Everyone says August is dead. I checked our market's daily data — it's down 8%, not 40%
I run a small tool that crawls the Shopify App Store every day, so I have a daily record of the whole market. When the usual "August is dead" wave started in the dev forums this month, I could actually go check instead of nodding along.
Short version: August is down about 8% vs late July. Not 40%.
I used reviews-per-day as a rough proxy for installs, across a fixed cohort of 501 apps — the same apps in every window. June ran ~119 reviews/day, second half of July ~115, first half of August ~106.
Now the embarrassing part, which is honestly the more useful half of this post. My first pass gave a different answer and it was wrong. I'd measured reviews per day across everything we had crawled, which meant I was mostly measuring when our crawler ran, not when reviews were actually left. The dips lined up with our own scheduling gaps. I had to bin the numbers and redo it on a cohort with complete history through the whole window. If you're pulling stats out of your own product's database, that's the trap waiting for you: the data describes your infrastructure before it describes your market.
What I think this means if your August looked terrible:
If the market moved 8% and you're down 40%, the other 32 points are yours. At the scale most of us operate at, distribution moves way more than demand does. You get a handful of organic installs a day, and slipping two positions on a single search term can halve that. A competitor rewriting their listing, a new entrant, a routine reshuffle — in your own install chart they all look identical to "a quiet week with no obvious reason". And a quiet week with no obvious reason is exactly what gets filed under seasonality.
The reason the distinction matters: if the market is asleep, waiting is the correct move. If your position changed, waiting is the one response that cannot possibly help. Same chart, opposite decision.
Method and the daily chart here if you want to poke holes in it: https://bestappify.com/blog/is-august-really-the-shopify-app-store-s-dead-month-we-measured-it-41
If anyone here sells into Shopify and wants a cut of this for their own category, say the word and I'll pull it.
a fixed cohort w/ complete history quietly drops the apps that died mid window tho. does that pull the 8% down a bit or is the survivor effect too small to matter?
Real point, and it doesn't get waved away as noise.
The 501 is a fixed cohort of apps with complete review history through the end of the study — so by construction it drops anything that stopped getting reviews (or got delisted) partway through the window. That cuts the way you'd think: an app dying mid-window would show a cliff to near-zero, not an 8% dip, so leaving it out almost certainly makes the market look healthier in August than it actually is.
We don't have a clean way to size that effect from what we measured — we held the cohort constant on purpose and didn't separately track apps that dropped out. So I can't tell you if the true number is 10% or 25%. What I can say honestly is that 8% is a floor, not the full answer — August is down, and probably down more than this cohort shows.
The positioning fights itself in two ways worth naming. First, "more than just an AI tool, a community, or a business plan generator" defines you by what you're not, three times, which leaves a reader with no idea what you ARE. "Ecosystem bringing three things together" is what everyone says right before nobody can describe the product back to them. Negative space isn't positioning.
Second, and harder: Explorers, Inventors, and Founders want opposite things. An explorer wants to learn with zero stakes; a founder building something real wants feedback and users, and has no patience for a curious dabbler's questions. Serving all three means the founder shows up and finds tire-kickers, while the explorer shows up and finds intimidating real builders. A product for everyone at every stage is a product for no one at any stage.
And the whole thing rests on a two-sided problem you don't name: "real community feedback" only works if the community is already there and motivated to give it. Early on you have neither side. Who gives thoughtful feedback to the first ten ideas, before there's an audience to reward it?
Which of the three, explorer, inventor, or founder, would you lose last if you had to drop two? Start there.
I think this landed on the wrong post — there's no AI tool or explorer/inventor/founder tiering here. This one's about whether Shopify App Store activity actually drops in August, measured against a fixed cohort of listings.
Worth reposting wherever you meant it, though; it reads like careful feedback and it's wasted sitting here. And if you did mean this post and I'm missing the connection, say which part and I'll answer it properly.
