Can't stop thinking about something you said?
Published November 13, 2026 · By The 3AM Overthinker, from the research behind the 3AM Coach
On this page
First, check one thing: did it hurt someone, or leave a wrong fact that someone will act on? If it did, send one short, specific message, once. If it didn't, there's nothing to send. Other people usually notice a slip like this less than you fear, and another pass through it is unlikely to turn up anything new.
Try this now
- Write down what you actually said, as close to word for word as you can, and what happened right after. Just what happened, not what you think it meant.
- Ask one question: did it hurt someone, or leave a wrong fact out there? If not, write "nothing to send" under it.
- If it did, send one short message that says what happened and fixes it, like the examples further down. Two or three sentences is plenty. After that, only follow up if they reply.
- When it comes back, swap "why did I say that?" for "what actually happened?" or "what would I tell a friend who said this?"
- If it keeps coming back, pick a set time later, 30 minutes at most, and save it for then.
Others usually notice less than you fear
In a Cornell study, students wearing an embarrassing Barry Manilow T-shirt guessed that about half the room had noticed it. About a quarter had.
In later studies by the same team, people who imagined a social blunder, failed at something in front of others, or were described in an embarrassing way expected to be judged more harshly than they really were (Savitsky et al., 2001). They focused on their own mistake and overlooked what softened how observers saw it, like the situation they were in. These were small studies, mostly with college students, so it won't hold every time.
Why it keeps coming back
Going back over how you came across, again and again after the moment has passed, is well studied in social anxiety (Brozovich & Heimberg, 2008).
"Why did I say that?" usually keeps it going, because it doesn't have an answer you can check. Plain questions about what and how slow it down: what were the exact words, what happened right before, what did the other person actually do next, and what would someone neutral in the room have seen? If it's mostly "I'm such an idiot", write what you'd say to a friend who told you the same story, next to the facts from step 1.
Edward Watkins built a short self-help training around this switch. In a trial of 121 adults with depression, it reduced the habit of going over things more than usual care or relaxation training did. That was depression, not everyday cringe, so it's worth trying rather than a sure fix.
Decide if it needs fixing, then do it once
Plenty of these need nothing sent. A joke that landed flat, an odd phrase, talking too much in a meeting: none of that hurt anyone. An apology for one of those may just bring it back up for people who'd moved on.
Some things do need a repair. If you said something unkind, shared something that wasn't yours to share, or got a fact wrong that someone will act on, a short message helps. For example: "I was short with you in the meeting today. That wasn't about you, and I'm sorry." Or: "Quick correction from earlier: the deadline is the 14th, not the 12th." Say what happened and what you'd like them to know. A long explanation, or asking whether it's okay now, can bring the moment back for both of you.
Asking for reassurance over and over may backfire. "Was it weird? Are you sure? Did she seem off?" Across 38 studies, people who did this a lot tended to have lower mood, and it went with a bit more strain in their relationships. Talking the same problem through again and again with a friend has been linked to more anxiety and low mood as well. These are links, not proof of cause, but they're a reason to tell the story once and leave it there.
How the other person may have seen it
In studies of conversations with new people, people tended to underestimate how much the other person liked them, partly because they were busy worrying about what they'd said (Boothby et al., 2018). It's an average, so it can't tell you how any one conversation went.
If it comes back at night
The same thing can hit harder in bed. For the bedtime version, with what to do if you're lying awake, see How to stop replaying a conversation at night.
If it keeps happening
If this happens most days for weeks, or the dread before and after social situations is getting in the way of work, friendships or going out, that's worth taking to a doctor or therapist. Coaching apps, this one included, aren't therapy. If you're thinking about hurting yourself or don't feel safe, call or text 988 (the Suicide and Crisis Lifeline in the US) or your local emergency number.
How the 3AM Coach uses this
The 3AM Coach asks what was actually said in a line or two, steers "why" toward what and how, and helps you separate what you know from what you're assuming. If you're being hard on yourself, it asks what you'd tell a friend. It won't keep answering "was it okay?", and it helps you land on either a short repair message, which it can help you word, or "nothing to do" with a time to come back to it. Try the coach
The numbers, for the curious
- Spotlight effect (Gilovich et al., 2000): wearers of a Barry Manilow T-shirt estimated about 46% of observers noticed it; about 23% did (15 wearers, 64 observers). With a T-shirt of their own choosing, about 48% estimated against about 8% actual. Small samples from one lab.
