A quick, important clarification before anything else: Ghost doesn't have a native, automated A/B testing feature, the kind where you write two subject lines, Ghost sends each to half your list, and automatically picks a winner before sending the rest. If you came here looking for that specific button, it doesn't exist in Ghost today.
What Ghost does have is something genuinely useful for the same underlying goal: rich post and newsletter analytics that let you test ideas deliberately over time and actually see what's working. It's a more manual process than an automated split test, but it's far from useless, and once you know what data you actually have access to, you can run a perfectly effective testing practice without needing a feature Ghost doesn't offer.
Here's what's real, what's not, and how to actually improve your titles and newsletter content using the tools genuinely available.
What Ghost's Post Analytics Actually Show You
Every post you publish, especially ones sent as a newsletter, comes with a real analytics page. From your Posts list, you can see at a glance how each post performed, and drilling into any individual post gives you considerably more:
- Unique visitors, the number of people who viewed the post on your actual site
- Opens, the number of unique members who opened the email version
- Open rate, opens divided by total recipients
- Clicks, the number of unique members who clicked any link inside the newsletter
- Click rate, clicks divided by recipients
- Link-level breakdown, showing exactly which specific links inside your newsletter got clicked, and by how many people
- Audience feedback, if you've enabled it, showing how many readers reacted "more like this" or "less like this" to a given piece
- Signups and conversions attributed to that specific post, showing how many free signups or paid conversions a particular piece of content actually generated
Crucially, Ghost doesn't just hand you raw numbers in isolation. It automatically compares each post's open and click rates against your typical averages, calculated from your last 20 newsletters sent. This context matters enormously: a 35% open rate might be a disappointment or a triumph depending entirely on what's normal for your specific list, and Ghost does this comparison for you rather than leaving you to track it manually in a spreadsheet.
A Fair Warning About What "Opens" Actually Means
Worth knowing before you lean too heavily on open rate as your primary success metric: privacy-focused email clients, Apple Mail being the largest example, now pre-fetch and effectively "open" emails automatically as a privacy measure, regardless of whether a human actually read anything. This means open rate, across the entire email industry, not just Ghost, has become a noisier signal than it used to be.
Click rate, and signups or conversions attributed to a specific post, tend to be more trustworthy indicators of genuine engagement, since those require an actual reader taking a deliberate action, not just a privacy proxy pinging your server in the background.
How to Actually "Test" Titles and Content Without a Built-In A/B Tool
Since Ghost won't split-test automatically for you, the practical approach is sequential testing: trying one deliberate approach, measuring it properly, then trying a genuinely different approach next time, and comparing the results against your own historical baseline rather than against a simultaneous variant.
1. Change one variable at a time, and actually note what you changed
If you write a punchy, curiosity-driven subject line this week and a completely different, descriptive, literal subject line next week, while also changing your send time and your opening paragraph structure between the two, you won't actually learn anything useful, you've changed too many variables to know which one moved the needle. Pick one element to deliberately vary at a time: the subject line style, the send day, the opening hook, the call-to-action placement, and keep a simple running note (even a basic spreadsheet) of what you changed and what happened.
2. Compare against your own baseline, not a single previous post
A single post performing better or worse than the one before it isn't necessarily meaningful, individual posts vary for all kinds of reasons unrelated to your title or content choices. This is exactly why Ghost's rolling 20-newsletter average is genuinely useful: judge a specific test against that baseline rather than against just the most recent send, which smooths out a lot of natural noise.
3. Use the link-level click data to learn what readers actually want
This is one of the most underused pieces of data available. If your newsletter includes multiple links, a few different topics, a couple of recommended reads, a call-to-action, the breakdown of which specific link got clicked tells you directly what your audience found compelling that particular week. Over several sends, patterns emerge: maybe practical how-to content consistently outperforms opinion pieces, or a particular kind of call-to-action consistently underperforms. That's genuinely actionable, even without ever running anything you'd formally call an "A/B test."
4. Pay attention to audience feedback if you've enabled it
The "more like this" and "less like this" reactions, available in Settings → Email Newsletters, are about as direct a signal as you'll get. It's not as granular as a full survey, but it's effectively a built-in, frictionless feedback mechanism that costs your readers one click and tells you plainly whether a specific piece landed.
5. Track signups and conversions by post, not just engagement
A post can have a mediocre open rate but a genuinely strong signup or paid-conversion rate, or the reverse. Since Ghost attributes signups and paid conversions to the specific post someone was reading when they converted, this is worth watching as its own metric, separate from opens and clicks, especially if growth or monetization is your primary goal rather than pure engagement.
What You Can Actually Test This Way
A reasonable, ongoing testing practice on Ghost, using what's genuinely available, might look like cycling through these one at a time over several sends:
- Subject line style: curiosity-driven versus direct and descriptive
- Subject line length: short and punchy versus longer and more specific
- Send day or time, comparing performance against your rolling average for that same day or time historically
- Opening paragraph structure: leading with a hook or anecdote versus leading directly with the main point
- Call-to-action placement and wording, tracked through your link-level click data specifically
- Content format, comparing how a listicle-style post performs against a single deep-dive narrative piece, using your conversion and click data as the real signal
If True Automated Split-Testing Genuinely Matters to You
If running real, simultaneous A/B splits, sending Subject Line A to half your list and Subject Line B to the other half, automatically, is a hard requirement rather than a nice-to-have, it's worth being honest that this isn't something Ghost's newsletter system does natively today. Dedicated email marketing platforms (Mailchimp, beehiiv, and others built specifically around campaign-style sending) do offer this as a built-in feature.
Some publishers handle this by using Ghost for everything else, writing, the site itself, membership and Stripe billing, while syncing their member list to a dedicated email tool specifically for the rare campaigns where formal split-testing genuinely matters, then syncing engagement data back. This is more setup than most people need, and for the overwhelming majority of newsletter writers, the sequential testing approach using Ghost's own analytics gets you almost all of the same practical benefit without an extra tool and monthly cost layered on top.
The Honest Takeaway
Ghost gives you real, useful data, opens, clicks, link-level detail, feedback, and conversion attribution, all benchmarked automatically against your own historical performance. What it doesn't give you is the specific mechanic of simultaneous, automated split-sending. For almost everyone writing a newsletter, deliberate, one-variable-at-a-time testing over a series of sends, paired with genuinely reading your own analytics rather than just glancing at the open rate, gets you to meaningfully better titles and content without needing a feature that isn't actually there.
