Verke Editorial

CBT exercises for self-esteem: 5 techniques you can start today

Verke Editorial · May 5, 2025

A meta-analysis of CBT interventions for low self-esteem found an effect size of d=1.12 — larger than most antidepressants. That number means CBT exercises for self-esteem don't just help a little. They shift the whole distribution. The catch? You have to actually do the exercises, not just read about them. This article gives you five. Pick one. Do it today. Not tomorrow.

Each exercise below targets a different point in the cycle that keeps self-esteem low. You don't need all five. You need one that you'll actually do, done consistently enough that your brain starts updating the evidence it pays attention to. If you want the full picture of why self-esteem gets stuck, read Building self-esteem: exercises that actually work. If you want to start working, keep scrolling.

Melanie Fennell's CBT model of low self-esteem works like this: you carry a core belief — a "bottom line" like "I'm not good enough" — that acts as an evidence filter. Information that confirms the belief slides right through. Information that contradicts it gets discounted, explained away, or simply not noticed. The belief strengthens not because it's true but because it controls what you see (Fennell & Wahl, 2023). The five exercises below interrupt this cycle at different points. That's all the theory you need. Now do something.

When you notice self-critical thinking, stop and write down seven things. This is the core CBT exercise for self-esteem — the one clinicians assign most often and the one that produces the most change when done consistently. It takes 5–10 minutes per entry.

1. The situation (what happened, just the facts)
2. The automatic thought (the sentence your mind produced)
3. How you feel right now, 0–100
4. Evidence FOR the thought
5. Evidence AGAINST the thought
6. A balanced alternative thought
7. How you feel now, 0–100

Worked example

Situation: You gave a presentation at work. Your manager didn't comment on it afterward. Automatic thought: "She thought it was terrible. Everyone could tell I didn't know what I was talking about." Feeling: Shame, 75/100. Evidence for: She usually says something after presentations. She looked at her phone once. Evidence against: Two colleagues said it was clear. She had a meeting immediately after — she may not have had time. Last month she said my project summary was strong. I prepared for six hours and knew the material. Balanced thought: "She might have been busy. The presentation was well-prepared and colleagues responded well. No evidence it was bad — just silence, which my filter read as rejection." Feeling now: 40/100.

The hard column is "evidence against." That's not a design flaw — it's the negative belief doing exactly what Fennell's model predicts: filtering out contradictory data. If you're stuck, try: "What would I say to a friend who told me this?" You can almost always find counter-evidence for someone else. The exercise teaches you to do it for yourself.

Each evening, write down 1–3 things that went reasonably well and what you did that contributed. Not "everything was amazing" — just accurate evidence. Three minutes, every day. The key word is attribution: not luck, not circumstance, but your specific action.

This is not gratitude journaling. A gratitude journal says "I'm grateful for the sunny weather." A Positive Data Log says "The meeting went smoothly and I contributed to that by preparing the agenda in advance." The self-efficacy is in the attribution. You're not cataloguing good fortune — you're building a factual record that your negative filter has been editing out.

This exercise shows results fastest of the five. After two weeks, most people notice they're catching positive evidence in real time, not just at night. The attentional filter starts shifting. It also softens the ground for the Thought Record: once you've been logging positive data for two weeks, the "evidence against" column stops feeling impossible.

Identify one thing you've been avoiding because you predict it will go badly. Write the specific prediction before you act — not a vague "it won't go well" but the exact outcome you expect. Rate your confidence in that prediction, 0–100. Then do the thing. Record what actually happened. Compare.

Example: "If I ask a question in the team meeting, people will think I should already know the answer. Confidence: 80%." You ask the question. Two people nod. One says "good question." Nobody rolls their eyes. Actual outcome: nothing bad happened. Prediction accuracy: about 15%. Bennett-Levy and colleagues found that this prediction-testing framework is one of the most powerful tools in CBT because it generates evidence you can't argue with — you lived it (Bennett-Levy et al., 2004).

Start small. Low-stakes situations first — asking a stranger for directions, giving an opinion in a casual conversation. Build toward the situations that carry real weight. One experiment per week is enough. For behavioral experiments specifically around saying no, see how to stop people-pleasing.

You live by rules you've never written down. "If I make a mistake, people will lose respect for me." "If I ask for help, I'm weak." "If I'm not the best, I'm the worst." These conditional assumptions are the load-bearing infrastructure of low self-esteem. They sit between your core belief and your daily behavior, shaping what you attempt and what you avoid.

Write down three rules you live by. For each one, answer: Where did this rule come from? What does it cost me to follow it? What would happen if I bent it slightly — not broke it, just bent it? Then design one small experiment to test one rule this week. If your rule is "I must never make mistakes at work," the experiment might be deliberately submitting a draft you know is imperfect and observing what actually happens. Takes about 15 minutes to write. The experiment takes a week to run.

If perfectionism is a major pattern for you, the rules audit is where to start. See also perfectionism: when good enough never feels enough for a deeper look at how these rules entrench themselves.

Your core belief is probably all-or-nothing: "I'm worthless" or "I'm worthy." All-or-nothing beliefs resist change because no single piece of evidence can disprove a global judgment. But "I'm at 35 out of 100 in this context" can be moved to 45. The continuum makes change possible by converting a binary into a spectrum.

Draw a line from 0 to 100. Place yourself on it. Now place five other people you know — a colleague, a friend, someone you admire, someone who struggles, someone in between. Notice: nobody is at 0 or 100. It's a spectrum, and you're somewhere on it. Now rewrite your core belief as a continuum statement: "My sense of competence varies by context and is currently around ___." This takes about 10 minutes. The shift from binary to spectrum is one of the most reliable moves in cognitive restructuring (Beck, 1976).

The two-week dip is real. Initial enthusiasm fades before the habit forms. Most people who drop CBT exercises quit in week 2 or 3 — right before the attentional shift starts to kick in. Treat these exercises like a physiotherapy prescription, not inspiration. You wouldn't stop doing knee exercises because you "didn't feel like it" three days in.

A specific schedule helps: Thought Record daily for the first two weeks, then as needed when you catch self-critical spirals. Positive Data Log daily, ongoing — this is the one that compounds. Behavioral experiments: one per week. Rules Audit: once, then revisit monthly. Core Belief Continuum: once, then re-rate every two weeks to track movement.

If you're new to working with a coach on this, see your first week with AI coaching for what the rhythm looks like in practice.

These exercises work on paper. They work better with someone who can spot the evidence your filter hides, push back when your "balanced thought" is still 90% self-critical, and keep you honest about running your behavioral experiments. Judith is a CBT-trained AI coach built to do exactly that. She walks you through thought records in real time, remembers your rules and experiments across sessions, and won't let you skip the hard column. For more on the approach, see Cognitive Behavioral Therapy.

Try a CBT exercise with Judith — no account needed