How To Write A Good Blog Post Title

And why most of the things you think matter actually don’t

Stop following bad “How To” guides

artist Taylor White

Who cares?

You, probably. Most writers think they should.

And I get why…

“According to Copyblogger, 80% of your visitors will read your headline — but only 20% will go on to finish the article.”

So, people care.

But: they care about — and are looking at — irrelevant things.


WE’RE LOOKING AT THE NUMBERS ALL WRONG

I’m amazed at how many people slap together averages of top posts and call it “analysis,” actually thinking this is sufficient, let alone “significant” on any intelligent level.

First: “top post” stats say something about them, but nothing about how they’re different than the low performers.

Without comparing to poor performers, you can’t know anything significant about top ones. You have to look at the entire data set for any kind of meaningful insight.

Second: averages can mean absolutely nothing. Just because you can calculate an average (and god, do we freaking love calculating them…) doesn’t mean it should be some guiding north star.

I could tell you the average height of best-selling authors or the average name length of a baseball player, and while those may be interesting (fascinating really!), they do not tell you what actually makes a good writer or athlete.

Quit with the “averages.” We love them, but they’re illogical and we need to stop.


STATISTICAL ANALYSIS

I ran actual correlation analysis on the titles of my 470+ posts to date. (Feel free to skip this part if you don’t care about the details.)

Model used: Pearson correlation coefficient

PCC measures the linear correlation between two variables, X and Y (independent and dependent.) It’s the covariance of the two variables divided by the product of their standard deviations.

It yields a value between +1 and −1:

  • ±1 means total linear correlation (positive or negative)
  • A score of 0 means there is no linear correlation

The closer to ±1, the more correlated. The closer to 0, the less.

An Example:
I ran the correlation between “Reads” and “Views.” (Not whether everyone who opened an article read it, but rather whether everyone who read an article first opened it.) Obviously this PCC should pretty much be 1… and it came out to 0.977, with a p-value of 0.0. (i.e., yeah. We’re highly confident that it’s highly correlated.) Duh.

Makes sense, right? Great.

Let’s call “Reads:Views” our “actually freaking correlated” control.

p-value

p-value (or probability value) represents “the probability of the significance” of results —i.e., whether each PCC “matters.”

In stat talk: a small p-value (often ≤0.05) indicates strong evidence against the null hypothesis, so you reject the null hypothesis. A large p-value (>0.05) indicates weak evidence against the null hypothesis, so you fail to reject the null hypothesis.

In layman’s terms: a p-value <0.05 indicates the PCC “means something.” A p-value >0.05 suggests it may not.

(YMMV on PCC and p-values, but probably not much.)

Dependent Variable (y): “Views

Medium reports stats on “Views,” “Reads,” and “Fans.” Reads and Fans are influenced by the piece itself (content), but “Views” is more purely influenced by the title (in addition to some other factors like image, writer reputation, shares, etc., but let’s just stay focused on title for now, mmkay?) Of the 3 stats, “Views” is the “cleanest” for analyzing title effectiveness.

Independent Variables (x): [The Bullshit Tips People Mention]

I ran correlation analysis on several of the most popular factors people love to talk about.

Let’s walk through these one by one…


PART I — WHAT PEOPLE THINK MATTERS

Hint: a lot of shit that actually doesn’t.

Read most any article on “HoW tO wRiTe A gOoD bLoG tItLe” and it’ll include a lot of the same HoT tIpS.

But what ACTUALLY matters most? Almost none of them.

1. TITLE WORD COUNT

What they say: “6–7 words!”

Or, my favorite dumb insight:

“The average title word count of the top posts is 6–7 words!”

First thing I think in response?

“What was the average title word count on the worst performing posts, champ??”

Because if it’s also 6–7, guess what?? This “brilliant insight” is totally useless. (And, spoiler: it is. The average across ALL of my own posts, from best to worst, is also 6.6 words.)

What I found: IT’S NOT CORRELATED.

Yes, I did manually count up the number of words in each title of my 470+ posts and dutifully type the number in an Excel column, then ran the PCC on them (vs each post’s Views) to prove: there’s very minimal correlation.

