Twenty minutes gets you from nothing to a Snapshot of your own channel, a competitor set you're tracking, and one decision memo you could hand to someone.
Here's the order and, more usefully, what each step is actually buying you, so you can skip the parts that don't apply.
Minute 0 · What you need before you start
- A YouTube handle or channel URL. This is the one hard requirement. Everything in the first read comes from public YouTube data.
- A website URL works as a fallback if you don't have YouTube, but it's a weaker starting point.
- Instagram and TikTok handles, if you have them.
- Two to five competitor names. Optional at intake, and worth having ready. See below.
You'll need to create an account. The Snapshot is free but not anonymous. It's tied to a brand record that persists, which is what lets you compare against it later.
Minutes 1–4 · Intake
Start here. The form asks for more than a handle, and each field changes the output rather than being data collection for its own sake.
Outcome lens. More sales, more qualified leads, better creator ROI, more efficient production, stronger engagement, faster growth. This sets what the analysis optimises its reading for. The same channel read through "more sales" and through "faster growth" produces genuinely different recommendations, because the definition of a good post differs.
Pick the one your next quarter is actually judged on, not the one that sounds most ambitious.
Content format. Short-form, long-form, carousels and static, or everything. This narrows what gets analysed. If you're a long-form channel that also posts Shorts, choosing "long" stops the Shorts from dominating the read on volume alone.
Competitors. Up to five, typed in by hand. Worth spending two minutes on rather than thirty seconds. This set decides what "how you compare" means, and a badly chosen cohort produces a confident, useless comparison. The rule of thumb: same platform, same rough audience scale, same category. One channel a step above you is instructive; one a hundred times your size is not.
Minutes 5–8 · Generating
It takes a minute or two. What's happening: your public uploads are being fetched and analysed, your competitors triangulated, the patterns extracted, and the summary composed.
Worth knowing about the division of labour, because it explains the shape of what comes back. The numbers are computed by code from fetched data. The words are written by a language model. The confidence comes from how much data there actually was. The model composes sentences around figures it isn't permitted to originate, which is why it will sometimes tell you there isn't enough here to say, instead of producing a confident-sounding number. That's deliberate.
Minutes 9–14 · Read the Snapshot
The Snapshot has a specific structure, and each part is answering a different question. Reading your first Snapshot walks through it properly, but the short version:
- The header carries your evidence tier: what this read is based on, and therefore what it can support.
- Totals are the volume view: how much you've made, and over what period.
- Per-channel sections break it down by platform, with a cadence heatmap showing your actual publishing rhythm rather than your intended one.
- The headline is the single strongest pattern found, with its confidence band and evidence tier attached. Read the band, not just the sentence.
- What's working ranks your top performers by format, from real public view counts.
- How you compare is the competitor triangulation.
Spend most of your time on the headline and its band. A strong-sounding statement with a wide band is a hypothesis; a duller statement with a narrow one is something you can act on.
Minutes 15–20 · Pick one of three next moves
Three doors, depending on what you came for.
If you have a decision to make → generate a Signal Brief
The Snapshot describes. The Signal Brief decides. It's a memo with a recommendation, options, risks, and every claim carrying its confidence and its evidence tier. It's the thing you'd actually forward to someone.
The free account includes a one-time credit grant, enough for one generative read: a Signal Brief or a Concept Read. Not a monthly allowance: a single taste, so you can see the real output before deciding whether the engine is worth paying for. Spend it on the question you actually have.
If you have an idea you're unsure about → run a Concept Read
Describe something you're considering making. You get a confidence band on it (Promising, Mixed or Weak), grounded in your own data and your category.
It's a band, not a prophecy. What it's genuinely good at is killing ideas cheaply, which is worth more than it sounds.
If you're here for the category → build a watchlist
Competitor Snapshot and watchlists track the set you named, continuously. The Morning Brief on your Overview is what tells you when something changed (a competitor's cadence shifting, a post breaking out) rather than you finding out a quarter later.
This is the highest-leverage thing in the product for the least ongoing effort, and it's the one most people set up last.
What to do in week two, not today
Deliberately not in the first twenty minutes:
Connect Instagram, TikTok and YouTube analytics. This is the biggest single upgrade available to you. It moves the read from public view counts to your real reach and retention, which is tier 2 evidence and a genuinely different instrument. It's read-only. It takes a few minutes per platform. But do it once you've seen the tier-1 read, so you can feel the difference rather than being told about it.
Explore the rest. There's a lot: Ad Lab, Ad Research, Experiments, the Content Calendar, the Creator Network, Delivery Rooms. None of it is useful before you've got one honest read on your own channel and one decision that came out of it.
The honest limits
Worth knowing on day one rather than discovering later:
- The first read is YouTube-only unless you connect other platforms.
- Meta ad results are pending app review. Ad Lab builds campaigns from organic winners today; the results half of that loop isn't live yet.
- Engagement replies and DMs are drafted by Acumin, sent by you. Every message goes out under your hand.
How to use this tomorrow
Do the twenty minutes. Then answer one question out loud: what did the headline say, and how wide was its band?
If the band was wide, your next move is more evidence: connect a platform. If it was narrow, your next move is a brief.
Related: Reading your first Snapshot is the detailed walkthrough of minutes 9–14.