A/B testing

Find the pin style your readers actually click

Every Pinterest account you connect runs its own design test. Bloomtack tracks outbound clicks for each design, sends more new pins to the ones that work, and keeps testing challengers against the leader.

What gets tested, and how it’s measured

A design is a set of choices: the template, the font pair, the palette, the headline length and case, the overlay, the image type, where the logo sits, and whether the pin is an image or a video. Each combination is one variant in your test.

The score is the outbound click rate: clicks through to your blog divided by impressions. When an account gets almost no clicks yet, the save rate takes over, and it breaks ties otherwise.

Only pins between 7 and 60 days old count, because Pinterest spreads pins slowly. The 60-day window also lets the test follow seasons in your niche.

How the test runs over time

The method is a multi-armed bandit (Thompson sampling), which moves pins toward what works while it’s still learning, instead of waiting for a test to end.

  1. 1

    Warm-up

    A new account starts with 8 designs: 6 ranked for its niche and 2 built on its brand kit, including at least one video style. They take turns until each has about 1,000 impressions, and the app shows “Learning” with the progress.

  2. 2

    Learn while posting

    New pins favor the designs that are doing better, but at least 20% still go to designs that aren’t leading, and every post gets pins in at least two designs, so designs are compared on the same topic.

  3. 3

    Challengers and a winner

    Every 14 days, two challengers join, each changing one thing about the leader. Designs with under 5% chance of being best after 2,000 impressions retire. A winner is declared at a 90% chance of being best.

The winner report

The report names the winning design and how it compares to your account’s average, for example “Kinetic title video: +22% outbound click rate” (an illustrative example). Because challengers change one thing at a time, you can see which change made the difference: a darker title band, a shorter headline, a video instead of a still.

A weekly email sums it up. Accounts that don’t have enough data yet show “Needs more impressions” instead of a winner that isn’t real.

Your data stays yours

The test runs on each Pinterest account separately. Results are never pooled across customers, and customer data never feeds the starter templates, which come from the founder’s own test accounts.

Analytics are synced once a day, only for the pins Bloomtack created, and Pinterest data is never used to train AI. Read more in the privacy policy.

A/B testing questions

How long does it take to find a winner?

It depends on how many impressions your pins get. Each design needs about 1,000 impressions during warm-up, and pins only count once they’re 7 days old, so busier accounts see a winner sooner. Until then, the app shows the progress instead of guessing.

Do I still approve every pin during a test?

Yes. The test only decides which designs new pins use. Every pin still waits for your approval in the review queue, and you can edit or reject any of them.

Why outbound clicks and not saves?

Outbound clicks are readers arriving on your blog, which is what brings you traffic and income. Saves still count as the tie-breaker, and as the main score for accounts that don’t get clicks yet.

Get your blog on Pinterest without the busywork.

Join the waitlist for early access. One email when your spot opens, nothing else.

Opens your email app with a message to hello@bloomtack.com. Nothing is stored on this site.