Pearson correlation coefficient: 0.116
(“correlated” is ±1; “uncorrelated” is 0)

p value: 0.012
(PCC “matters” when p-value is <0.05)

My top 10 posts (by views) ranged in title length from 4 words (“Good Love Is Boring”) to 12 words (“8 Things I Learned Reading 50 Books A Year For 7 Years”), and of all 470+, I had titles as short 2 and as long as 16.

Conclusion / recommendation: this doesn’t matter that much. You may want to keep it shorter for display (longer titles get cut off in search results, social media, and feeds), but beyond that, whatever.

If you’re thinking: “Wait! Your post with the highest views had 7 words! Doesn’t this prove that it’s The Best?” 
The answer is: *sigh* No. (Please stop before you hurt yourself.)

2. “HOW TO,” “WHY,” etc.

i.e., “interrogative words” for those in the know.

What “they” say: “use them!”

What I found: IT’S NOT CORRELATED.

Pearson correlation coefficient: 0.041
(“correlated” is ±1; “uncorrelated” is 0)

p-value: 0.374

Only 2 of my top 10 posts (by views) had “interrogative” words in the title — “How to *really* know you’re in love” and “How To Get Over Someone You Never Dated.” There were only 6 in the top 20, and 124 (26%) total.

Conclusion / recommendation: doesn’t matter that much.

3. NUMBERS! (“TOP TEN,” etc.)

What “they” say: “use them!”

What I found: IT’S NOT CORRELATED.

Pearson correlation coefficient: 0.083
(“correlated” is ±1; “uncorrelated” is 0)

p-value: 0.071
(PCC “matters” when p-value is <0.05)

Only 2 of my top 10 posts (by views) had numbers in the title — “8 Things I Learned Reading 50 Books A Year For 7 Years,” and “The Only 3 Things I Need In A Partner.” There were 3 in the top 20, and only 31 (~6%) total.

Conclusion / recommendation: use the necessary words to write the title.

4. ADJECTIVES

What “they” say: “use them!”

What I found: IT’S NOT CORRELATED.

Pearson correlation coefficient: -0.009
(“correlated” is ±1; “uncorrelated” is 0)

p-value: 0.845

Only 2 of my top 10 posts (by views) had adjectives in the title —“The Most Important Thing In A Relationship,” and “Good love is ‘boring.’” There were 6 in the top 20, and 155 (33%) total.

People don’t internalize adjectives as “Adjectives!” For example, we read “happy” the same as “happiness,” even though one’s an adjective and the other a noun.

It’s not really about “the part of speech,” people.

Conclusion / recommendation: use the necessary words to write the title.

5. Oh, wait — we meant “POWER” WORDS!

e.g., “super,” “top,” “best,” etc

What the hell counts as a “power word?” I get the concept, but what is and what is not included?

Answer: nobody knows. Bc it’s a bogus, deliberately-ambiguous, catch-all phrase meant to stand for “anything that elicits emotion!” It means nothing.

Yet people love emphasizing them.

Ryan McCready wrote,

“6 out of the top 10 recommended articles use at least one power word…!”

*sigh*

Guys — this is NOT proof that the “recommends” are correlated to — let alone caused by — the “power words.”

How many “bad” articles have “power” words??

(The answer, for me: just as many.)

I ran my top ten and bottom ten titles through CoSchedule’s “headline analyzer,” which gives a “percentage” for “power” words (among others.)

CoSchedule’s analyzer gave SIX of my top ten titles a score of “0%” on “power words,” and the average score across all top ten was 11% — the same score as my worst ten.

“Power words” (whatever they are) are looney tunes and don’t matter that much.

6. ALLITERATION

e.g., “Ways In Which We Warp Attraction” or “The Reasonableness of ‘Being Reachable’”)

What “they” say: “use it!”

What I found: Dawg, wat… this doesn’t matter.

Pearson correlation coefficient: -0.023
(“correlated” is ±1; “uncorrelated” is 0)

p-value: 0.703

Conclusion / recommendation: dudes, whatever.

7. PLAY ON WORDS AND DOUBLE MEANINGS

What “they” say: “use it!”

