The Headlines Cruel Instruction

Writing Headlines Is Just Deleting Until Something Stays

The headline cruelty part is not about being harsh on readers. It is about being harsh on your own first thirty attempts. Most people write the thing they want to say, not the thing a stranger will click. There is a gap between those two and the only way to close it is iteration, not inspiration. I call the whole process the cruel instruction because it forces you to kill your favorite line. You start with one you think is clever. Then you test it against a bunch of variations that are blunter, uglier, more specific. The ones that win are usually the ones that would make you wince to put your name on. That is the cruelty. Not the mechanics. Here is how it actually works when you sit down to do it.

The Method

Write twelve headlines in twenty minutes. No editing pass while you draft. Just spit them out. Use patterns you know. Numbers, contradictions, direct promises, the question, the negative frame. Pick a lane for each one so you are not making twelve variations of the same thought. Then you rank them on three axes: specificity, clarity, curiosity. Specificity is how concrete the headline is. Clarity is whether a tired human reading it on their phone understands the topic in under two seconds. Curiosity is the gap between what the reader knows and what the headline suggests they will learn. If a headline scores high on curiosity but low on clarity, it is going to get clicks and burn your content immediately. That is the most common failure mode I see, by the way. After ranking, cut it down to five. Not your five. Pick the five that cover the most ground. You want range in that final set. If all five sound like they belong to the same product, you did not get enough range. Range matters more than quality at this stage. Quality comes later.

Then you test. If you have real traffic, you run an A/B or multivariate test. If you do not, you use cheap signals: scroll depth on a landing page, click-through from an email subject line, engagement on social. Subject lines and headlines are close enough cousins that running a cruel instruction loop on subject lines is a legitimate proxy when you are small. One detail beginners miss: put the headline in context. A raw string of words does not reveal how it reads against the image, the opening paragraph, or the surrounding layout. Test it where it will actually live. I learned this the hard way on a B2B SaaS piece where a headline performed two points better in a vacuum than it did in the actual page layout. The hero image was stealing attention from the text. Moving the headline above the fold and reducing the image opacity fixed it.

Edge Case I Hit Recently

Last month I ran a cruel instruction pass for a fintech explainer. The front-runner was something like "Your Savings Are Losing Money While You Check Your Balance." It scored well on clarity and specificity. The curiosity angle was fine too. I ran it through a quick lander test. It underperformed my backup headline by a wide margin. I spent two days staring at it. The problem was not the headline. It was the post-click experience. The article used jargon like "real rate of return" and "inflation-adjusted yield" in the first three paragraphs. Readers who clicked the headline expected a simple, emotional hook and then got hit with textbook language before they were oriented. I rewrote the opening to match the headline's voice and brought the explanation down a notch. The headline then outperformed by four points. The lesson was blunt: the cruel instruction works on the headline in isolation only if the rest of the piece can actually deliver what the headline promised.

Counter-Intuitive Stuff

First, longer headlines often beat shorter ones in non-social contexts. I know that goes against every shorthand you have ever read. A headline like "How Small Budgets Beat Big Competitors Without Paying For Ads" can outperform "Beat Competitors Without Ads" because the specificity reduces the bounce. Length is fine when every word earns its place. It is not fine when you are padding for rhythm. Second, negative frames are stronger than positive frames for awareness but weaker for conversion. "Stop Doing X" pulls attention. "Start Doing Y" pulls action. If your goal is a CTR, lean negative early in the funnel and positive closer to the offer. This is not a rule. It is a pattern I have seen across dozens of landing pages and email campaigns. Treat it as a starting bias, not a law. Third, numbers in headlines are not a magic bullet. They help when the number is surprising or specific. "7 Tips" is noise. "We Cut Churn By 14%" is signal. If you are using a number, make sure it changes the meaning. Otherwise drop it.

Pitfalls To Avoid

Do not optimize a headline in isolation from the audience segment. A headline that works for founders will trash a headline that works for mid-market operators. Segment your tests. If you can only test one group, acknowledge that limitation and do not generalize the result. Do not let the cruel instruction turn into endless tweaking. Twenty minutes per round is a hard cap I use. After that, the variations stop being variations and start being paraphrases. You gain nothing from that. Do not confuse engagement metrics with outcome metrics. Likes, shares, and even time on page can lie. If a headline gets clicks but the readers leave at paragraph three, you solved the wrong problem. Pick an outcome metric that matches your goal and optimize toward that.

What This Approach Loses

It is slow if you do not already have a workflow. A clean cruel instruction pass takes about forty-five minutes when you are practiced. First few times it will take two hours because you will second-guess yourself. That is normal. The process also does not recover from a weak core idea. No amount of headline cruelty will fix a piece that is actually uninteresting. You need something worth saying before you can be ruthless about saying it well. If you are dealing with a situation where you cannot run tests at all, use heuristic scoring instead of guesses. Write a simple rubric: specificity, clarity, curiosity, and tone match. Score each headline one through five on each axis. Multiply clarity by specificity and add curiosity. Pick the highest total. It is not as good as a live test but it is faster and it keeps you from picking the same headline three times in a row.

A Practical Walkthrough

Take a working topic. Let us say it is about switching project management tools. Draft twelve. Some with numbers. Some with a negative frame. One as a question. One as a direct promise. One that calls out the reader. One that uses a counterintuitive claim. Stop after twenty minutes even if you only have eight. Rank on the three axes. Cut to five that cover different angles. Test the five against a real audience slice. Measure the metric that matches your actual goal. Update the headline. Repeat until you are hitting the target or you are running out of distinct ideas, which usually happens after the third or fourth pass.

I keep a running log of what wins and what fails across projects. It is boring but it works. After a dozen projects you start seeing patterns that are specific to your niche. Those patterns are more valuable than any headline formula you will find online.

The Downloadable Bit

There is no official template from the people who talk about the cruel instruction. What I use is a simple spreadsheet. Columns for headline text, axis scores, test variant, audience segment, sample size, metric, result, and notes. Rows are your rounds. Over time you build a reference you can query when you are stuck. If you want one, I can share the sheet format. I keep it in a public gist and the link stays stable. Look for it under the name "headlines-cruel-instruction-log" on my profile if you want to clone it and fill it in. That said, the sheet is not the method. The method is the discipline of writing, ranking, cutting, testing, and recording. The sheet just stops you from forgetting which headline lost and why. People who skip the recording step tend to repeat the same mistakes for months.

When It Fails

Brand-led messaging sometimes fights this approach. If your editorial voice is built around a specific tone and a blunt headline breaks that tone, you can get a short-term CTR bump and a long-term brand drag. I have seen it happen with luxury brands trying to run cruelty loops meant for direct response. It does not map cleanly. In those cases, use a milder version: write the headline set, but score it against a tone consistency axis in addition to the usual three. Drop anything that breaks the tone unless the data shows you can afford to. Another failure case is highly regulated industries. If you are writing for healthcare or finance and your headline claims imply outcomes, you may not be able to test the strongest versions legally. Stick to descriptive headlines and optimize around accuracy instead of aggressiveness. The cruelty becomes about precision, not punch. If you need a quick reference, the three-axis ranking and the twenty-minute cap are the parts that matter most. Everything else is decoration. Run the loop, log the results, and let the next round teach you something you did not know from the last one.