Other people's impressions converted. Mine never got a fair test.
Other people's impressions converted. Mine never got a fair test.
I grew from 102 to 270 followers by replying into conversations that already existed. My own posts did not get better. They stayed in the same low-impression bucket they were in at 102.
That is the part I got backwards for weeks. I treated follower count as the thing that would finally let me test the product. It did not. A larger following made the next original post feel more legitimate. It did not give that post a room.
A reply sits under a question someone already asked. The audience is borrowed. The context is free. A cold post has to build both from zero, on an account the algorithm has not decided to show. Same handle. Same week. Completely different denominator.
What I actually do now:
- Find threads where someone named the problem, not a launch announcement.
- Answer the specific line. One number they can check. No “great post.”
- Hit send myself. Discovery can be queued. Unsolicited auto-replies are how you get banned.
- Count conversations and profile clicks, not impressions on my own feed.
The 102-to-270 jump was a side effect of step 2, not a content-calendar win. I still cannot use my own posts to tell whether the product is weak or just unseen. If I need a verdict, I go where the verdict already has an audience.
I built a tool that does the discovery half of that reply work. Free is 10 scheduled posts a month. https://sent2x.com/?utm_source=indiehackers&utm_campaign=alexecho1#pricing
If you are also stuck reading original-post analytics as a product test: whose thread last produced a conversation for you, and was it yours?
Yes, and it changed what I look at next.
Follower growth turned out to be a vanity denominator once the 27 were already inside the product. The question stopped being how do I get more of them and became what do they do on day two.
The thing I still cannot separate from a dashboard is activation without habit versus simply the wrong people. Both look identical from outside: signed up, did the thing once, never came back. One is a product problem and one is a targeting problem, and the fix for each makes the other worse.
The only test I have thought of is whether the ones who came from a place where somebody vouched for us behave differently from the ones who came from a cold click. If they do, it is targeting. If they churn the same way, it is the product. Have you ever managed to split it cleanly?
That 2.6K against 23 to 120 is the whole illusion in one screenshot, and I think it explains why this advice keeps propagating.
X Analytics lists a reply's impressions using the thread's number, not the number of people who read your sentence. So the tool itself tells every small account that replying works twenty times better than posting. It is technically true about the room and false about you. Anyone reading their own dashboard will reach exactly the conclusion you reached, and so did I.
The only place the real figure shows up is the per-post view count on the reply itself, read off your own profile timeline rather than the analytics tab. That is where my 2, 3 and 11 came from. If you pull those four, I would expect them to be two orders of magnitude under the 2.6K.
On the next column, the most useful one for us has not been "did they name a problem" or even "did they ask". It is whether the ask has a deadline inside it. "What are you building" stays open forever. "Drop it, I am trying things tonight" has an hour attached, and those are the threads where a reply gets read instead of scrolled past.
340 followers against 24 to 27 infrequent users with no traceable paid conversion is a more honest number than most people are willing to publish, and it deserves saying out loud.
The question I would move onto your board rather than mine: if the follower count cannot be joined to the product at all, is it a metric, or is it just a receipt proving you did the activity?
My tool is built to serve X accounts. It can help you find posts you might be interested in, generate draft replies for you, and via its Signal feature, locate recent posts from potential competing X users for your product. You can click through and connect with them.
Followers are just people within a circle or those curious about you. I’m not purely running a product‑driven account; growing followers is only a side effect. So not all of these new followers are here because they’re interested in my product. Hope I’ve answered your question properly.
You are right about the denominator. I think you are wrong about what actually gets borrowed.
I read ours this morning at three days out. Four replies, each sitting directly under the original post, none folded or hidden. The threads they sat in carried 196, 291, 303 and 617 views. Our own replies carried 2, 3 and 11.
So the reply did not borrow the audience. It borrowed the question. Those are two different assets and only one of them is free.
What the 11 had that the other three did not: that author had said out loud that he would try whatever people sent him. The other three were general "what are you building" rooms. Same handle, same hour, same shape of reply. The only variable was whether anyone in the room was actually waiting for an answer.
Which sharpens your step one. "Threads where someone named the problem" is close, but the harder filter is "threads where someone asked for a thing and said what they would do with it." A named problem still leaves you guessing whether anybody wants it solved this week.
The follower part is the bit I would put in bold. 102 to 270 and the same impression bucket is the cleanest evidence I have seen that follower count is a vanity input into a system that ranks rooms, not accounts. Most people will read that paragraph and still go and buy followers.
One question, because I genuinely do not know the answer for our own accounts either. Did any of those 168 new followers ever turn up inside your product, or did they only ever turn up in the follower count?
For where the numbers come from: I build AI Applyd, which applies to jobs on people's behalf. We sell to job seekers, not founders, so nothing here is competing with you.
You are right about the asset. I was sloppy calling it borrowed impressions. A reply sits under someone else's question. It does not inherit that thread's view count as if those people saw me.
I do not have reply-level views as clean as yours (2, 3, 11 against 196–617 on the parents). What I have from X Analytics this week is replies listed at 2.6K, 2.3K, 1.3K, 882 versus original posts at 23–120. That still describes the room, not how many people actually saw my sentence.
The "said they would try whatever people sent" filter is sharper than the one I wrote. I have been selecting on "named the problem." I have not been logging whether anyone said they would act this week. That is the next column.
On the 168: I do not have the join. 102 to 270 was that window. Today the account is 340 followers and I have 24–27 users who use the product infrequently. I cannot tell you those 168 ever opened it. Most of them are probably only in the follower count. That is the hole.
The 102→270 growth is useful evidence that the distribution channel works, but the harder signal seems to be what happens after the conversation. Are those profile clicks turning into product trials or customers, or mostly into more followers?
Mostly more followers. I cannot show you a paid conversion from those profile clicks.
Live numbers today, same account: 340 followers, +24 net in the last 7 days. Sent2X dashboard: 16 original posts, 703 impressions, 32 engagements. X Analytics on those posts: most sat at 23–120 impressions. One connect post hit 218.
Same 7 days, replies into other people's threads: 2.6K, 2.3K, 1.3K, 882 impressions on the top four. That is the borrowed denominator. It is still not a sales channel I can forecast.
On Sep 1 I wrote that we had 24 users and they use the product infrequently. A day later I had 27 on the Chrome extension. Revenue is still effectively $0. So 340 followers is a leading indicator that the rooms were right. It is not evidence the product got a fair test. If a paid conversion shows up from a reply-sourced click I will update this thread.
That distinction between finding the right rooms and actually getting a product test is pretty clear from those numbers. I’d be interested in digging into what the 27 users are actually doing with the product, especially given the low frequency. Happy to continue privately — what’s the best email to reach you on?
I'd rather keep it on the thread.
What I actually know about the 27: they signed up and they do not use it daily. I said that on Sep 1. I do not have a session table I can paste here. Dashboard today is still the same picture — 340 followers, 16 original posts, 703 impressions — not a heatmap of those 27.
So the honest read is activation, not a missing intro email. They are in the product. They did not build a habit. If I get a cleaner usage split I will post it here.
That’s a fair read. If they’re activating but not forming a habit, that’s a much more useful signal than the follower growth anyway.