The phrases that predict willingness to pay.

If you're scanning a subreddit and counting upvotes, you're measuring attention, not appetite. Ten people mildly annoyed about a slow dashboard is not the same signal as one person typing "I would pay real money for this." The words carry the intent. Volume is noise until you weight it.

This is the core mechanic behind how BuyerTell scores every post, comment, and review it mines: not just did someone complain, but what tier of intent did their language imply. There are five tiers, each with a weight derived from how close the speaker is to a purchase decision. Below is the full ladder, with example phrases, why each weight is calibrated the way it is, and a practical method for scanning your own communities the same way.

Tier 1 — Willing to pay (×1.8)

Phrases: would pay for, take my money, instant buy, shut up and take my money.

This is the ceiling. Someone has skipped straight past "this is annoying" to naming a transaction. They are no longer describing a problem — they are pre-qualifying themselves as a customer. It's rare (most people vent, they don't pitch themselves as buyers), which is exactly why it carries the highest multiplier. When this phrase shows up attached to a specific, nameable gap, treat it as a near-direct quote for your landing page.

Tier 2 — Switching (×1.5)

Phrases: alternative to, canceling my subscription, too expensive, looking to switch from.

Switching intent is powerful for a different reason than tier 1: it implies an existing budget line. Someone already pays for a category of tool — they're not being asked to discover a new expense, only to redirect one. That's a fundamentally easier sale than educating a market from zero, which is why it sits just below "would pay for" rather than far beneath it.

This is also where 1★ and 2★ App Store reviews concentrate almost entirely. A person who paid for a subscription, used the product, and is now angry enough to leave a public review is emitting switching intent at a density you won't find on Reddit or Hacker News, where most posters have never opened their wallet for anything in the category. It's the single richest — and most overlooked — vein of this tier, and one of the reasons App Store reviews are a first-class source rather than an afterthought.

Tier 3 — Solution-seeking (×1.35)

Phrases: is there a tool that, what do you use for, can anyone recommend.

This tier is a direct request for a product — the person has already concluded that a tool should exist and is actively hunting. It's slightly below switching because there's no confirmed budget yet; they might be shopping for a free option. But it's a much stronger signal than a passive complaint, and communities built around this exact phrasing (Stack Exchange's softwarerecs is the clearest example) are worth mining almost exclusively for this tier.

Tier 4 — Workaround (×1.2)

Phrases: built my own, zapier can't, cobbled together, hacked together a script.

Someone has already spent time and effort solving this themselves — usually badly, and usually because nothing off-the-shelf fit. That effort is a proxy for the value of the problem: people don't build janky internal tools for things they don't care about. The weight sits in the middle of the ladder because the intent is real but undirected — they solved it with duct tape, not a credit card, so you still have to convince them a paid tool beats their spreadsheet.

Tier 5 — Pain (×1.0)

Phrases: still doing this manually, so tedious, someone needs to build this.

The baseline. This is a real, describable problem — but with no evidence yet that the speaker has moved toward paying for, seeking, or even hacking together a fix. It's the tier most complaint-scraping tools stop at, which is also why it's the most abundant and least differentiated. Useful for validating that a pain exists broadly; useless on its own for predicting whether anyone will pay to remove it.

The most valuable sentence in your research isn't the one with the most upvotes — it's the one that starts with "I'd pay for," buried three replies deep in a thread nobody else read closely.

The formula: engagement × intent × source

Weighting phrases alone isn't enough — a "would pay for" comment with zero replies might be an outlier, and a generic pain post with 400 upvotes might reflect a problem so common it's already commoditized. The full score multiplies three things together:

• Engagement — upvotes, replies, reactions, or review helpfulness votes; a proxy for how many other people share the same view.
• Intent weight — the ×1.0 to ×1.8 multiplier from the ladder above, applied to the strongest phrase found in the text.
• Source weight — some communities skew toward higher base intent (App Store reviewers already pay for something; casual Reddit lurkers mostly don't), so the source itself carries a modifier.

Results are then deduplicated — the same underlying complaint showing up in three threads shouldn't count three times as much as it deserves — and capped per source, so one unusually chatty subreddit or one viral App Store review doesn't drown out every other signal in the run.

How to scan your own communities

You don't need infrastructure to start applying this. Pick a niche community — a subreddit, a Discord, an App Store review page for a tool your prospects already use — and:

1. Ctrl-F for the tier 1 and tier 2 phrases first. They're rare, so a hit is worth reading in full context.
2. Log every match with a link, the surrounding sentence, and an engagement count.
3. Widen to tier 3 and tier 4 phrases once you've exhausted the top two — this is where volume starts to matter.
4. Look for the same underlying complaint recurring across multiple people, not just multiple posts from one person.
5. Weight by hand: multiply engagement by the tier's multiplier, and rank what you've collected.

Do this across a handful of communities and a pattern usually surfaces within an afternoon — the same specific gap, named in switching or willing-to-pay language, by people in the same profession. That's the pattern worth building for. For a broader map of where to look first, see where to find SaaS ideas — it walks through the six source types worth prioritizing, App Store reviews included.

None of this replaces judgment — a high-scoring signal still needs a real look at who's saying it and whether they can actually pay. But treating language as a ranked signal instead of undifferentiated noise is the single highest-leverage change you can make to how you read a community for ideas.