What I found: I doubt I do this enough to even provide an accurate data set.

(That’s my excuse. Let’s move on.)

8. ACTIVE VOICE

e.g., “I/you succeed” vs “the success was achieved by you/me”

What “they” say: “use it!”

What I found: nothing. Because I didn’t analyze this. I didn’t need to…

This is “Writing 101.”

Unless you’re writing a user manual or some kind of cutesy story in which the passive object is also an implied subject (“the Christmas cookie tin was passed around and around— and golly was he getting tired of being felt up by grandma!”), then ACTIVE VOICE IS JUST WHAT YOU USE.

Conclusion / recommendation: uh, yeah. do the writings.

9. TITLE CASE (“In This Way” vs “In this way”)

My old pieces use sentence case, but now most all my pieces use title.

The only reason I changed it? Publications did. So I followed suit just to save them the edit (and bc I figured they’d done their homework.) That’s it.

Others have mixed perspectives,

“Articles using sentence case performed better than title case, averaging 20% more recommends… About 70% of the Business articles… were clearly using title case… Social articles about life lessons and the like were mainly using sentence case.”

ALRIGHTY, GUY. (Also, these “findings” are pretty wobbly.)

Either way… to repeat:

This is NOT proof that the “recommends” are correlated to — let alone caused by — the “title” case.

10. MEDIUM PUBLICATIONS

Not a “title” tip, but mentioned a lot. So…

What “they” say: “use them!”

What I found: IT DOESN’T MATTER (sorry, Dan Moore! ❤)

Pearson correlation coefficient: 0.115
(“correlated” is ±1; “uncorrelated” is 0)

p-value: 0.845

Four of my top 10 posts (by views) ran in publications — “How To Get Over Someone You Never Dated,” “Read This If You Think You’re An INFJ,” “Does Marriage Even Make Sense Anymore?” and “Good love is ‘boring.’” Of all 470+, 90 (~19%) were in Medium publications.

I didn’t publish in a Medium publication until my 58th post — “Why being broken up with by my boyfriend was so very traumatic and difficult for me,” (which is well beyond the recommended 6–7 words, yet was picked up by The Hit Job anyway.)

Pretty much any time a publication requests a piece (with the exception of one who spammed me and got their butts blacklisted), I submit. My percentage is “low” not because I have some Brilliant Strategy or Moral Issue, but rather because I never really care one way or the other. I’m just doin’ me.

Conclusion / recommendation: you do you, boo boo. I know publications have been amazing for a lot of writers, not just for readership but connections. (I certainly count Dan Moore as a comrade.) Follow your own north star.


PART II — OTHER THINGS I ANALYZED

Even though they weren’t on any “hot tips” lists.

1. PROFANITY

Because I use it.

What “they” say: (they don’t.)

What I found: it doesn’t matter, good or bad

Pearson correlation coefficient: 0.029
(“correlated” is ±1; “uncorrelated” is 0)

p-value: 0.529

(If you thinking: “wait — looks like profanity performs worse!” The answer is, “no.” While these 22 posts have fewer Views, the PCC suggests it is uncorrelated.)

And, since we all love averages so much (even though they are meaningless) and just to illustrate how much they can sway data in dumb directions: I will point out that the average views for titles with profanity is ~16K; the average views titles without is ~12K. (I say this totally irresponsibly, because I know at least a few readers are going to latch on to it despite what I said earlier about averages.)

NONE of my top 10 posts (by views) had profanity in the title. NONE of the top 20 did, either. Only 22 (<5%) do.

Conclusion / recommendation: just write how you wanna write.

2. THE WORD “LOVE”

Why? Because, again, I write about it.

What “they” say: (they don’t.)

What I found: it doesn’t matter

Pearson correlation coefficient: 0.089
(“correlated” is ±1; “uncorrelated” is 0)

Two of my top 10 posts (by views) had “love” in the title — “How to *really* know you’re in love” and “Good love is ‘boring.’” 95 (19%) total also did.

3. POINT OF VIEW (FIRST, SECOND, THIRD)

What “they” say: (they don’t.)

