Most product copy gets written the same way. Someone who knows the product well sits down, writes from memory and instinct, then edits toward what they already believe about the customer. The draft reads fine. It just doesn't convert, because it was never checked against anything real.
The fix isn't better writing. It's writing from a different source.
The gap between what you believe and what customers say
That 80/8 gap Bain identified back in 2005 hasn't closed because it isn't really about service delivery. It's about how differently a company and its customers describe the same experience. A founder describes their product in terms of what they built. A customer describes it in terms of what almost went wrong, what they compared it to, and what finally convinced them. Those are different vocabularies, and copy written in the first one rarely lands with someone thinking in the second.
A copywriter on Reddit put this well recently: before touching any ad copy or landing page, they go looking for the exact words people use when they talk about the problem, not their own version of it. One well-known case study found that a headline pulled straight from an Amazon review beat a written-from-scratch control by over 400% in clicks. Same product, same offer, different vocabulary.
Three buckets: pains, objections, desired outcomes
When you're mining real language, it helps to sort what you find into three categories instead of one big pile of quotes.
- Pains. What's actually frustrating people right now, described in their own words, not the clinical version you'd write internally.
- Objections. The specific reasons someone almost didn't buy, or almost churned. This is the most useful bucket for copy, because it tells you what your headline needs to defuse.
- Desired outcomes. Not features, the actual change someone was hoping for. This is usually a sentence, not a bullet point, and it's often more emotional than functional.
Copy that addresses one bucket without the others tends to feel incomplete. Copy that pulls from all three, in the customer's own phrasing, tends to feel like it was written by someone who already understands the reader.
Where to actually find this
You don't need customers of your own to start. A few sources work whether you have ten customers or none:
- Reviews on competing or adjacent products. Filter to 2-3 stars specifically. Five-star reviews are thank-you notes. One-star reviews are often about one bad experience. Two and three star reviews are where people explain, in detail, exactly what was almost good enough.
- Reddit and forum threads in your category. Search for your product category plus "worth it," "alternative," or "vs," and read the comments, not just the original post. The disagreements are where the real objections surface.
- Your own support tickets, if you have any. The tickets people write in frustration use more honest language than the reviews they leave later, once they've calmed down.
- G2, Capterra, or Product Hunt comments, if you're B2B. Same logic as reviews, filtered for the middle-rated, detailed ones.
Pull the sentences, not summaries. A collection of pasted quotes sorted into pains, objections, and outcomes is more useful for writing than a paragraph you've already generalized from them. The specificity is the point.
Why this has to be ongoing, not a one-time sprint
Collection like this is usually treated as a pre-launch task: do it once, write the copy, move on. That's a mistake for the same reason a one-time competitor analysis goes stale. Objections drift as a market matures. A competitor ships the feature people were asking for. Your own customer base shifts as you grow past your first buyers. A pain-and-objection list from six months ago describes a market that no longer quite exists.
The collection part is also the slowest part by far, which is exactly why it gets skipped after the first round. Reading through dozens of open tabs, copying quotes one at a time, is a full day of work most teams don't have free capacity for on a recurring basis. That gap between wanting continuous signal and having the hours to collect it manually is the specific problem we built Hugo to close: point it at a category or a competitor and it keeps surfacing real language, sourced and current, instead of a one-time list that starts decaying the day you finish it.
The words that convert aren't the words you'd choose. They're the words your customer already used, before you ever asked them a question.