It started with a bad meeting. I was presenting a project update to a group I didn't know well, and I could feel it going sideways in real time — not because the content was bad, but because I kept saying "basically" and "um" and "like, sort of" every other sentence. Afterward, a colleague mentioned it, gently. That was enough.
I decided to actually measure the problem. Every day for 30 days, I spent 5–10 minutes speaking and tracking my fillers. Here's what I found.
My first session was a 5-minute improvised talk on a topic I know well — my job, basically. I expected maybe 10–15 fillers.
There were 47. In 5 minutes. That's nearly one filler every six seconds.
The breakdown surprised me too. I thought "um" was my main problem. It was actually "basically" — I used it 19 times, sometimes twice in the same sentence. I had no idea. It had become completely invisible to me.
The rest of week 1 was similar. Some sessions were slightly better, some slightly worse, but the range was 38–52 fillers per 5-minute session. I wasn't improving — I was just measuring.
Around day 9 or 10, something shifted. I started hearing my fillers as I said them. Not before — but the half-second after, I'd register it. I'd think: "there's another one."
This sounds like progress. It wasn't, immediately. My counts actually went up — 54, 58 — because I was now monitoring myself while speaking, which added cognitive load, which triggered more fillers. I was aware of the problem and making it worse at the same time.
This is apparently normal. Several resources I read mentioned a "conscious incompetence" phase where you've become aware of the problem before you've built the skill to fix it. It's uncomfortable, but it means the process is working. I kept going.
By the third week, I had enough data to see patterns. I was saying "basically" most when I was summarizing something I wasn't totally confident about. "Um" appeared most at the start of answers to questions, not mid-sentence. "Like" clustered around comparisons and examples.
Knowing this let me practice specifically. Instead of just doing general speaking practice, I would ask myself a question and work on the opening of my answer — where most of my "um" use was concentrated. I'd do the same answer three or four times, trying to open cleanly each time.
By the end of week 3, my per-session count had dropped to the mid-20s. That was roughly half my baseline in two weeks of targeted practice.
The more meaningful change in week 4 wasn't in my practice sessions — it was in my actual speech. I caught myself pausing instead of filling during a team call. I stopped mid-sentence, stayed silent for a moment, and finished the thought. A couple of people responded positively without knowing why.
By day 30, my practice session counts were in the 12–18 range — down from 47 on day 1. "Basically" dropped from 19 uses to 2. I'd replaced most of my fillers with either silence or nothing — I'd restructured the sentences mentally so the filler wasn't needed.
The transfer to real conversation wasn't complete. High-pressure situations still triggered more fillers. But the floor had risen: even my worst days were better than my average from week 1.
A few things surprised me. First, the data is what makes the difference. Vague awareness ("I say um too much") does almost nothing. Knowing your actual count, your actual patterns, and seeing them change over time is what drives behavior change.
Second, 5–10 minutes a day was enough. I didn't need long sessions. I needed consistent sessions with clear feedback.
I used Fluent throughout this — it tracked the counts per session, broke down my fillers by type, and gave me AI coaching that was actually specific to what had happened in each session rather than generic tips. That specificity was the thing that made practice feel useful rather than performative.
Start tracking your own baseline
The hardest part is seeing the real number. After that, it gets easier.
Try Fluent free →