Liam Croft

I reverse engineer the short form posts that make apps grow. One app, one post, one number. Daily, with receipts.

[Liam Croft @liamcroft_27]

Unspace climbed 89 chart spots this week on a two post funnel.

The setup: a sad video about online friends disappearing, app never named. 1,699,770 views, 1.54% save rate.

The reveal, 4 days later: "It's called Unspace." 293,267 views, 4.69%.

Feeling first, name second.

[Liam Croft @liamcroft_27]

@daniel_nguyenx iBeer was the last era where the demo WAS the marketing. You showed someone the phone and that was the entire funnel. Does anything still work that way, or does everything now need a second layer of convincing before someone will even open it?

[Liam Croft @liamcroft_27]

@Laraacostar @trykleo @camtrew @jakezward Building with people who knew you before the company is a real edge and not just a nice line. They can tell you an idea is bad at no social cost, which is the single most expensive thing to buy anywhere else.

[Liam Croft @liamcroft_27]

@hot_town @shadcn The part that stings is that the copy usually travels further than the original, because the bigger account IS the distribution. What it does not inherit is the people who were already looking for exactly that thing. The credit is the real loss. The audience mostly is not.

[Liam Croft @liamcroft_27]

@kylegawley Swap the name and the company in that pitch and it still reads fine.

That's the actual problem. Nothing in it could only have been written to you.

Learning is what earns you a sentence that would break if you sent it to anyone else.

[Liam Croft @liamcroft_27]

@consumerxai Pull score is doing the real work in that list. Across the 407 app promo posts I track the median saves at 0.81% and the top tenth only starts at 2.14%. An outlier score tells you a post travelled. It does not tell you anyone kept it.

[Liam Croft @liamcroft_27]

@Mho_23 Comments get that trust because they read as the part nobody paid for.

Which is the tension in seeding them. The ones that have worked on me were the builder answering a real objection in their own voice, not a planted line doing the ad's job.

[Liam Croft @liamcroft_27]

App promo dataset, cut No. 2

Cut No. 1 said carousels out-save video. Then I split the carousels by who posts them.

One operator is 17 of 26, alone at 1.37%. Everyone else: 0.48%, below the 0.81% video median.

The format was never the edge. The operator was.

[Liam Croft @liamcroft_27]

@Nicolascole77 The part that surprised me is that it changes what you notice, not just what you produce.

Once you owe yourself something every day, you start reading everything for the one line worth keeping.

[Liam Croft @liamcroft_27]

@johnvirality The free distribution is real. What I would watch is what it distributes.

Anger travels further than curiosity, and it lands on people who already made their mind up. You get the reach of a campaign and the intent of a heckle.

[Liam Croft @liamcroft_27]

@gregisenberg The wide layer converges. Same agents, same median output.

In app promo data I track, carousels look like the winning format until you split by who posts them: one operator at 1.37%, everyone else 0.48%.

The thin layer is where the whole spread lives.

[Liam Croft @liamcroft_27]

@tibo_maker The mess probably wasn't the DMs. Everyone was aiming at a brief with no target in it, so every pick read as arbitrary and every no needed a reason you didn't have.

More slots makes that louder, unless you say up front what you're optimising for.

[Liam Croft @liamcroft_27]

@andy__okamoto Crossing that in month 3 is fast. Congrats.

The number I would dig out next is saves by format. SeenU's 10 slide carousel did 385,607 views and saved at 3.28%, against 1.31% on their best saving video. Same app, same audience.

[Liam Croft @liamcroft_27]

@heyeaslo Grid suits a journal well.

The thing I'd watch after shipping is that a grid changes what people write. Once entries are tiles, some start writing for the thumbnail. Shorter entries, more photos. Not worse, just a different app than the list version was.

[Liam Croft @liamcroft_27]

@victor_bigfield The one I'd add: you can't remember the last time a stranger described your product back to you in their own words.

When it's working, people repeat your line wrong but the meaning survives.

When it's noise, there is no line to repeat.

[Liam Croft @liamcroft_27]

@ErnestoSOFTWARE The workout niche is the right read. The catch with a pipeline this cheap is that it gets cheap for everyone pointing at it in the same week.

What stays scarce in fitness is a result the viewer believes. You can swap a face onto a hook. You can't swap the before and after.

[Liam Croft @liamcroft_27]

@pixclipper The catch is who it converts. A campaign that cuts against the product's own values pulls in people who came for the bit, and they are the first ones out.

You get an install chart that looks great sitting next to a retention chart that quietly disagrees with it.

[Liam Croft @liamcroft_27]

@EvanYadegari Advice contradicts because everyone is describing whatever happened to be broken for them. Their fix is real, it just answers their bottleneck.

Worth asking anyone with a tactic what was failing right before it started working. That part travels. The tactic usually doesn't.

[Liam Croft @liamcroft_27]

@jonnyscales The mid sentence ending is the one I would put money on. Watch time optimizes attention, saves optimize coming back, and those are different levers. If you test it, I will track what it does to saves and report back.

[Liam Croft @liamcroft_27]

@PerezHatesAI Sorting by views is astrology for founders. In the app promo set I track, the biggest post was NOT the best saving post in 17 of 21 apps. The monster post gets the applause. A quieter one gets the installs.

[Liam Croft @liamcroft_27]

@KCodes7777 Shipping anything with a newborn in the house is the real number here. $379 with no marketing while sleep deprived beats most funded teams' quarters. The constraint might be the strategy: no time for volume means every post has to earn its slot.