📊

Title Split Test: Paid Spec

Enter the hub password to continue.

That is not right. Try again.

← Back to Hub

Title Split Test: The Paid Spec

Written 2026-07-26 · $150 to $300 · 5 to 7 days

Written 2026-07-26. Cheapest defensible way to find out whether the new search-intent titles actually pull readers better than the original editorial lines.

Total cost: $150 to $300. Runs 5 to 7 days.

What we're actually testing, and why not the obvious thing

The tempting test is "which of my five candidate titles wins." Five variants per article across three articles is fifteen ads, which needs roughly 4,500 clicks to separate cleanly. That's a $1,500 test to answer a question worth maybe $200.

Test the decision you actually made instead: new title versus original. One pairwise test per article, six ads total. It answers the only live question, which is whether retitling helped or hurt, and it does it at a tenth of the cost.

Article A (original) B (new, live now)
testosterone-women "My brain went from black and white to technicolor." Testosterone for Women: What It Actually Does to Your Brain
antidepressant-testosterone-anhedonia "The flatness your antidepressant can't reach." Low Testosterone or Depression? How to Tell the Difference
testosterone-ambition-manifesto "She's 57, on testosterone, and building companies with AI." Low Testosterone in Women Over 40: Drive, Ambition, and Energy

If the budget only stretches to one test, run the anhedonia pair. It's the validated whitespace gap and the biggest single difference in register between the two options.

Platform: Meta, not Google

Google Ads tests keywords, not headlines. You'd bid against Hims, Midi and every telehealth brand in one of the most expensive categories in search, and you still wouldn't learn which title reads better, because the headline isn't what the auction is about.

Meta puts three headlines in front of the same audience and reports which one got clicked. That's exactly the question.

Campaign structure

Campaign:  Title Test (objective: Traffic, optimize for Link Clicks)
  Ad set:  Women 40-60 US                    <- ONE ad set, all ads inside it
    Ad A1  original title, article 1
    Ad B1  new title,      article 1
    Ad A2  original title, article 2
    Ad B2  new title,      article 2
    Ad A3  original title, article 3
    Ad B3  new title,      article 3

One ad set, not one per ad. Ads inside a single ad set compete for the same audience, which is what makes the comparison fair.

Settings that decide whether this works

  • Objective: Traffic, optimizing for link clicks. Not conversions, not landing page views. You will never get enough signups for statistical significance, so measure the click.
  • Turn OFF Advantage+ campaign budget. On by default. It hands nearly all spend to whichever ad wins the first few hours, so you end up measuring Meta's early guess rather than your readers' preference.
  • Turn OFF Advantage+ audience and dynamic creative.
  • Set spend per ad, not per campaign, so each headline gets an equal shot.
  • Placements: Facebook and Instagram feeds only. Turn off Stories, Reels, Audience Network and Marketplace. Feed is where a headline is actually read.

Audience

  • Women, 40 to 60, United States
  • No interest targeting. Counterintuitive, but broad performs better than hand-picked interests now, it's cheaper, and interest stacks quietly bias which headline wins.
  • Exclude anyone who already visited the site, once a pixel exists.

Budget

$25 per ad per day for 6 days = $150 per pair, or $300 for all three pairs. Target roughly 300 to 500 clicks per variant.

The creative

Image and body copy stay identical within each pair. The headline is the only variable. If you change two things you learn nothing.

Use the wide hiking hero for all six. It's already sized right for the feed.

Pair 1: testosterone-women

Primary text (both ads):

Most women are never offered a testosterone test, and the ones who ask are usually told their level is normal. I'm a medical technologist. I went and read what the research actually says, including the part where 142 laboratories measured the same blood and disagreed wildly.

Headline A: "My brain went from black and white to technicolor." Headline B: Testosterone for Women: What It Actually Does to Your Brain Description: Written by a medical technologist. Every claim cited. CTA button: Learn more Link: https://testosteroneinwomen.com/articles/testosterone-women/

Pair 2: anhedonia (run this one first if you only run one)

Primary text (both ads):

I was on an antidepressant for thirty years. It worked, in the sense that it did the job serotonin drugs do. What it never touched was the flatness. Here's the difference between depression and something else that looks exactly like it.

Headline A: "The flatness your antidepressant can't reach." Headline B: Low Testosterone or Depression? How to Tell the Difference Description: A medical technologist on what the trials actually measured. CTA button: Learn more Link: https://testosteroneinwomen.com/articles/antidepressant-testosterone-anhedonia/

Pair 3: drive and ambition

Primary text (both ads):

By around 40, a woman's testosterone is roughly half what it was in her twenties. That's the hormone that quietly runs drive. Nobody mentions it, because there's no FDA-approved product for women and therefore no marketing budget behind the answer.

Headline A: "She's 57, on testosterone, and building companies with AI." Headline B: Low Testosterone in Women Over 40: Drive, Ambition, and Energy Description: Ten approved testosterone products. None of them for women. CTA button: Learn more Link: https://testosteroneinwomen.com/articles/testosterone-ambition-manifesto/

Reading the result

Compare link click-through rate, not clicks, not cost per click, not impressions.

  • Under a 20% relative gap: call it a tie. At 300-500 clicks per variant that's inside the noise. A tie is a real result: it means keep the new titles, because they also help in search, and the ad test gave you no reason not to.
  • 20% or more: believe it, and consider it for social and email subject lines.

The trap to avoid

An ad test measures scroll-stopping power in a feed. Search ranking rewards matching what somebody typed into a box. These are different jobs, and the quote-style originals may well win the ad test while still being the wrong choice for the title tag.

If the results conflict, don't split the difference. Use both: - title tag and H1 → the search-intent version (what's live now) - ad headlines, email subject lines, social posts → whichever wins here

That's not a compromise, it's correct. The two placements are read by people in completely different states of mind.

Before spending anything

  1. Install the Meta pixel on testosteroneinwomen.com. Without it you can't exclude repeat visitors or retarget later, and the data is gone forever once the campaign ends. Ten minutes of work.
  2. Add UTM parameters to all six links so the result is visible outside Meta's own reporting, which grades its own homework: ?utm_source=meta&utm_medium=cpc&utm_campaign=title-test&utm_content=a1
  3. Consider whether ad spend is worth it at all right now versus waiting until the email capture is armed. Right now a click costs money and produces a reader who leaves. With the six-part series live, the same click can produce a subscriber. My honest recommendation: arm the email sequence first, then run this.

Written 2026-07-26 · annettethompson.com/hub