What I found: doesn’t matter

Pearson correlation coefficients:
You / Your / Yours:
0.099
Me / My / Mine / I:
-0.004
Our / We / Ours / Us: 0.001

(“correlated” is ±1; “uncorrelated” is 0)

Conclusion / recommendation: whatevs

PART III — WHAT *ACTUALLY* MATTERS

a.) Title success is not black and white

The tongue-in-cheek point of this entire piece (and the reason I spent literally hours upon hours running analysis on it all) is this:

Sure, your title matters.

— but(!)—

YOUR TITLE’S SUCCESS IS NOT BLACK AND WHITE!

Stop forcing arbitrary “BeSt PrAcTiCeS” where they aren’t what matters most.

Like good posts, “good titles” are not built by following check lists.

b.) What makes a good title, then?

Like I always say about anything regarding “writing”:

It’s not the shit people think.

What matters is simple:

  • Write a title about something people care about — a Big Important Question or a topic that otherwise implicitly evokes emotion
  • Write a title that suggests / promises that you will say something or provide an answer (that’s interesting! and matters!) on this meaningful topic. (Do NOT skip this step, you mongoose!!) Bonus points for the suggestion that it’s going to be new / fresh (“CoNtRoVeRsIaL.”)
  • That’s it.

Want examples?

No problem! Here are my top ten titles:

  • How to *really* know you’re in love
  • Read This If You’re Not Sure You Want Kids
  • 8 Things I Learned Reading 50 Books A Year For 7 Years
  • The Only 3 Things I Need In A Partner
  • How To Get Over Someone You Never Dated
  • Read This If You Think You’re An INFJ
  • Does Marriage Even Make Sense Anymore?
  • Oral Sex Is Not The Measure Of A Partner
  • The Most Important Thing In A Relationship
  • Good love is “boring”

And, because I love you guys so much, HERE ARE MY BOTTOM TEN, too!

  • Happy independence for you and me, too
  • What I’m doing at your place when you’re not around
  • A few whiskies and the men they remind me of
  • Darling, let me pour you a whisky
  • Mansplaining makes me laugh
  • The definitive ranking of “acts of service,” and why breakfast in bed is the worst
  • Fear and loathing in the south
  • How I’ll love you (both lightly and forever)
  • The stages of falling asleep with you
  • You’re never going to meet The One on Tinder

SEE THE DIFFERENCE??

Here, I’ll help…

  1. The top posts deal with Big Topics people actually care about. What is love? Am I in love? Should I get married? Do I want kids? How do I choose a partner? (The worst titles address… well, we’re not even sure on some. But certainly not The Big Topics. There is far less implicit emotion.)
  2. The top posts suggest that I will share something new or provide an answer on this Big Topic. (The bottom posts… make no such promise.)

THE SUCCESS OF THESE POSTS HAD NOTHING TO DO WITH “TRICKS”

Feel free to reread that until it sinks in. (For proof, reread above.)

c.) The “Secret”

…to a good title is:

a.) The suggestion (promise) of…
b.) Something informative, entertaining, or new on
c.) a Big Topic for the reader.

(And the secret to a good post? Delivering on b while honoring c.)

The “Secret” is:

Human emotion and human language — the beauty of implicit feelings, needs, values, fears conveyed in the intricacies of our communication — not black and white rules…

“How to” or “numbers” or whatever else you wanna use is just the way in which you deliver on those two things above— human to human.

So stop trying to strip the emotion and humanness out of something to get to “the secrets” — esp. while actually still expecting readers to feel emotion on the other end!

Skeptical?

I know some people will go on swearing these things matter. (Some have even run their own stats.) That’s fine.

But I hope that I’ve convinced at least some of you: it doesn’t really.

Not like some people act like it does.

The only thing that matters is writing about a subject people care about, and writing a title that suggests you’re giving them something of value. That’s it.

(Well, and write about something that not just the reader but also you feel strongly about.)


“WAIT! WHAT ABOUT TIPS FOR THE REST OF A POST THO?!!!”

Wow. You sure are being a greedy little pumpkin, aren’t you?

We’ll see. Maybe a future piece.