BeginnerLesson 5 of 305 min read

Headlines and Strong Openings

Learn what makes a headline specific and honest, run a simple headline check with a program, and compare two headlines fairly with a range that shows how much chance matters.

In this lesson you will

  • Write specific, honest headlines
  • Check length and hook
  • Compare two headlines with ranges
  • Know when a test is too small to trust

About 22 minutes

The headline decides whether anyone reads

Priya’s windowsill post is ready, and it needs a title. She knows that, on a search results page or in a social feed, the title is often all a stranger sees. A headline has one job: to make the right reader feel that clicking is worth their time. It should promise something true, be clear about what the post is, and make the reader want to see how. This lesson covers what makes a headline work, a simple check you can run, and how to compare two headlines fairly.

Habits of strong headlines

  • Specific. ‘How to Start Tomato Seeds on a Windowsill’ beats ‘Seeds’.
  • Honest. The post must deliver what the headline promises. A misleading headline wins one click and loses a reader.
  • Useful words first. People skim the start. Put the topic early.
  • Sized for display. Search results and feeds cut off long titles, and the exact point varies, so a length of about 40 to 65 characters is a sensible rule of thumb, not a rule.
  • A reason to read. A number, a question or a how-to gives a hook, as in ’10 cheap tools’ or ‘Why seedlings fall over’.

A quick headline check

The program below checks five made-up headlines against three simple tests: not too short, not too long and some kind of hook, meaning a number or an opening word such as how, why or what. It is deliberately simple. Rules like these catch only obvious weaknesses. They cannot tell you whether a headline is interesting.

import math
heads = ["Seeds", "How to Start Tomato Seeds on a Windowsill", "10 Cheap Tools Every Small Plot Needs",
         "Why Your Seedlings Fall Over (and How to Stop It)", "Some Thoughts About Gardening This Month, Part One"]
def check(h):
    notes = []
    n = len(h)
    if n < 25: notes.append("too short")
    if n > 65: notes.append("too long")
    if not any(c.isdigit() for c in h) and not h.lower().startswith(("how", "why", "what")):
        notes.append("no hook")
    return n, notes
for h in heads:
    n, notes = check(h)
    print(f"{n} chars: {h[:30]}")
    print("   " + (", ".join(notes) or "ok"))

def wilson(k, n, z=1.96):
    p = k / n
    d = 1 + z * z / n
    c = (p + z * z / (2 * n)) / d
    h = z * math.sqrt(p * (1 - p) / n + z * z / (4 * n * n)) / d
    return c - h, c + h
# MADE-UP test: two headlines, 1,000 views each
for name, k in (("A (plain)", 32), ("B (how-to)", 51)):
    lo, hi = wilson(k, 1000)
    print(f"{name}: {k / 10:.1f}% clicked, range {100 * lo:.1f}-{100 * hi:.1f}")
a, b = wilson(32, 1000), wilson(51, 1000)
print("ranges overlap:", a[1] >= b[0])
5 chars: Seeds
   too short, no hook
41 chars: How to Start Tomato Seeds on a
   ok
37 chars: 10 Cheap Tools Every Small Plo
   ok
49 chars: Why Your Seedlings Fall Over (
   ok
50 chars: Some Thoughts About Gardening 
   no hook
A (plain): 3.2% clicked, range 2.3-4.5
B (how-to): 5.1% clicked, range 3.9-6.6
ranges overlap: True

‘Seeds’ is five characters with no hook, so it is flagged twice. The middle three pass. The last headline, ‘Some Thoughts About Gardening This Month, Part One’, is long enough, yet the check flags it for having no hook, since it promises nothing to a stranger. Rules cannot write headlines, but they can show you which ones need another try.

Test two headlines fairly

The second half of the program compares two versions of one headline in a made-up test, with 1,000 views each. Version A, the plain one, was clicked 3.2 per cent of the time, and version B, the how-to, 5.1 per cent. A single percentage is only an estimate, so the program also gives a range around each, a standard way to show how much a result could move by chance. A’s range runs from 2.3 to 4.5, and B’s from 3.9 to 6.6. Do the ranges overlap? True. That means the test has not settled it. B looks better, and with this much data you cannot be sure. Run it for longer, or get more views, before you declare a winner. For a small blog, with few readers, headline testing may not be possible at all. In that case use the habits above and learn from your results over many posts.

  1. Write ten headlines for each post, fast, without judging.
  2. Cross out the vague and the misleading ones.
  3. Run the quick check on the rest, and read the best three aloud.
  4. Choose one, and make sure the post delivers what it promises.
  5. Keep a list of the ones that worked, and learn your own readers’ taste.
Common mistake: clickbait

Writing a headline that promises more than the post gives. It may win clicks in the short term, but readers feel cheated, leave quickly and do not return. Honest headlines build trust.

Small tests mislead

Do not trust one result from a small number of views. Compare headlines only when you have many views, and treat a gap smaller than the ranges as unproven.

Write eight headlines for your next post. Run the three checks by eye. Pick two, and say aloud what each promises.

Practice for this lesson

Check what you learned

Five quick questions. Pick an answer and you will see straight away whether it is right and why.

What is a headline’s main job?

  1. To fill space
  2. To list keywords
  3. To make the right reader want to click
  4. To please search engines only

It must make clicking feel worth it for the right reader.

Which headline was flagged for no hook but good length?

  1. Some Thoughts About Gardening This Month, Part One
  2. Seeds
  3. How to Start Tomato Seeds on a Windowsill
  4. 10 Cheap Tools Every Small Plot Needs

It promises nothing to a stranger.

Did the sample ranges overlap?

  1. No, so B clearly won
  2. Yes, so the test did not settle it
  3. They were not measured
  4. Only on mobile

The ranges overlapped, so the result is unproven.

What should a headline do about honesty?

  1. Exaggerate a little
  2. Hide the topic
  3. Use capital letters
  4. Deliver what it promises

Misleading headlines lose readers’ trust.

What should you do when a test is unclear?

  1. Pick the one you like
  2. Collect more views before deciding
  3. Stop writing
  4. Delete both

More data narrows the range.

Summary

  • A headline makes the right reader want to click, and must be honest.
  • Be specific, put the topic early and give a reason to read.
  • A range shows how much chance could move a result.
  • Small tests rarely settle anything.

Next lesson

Lesson 6 covers editing for clarity.

Questions

How long should a headline be?

Search results and feeds cut off long titles at different points, so about 40 to 65 characters is a sensible rule of thumb rather than a fixed rule.

Can I test headlines on a small blog?

It is hard with few readers. Use the habits in this lesson, and compare what happens across many posts instead.

Are the click numbers real?

No. They are made-up sample numbers to show the method.

Progress is saved in this browser.