- Harsh judgment (Savitsky et al., 2001): across four studies, people who imagined a social blunder, failed in front of others, or were described in an embarrassing way expected harsher judgments than observers actually gave.
- Liking gap (Boothby et al., 2018): in a lab study of 34 strangers after a 5-minute chat, actual liking averaged 5.82 and perceived liking 5.17 on a 7-point scale. In shyer people the gap was 6.16 against 5.02. In 102 first-year students, the gap held at four of five surveys from September to February and was gone by May.
- Concreteness training (Watkins et al., 2012): d = 0.76 on clinician-rated depression against usual care after treatment (intention to treat), with less rumination than both usual care and relaxation. The depression benefit had faded by 3 months on the clinician measure. As a rough guide, 0.2 is small, 0.5 medium and 0.8 large.
- Self-compassion programs (Ferrari et al., 2019): g = 1.37 for rumination and 0.56 for self-criticism, across 27 trials. The authors flag possible publication bias, so the rumination figure likely overstates the effect.
- Reassurance seeking (Starr & Davila, 2008): r = .32 with depression (38 studies, 6,973 people) and r = .14 with rejection (16 studies). Correlations, not cause.
- Co-rumination (Spendelow et al., 2017): correlations of .14 to .26 with anxiety and low mood, across 38 studies and 12,829 people. Small to moderate, and not cause.
Sources
- Gilovich T, Medvec VH & Savitsky K (2000). The spotlight effect in social judgment: an egocentric bias in estimates of the salience of one's own actions and appearance. Journal of Personality and Social Psychology, 78(2), 211–222. https://doi.org/10.1037/0022-3514.78.2.211 (PubMed: https://pubmed.ncbi.nlm.nih.gov/10707330/)
- Savitsky K, Epley N & Gilovich T (2001). Do others judge us as harshly as we think? Overestimating the impact of our failures, shortcomings, and mishaps. Journal of Personality and Social Psychology, 81(1), 44–56. https://doi.org/10.1037/0022-3514.81.1.44
- Boothby EJ, Cooney G, Sandstrom GM & Clark MS (2018). The liking gap in conversations: do people like us more than we think? Psychological Science, 29(11), 1742–1756. https://doi.org/10.1177/0956797618783714 (accepted manuscript: https://repository.essex.ac.uk/23025/)
- Brozovich F & Heimberg RG (2008). An analysis of post-event processing in social anxiety disorder. Clinical Psychology Review, 28(6), 891–903. https://doi.org/10.1016/j.cpr.2008.01.002
- Watkins ER, Taylor RS, Byng R, Baeyens C, Read R, Pearson K & Watson L (2012). Guided self-help concreteness training as an intervention for major depression in primary care: a Phase II randomized controlled trial. Psychological Medicine, 42(7), 1359–1371. https://doi.org/10.1017/S0033291711002480 (full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC3359637)
- Ferrari M, Hunt C, Harrysunker A, Abbott MJ, Beath AP & Einstein DA (2019). Self-compassion interventions and psychosocial outcomes: a meta-analysis of RCTs. Mindfulness, 10, 1455–1473. https://doi.org/10.1007/s12671-019-01134-6
- Starr LR & Davila J (2008). Excessive reassurance seeking, depression, and interpersonal rejection: a meta-analytic review. Journal of Abnormal Psychology, 117(4), 762–775. https://doi.org/10.1037/a0013866
- Spendelow JS, Simonds LM & Avery RE (2017). The relationship between co-rumination and internalizing problems: a systematic review and meta-analysis. Clinical Psychology & Psychotherapy, 24(2), 512–527. https://doi.org/10.1002/cpp.2023
More guides
- How to stop replaying a conversation at night
- Writing a To-Do List Before Bed: Does It Help You Sleep?
- Brown noise vs white noise vs pink noise for sleep
- Should I get out of bed if I can't sleep?
- Breathing exercises for sleep
- Why do I wake up at 3am and overthink?
- Worry time: how to schedule your worries
- Does music help you sleep?