The Review-Mining Playbook: Find Ad Angles Hiding in Your Reviews and Competitor Comments

Roughly nine out of ten ad concepts we write start with a sentence we didn't write. A customer did, in a review, months ago, and forgot about it.

That's not me being lazy. It's the opposite. The most expensive mistake I see Shopify brands make with creative is sitting in a room trying to invent what their customer wants, when their customer has already told them, in their own words, for free, hundreds of times over.

Your reviews, your comments, your competitors' one-star rants. That's a copy bank. And almost nobody mines it properly.

Here's the workflow our creative team runs during onboarding for every brand we take on. It's not clever. It's just disciplined. You can run the whole thing yourself this week.

Why mined copy beats written copy

Quick bit of theory first, because it changes how you read everything below.

People don't buy what your product does. They buy who it helps them become, and the real reason is usually one they'd never say out loud. A bloke buys the premium version "for the build quality." In reality he wants to feel like the kind of person who owns the premium version. The stated reason and the true reason are different, and the true one is the one that sells.

You will not invent that true reason from a whiteboard. But customers leak it constantly in reviews, usually by accident, in the throwaway line at the end. "Honestly I just feel put-together when I wear it" tells you more than any feature list. The job of review mining is to catch those leaks and put them to work.

The other reason this works: you're writing for what one copywriter I rate calls the idiot brain. Not because your customer is dim, but because they're scrolling, half-watching telly, thumb already moving. Copy that mirrors the exact words in their own head gets understood at a glance. Copy you invented makes them translate. On a feed, the translation tax is fatal.

1. Set up the swipe file before you read a thing

Before you open a single review, make somewhere to put what you find. A simple doc or sheet with four columns: the exact quote, the source, the desire or problem behind it, and a tag for which feeling it's hitting.

The rule that makes this work: copy and paste verbatim. Do not tidy it up. Do not fix the grammar. The clumsy phrasing is the gold, because it's how a real person actually talks. The second you polish "it doesn't make my skin freak out" into "suitable for sensitive skin," you've thrown away the thing that would have stopped the scroll.

A swipe file you build once is yours forever. We've got brands where the file we started in week one is still feeding hooks two years later.

2. Start with your own reviews and read for I-want statements

Your own product reviews come first, because that's your actual buyer, not a lookalike.

Read with one filter on: you're hunting for I-want and I-hated statements. Not the star rating, the language. Someone writing "I wanted a pre-workout that didn't leave me jittery" has just handed you a desire and the objection it's solving in one line. Someone writing "my old one chipped within a week, this hasn't" has handed you the before-state your ad should open on.

Three kinds of line are worth grabbing every time:

• The desire stated plainly. "I just wanted something that actually lasts." That's your promise, in their words.
• The problem they're escaping. "I was so sick of the cheap ones breaking." That's your hook's opening frame.
• The surprise. "Didn't expect to use it every single day." Surprise is conversion gold, because it pre-handles the doubt the next buyer has.

Don't rush this. Twenty good lines beat two hundred skimmed ones.

3. Mine the comments, where people are blunter

Reviews are semi-public and slightly polished. Comments are where people stop being polite, and that bluntness is useful.

Go through the comment sections on your own organic posts and, more importantly, your own ads. The objections that come up again and again ("but does it work on X", "looks pricey", "is this just a dupe of") are the exact doubts killing your conversion rate. Each repeated objection is a future ad that answers it head-on.

Social comments, your reply DMs, even the questions your support inbox gets ten times a week. All of it goes in the file. The phrasing people use when they're frustrated or sceptical is sharper than anything a brand writes about itself.

4. Raid your competitors' reviews, especially the angry ones

This is the step most brands skip, and it's often the richest, particularly if you're newer and don't have a deep review pile of your own yet.

Go to the bigger players in your category and read their reviews. Two goldmines here:

The five-star reviews tell you what the category loves, the language that makes people happy, the benefits worth claiming. The one and two-star reviews are even better. Every complaint is a gap. "Great product but the strap broke in a month" is a promise you can make if your strap doesn't. Their worst reviews are your positioning, served on a plate.

Where to look: Amazon listings, Reddit threads, Quora, the comments under competitor ads, and review widgets on competitor sites. You're not stealing a product. You're listening to a market that's already told someone else exactly what it wants and isn't getting.

5. Sort the haul into desires, not features

Now you've got a messy pile of quotes. Time to make it usable.

Group the lines by the underlying desire or problem, not by product feature. You'll usually find your customers cluster around three or four core desires, and they are rarely the ones you'd have guessed. A homewares brand might assume people buy on looks and find half the reviews are actually about a quiet feeling of having their home sorted.

For each cluster, write the plain-English version of the desire at the top, then stack the customer quotes under it. Now you can see which desires show up most, which have the most emotional language, and which you've never once mentioned in your ads. That last group is where the easy wins usually hide.

I believe this sorting step is where review mining goes from a nice idea to an actual creative engine. The raw quotes are ingredients. The clusters are the menu.

6. Turn the quotes into hooks, headlines, and scripts

This is the payoff. Your sorted file now writes three things for you.

Hooks. The first line of the ad, the bit that has to stop the thumb. Pull the bluntest customer line in a cluster and use it almost verbatim. "I was so sick of pillows that go flat by 2am" is a better hook than anything with the word "introducing" in it. Apply the squint test: if you squint at the phone, is it instantly clear? If it needs a second read, it's too long for the feed.

Headlines. The on-image or on-screen text. Same source, tighter. Often it's just the desire cluster's title turned into a line of three to six words.

UGC scripts. This is where mined copy really earns its keep. Hand your creator the actual review quotes as their talking points. A creator reading a line a real customer wrote sounds real, because it is. Build the script as before-state, then the line that turned it around, then the proof. All three pulled straight from the file.

One thing worth saying plainly: the words alone aren't enough. If the cluster you're chasing is "this made me feel sorted," the footage has to look sorted too. You can't bolt a calm, put-together promise onto frantic, cluttered footage and expect it to land. Mine the copy, then shoot to match it.

7. Treat every concept as a test, not a verdict

Last one, and it's a mindset more than a step.

You don't actually know which desire is strongest until you spend money finding out. So build one concept per desire cluster, run them against each other, and let the market vote. A few thousand dollars across three or four angles will tell you more than any amount of arguing in a meeting.

The winners aren't your opinion. They're a signal the market sends back. Mine the copy, build the angles, ship them, read the response, then pour budget into whatever the data points at. The reviews tell you what to say. The test tells you what's true.

So this week, before you brief another ad, spend an hour in your own reviews and an hour in your biggest competitor's. Build the four-column file and just see what's sitting there. Reply and tell me the strangest desire you find buried in there, the one you'd never have written yourself. That line is usually your next winning ad.