You're right, that was meant for a different post, apologies for the mix-up. Here's what I actually wanted to say about this one, because your buried conclusion is stronger than your headline.
"The data describes your infrastructure before it describes your market" is the best line here, and it generalizes past this chart, anyone pulling stats from their own DB should tattoo it somewhere. You measured when your crawler ran, not when reviews landed, and caught it. Most people ship the first number.
But the sharper thing is a product, not an observation. "If the market moved 8% and you're down 40%, the other 32 points are yours" is the whole game, because inside your own install chart, seasonality and position-loss look identical, a quiet week with no obvious reason. A founder physically cannot tell them apart from inside their own data. Your cross-market baseline is the only thing that separates them, and that separation is the difference between "wait it out" and "your listing got beaten," which as you said is the same chart, opposite decision.
So your real product isn't "August numbers," it's subtract the market from my chart and show me what's left. Market-adjusted install signal. Your offer to pull a cut per category is circling this without naming it, the value isn't the market data, it's the founder's own delta against it.
For your 501-app cohort, can you already compute a single app's expected-vs-actual gap? Because that gap is the thing people would pay for.
That's the right name for it, and no, we don't have that built for a single app yet. What we ran was the aggregate: 501 apps with complete review history, coming in 8% below late July. That gives you the market baseline, but turning it into "here's your app's gap against that baseline" is a different product from what we shipped — right now it'd be a manual pull, not something live.
The mechanic you're describing is exactly right though: expected volume from the category curve minus actual volume is the only way to tell a quiet market week from getting outranked, from inside your own chart alone. We just haven't built the per-app version.
(I work on BestAppify, that's where the numbers come from)
Yes and no, and the gap between those two is the interesting part.
Computable: yes. Same tables — it's the cohort query with n=1 on one side. Shipped as a feature: no. What's in the product is the position half — rank tracking, a visibility index, competitor moves — which tells you whether your position slipped, but the founder still has to hold the market context in their head.
Two things stop it being a one-click number, and both are the same failure mode that bit me in this post.
The baseline choice does most of the work. A 500-app store-wide cohort is stable but too coarse — a subscriptions app and a shipping-label app don't share a season. Cut it to the category and n drops to 20-40, where one competitor's review burst swings the whole baseline. So the error bars on the residual depend on a choice the user never sees, and a confidently wrong "you're 32 points below market" is worse than no number at all.
And my install proxy is reviews. Public, but noisy and lagged at the single-app level. Fine across 501 apps, shaky for one. The clean version needs the founder's own install data joined to my market line, which makes it exact only for people who connect their analytics and approximate for everyone else. That tradeoff is the actual product decision, not the maths.
But you're right that this is the thing, and naming it that way is sharper than how I've been describing it. "Subtract the market from my chart" is the one calculation a founder cannot do from inside their own dashboard. Everything else I sell is downstream of it.
If you want to test the claim rather than take it: name an app in the store and I'll pull its expected-vs-actual for the August window and post the number here, whichever way it comes out.
The strongest part is the corrected methodology. Separating market-wide movement from your own distribution changes the decision completely—“the market is down” and “my position is down” require very different responses.
Thanks — that split was the whole reason I redid it. What made it concrete for me is that the two look identical in your own install chart. Same shape, same week, and nothing in your own dashboard tells you which one you're looking at. You can only get the answer from outside your own data, which is an uncomfortable thing to realise about a number you check every day.
The rule I landed on: if you can't see the market's line next to yours, don't file a dip under seasonality. Default to assuming it's your position, because that's the one you can actually do something about.
That’s an interesting distinction. I’d be interested in hearing what you learn from applying that rule over time.
Will do — and I'll post it either way, including if the rule turns out to be wrong. The honest test is how often a dip I'd have filed under "quiet week" has a position change sitting underneath it. I don't know that ratio yet; I only know the two are indistinguishable from inside one app's chart. Once I have enough cases to count, that's a post.
That’s a useful test. I’d be interested to see what the first real cases reveal once you have enough volume.