm, but only if intelligibility is stable and transitions are smooth.
Q3: My accent is strong. Can I still get 26?
Yes, if listeners never struggle to decode you. Accent is fine. Unintelligibility is not. If a rater mentally says “huh,” fours disappear. Prioritize vowel clarity and consistent syllable shaping.
Q4: I keep getting 24–25. What is the fastest lever?
Lexical variety under pressure. Pre-load descriptive modifiers and practice placing one in your first sentence. This is often the quickest visible lift in SpeechRater™.
Q5: Should I book a test every week to stay in rhythm?
No. That behaves like a lottery strategy and increases stress. Build clarity, set milestones, then book a focused window.
Q6: Can My Speaking Score help if I already tried tutors?
Yes. MSS gives objective data so you can coach yourself with precision. It shows the exact features that stall your score and lets you test whether changes hold up under rules.
The Turn
At the end of our call I told her the plan, added credits, and gave her direct access for accountability. We talked about finding small ways to put some kindness back into the world, even during a hard season. She said, quietly, that she would try.
If you are a pharmacist who needs a 26, today is your turn. Open My Speaking Score. Run one Real Mode test. Read your SpeechRater™ report. Pick one lever. Fix it. Then test again.
Your 26 is waiting.