Playbook
Every play Tailor draws on when it proposes a test, and the controls to steer which ones it uses.
Overview
Playbook is under Signals in the left sidebar. It is the library Tailor proposes from, laid out so you can see it rather than infer it from the tests that show up in your queue.
Plays are grouped by what they change (CTA, form, navigation, and so on) and sorted with the strongest evidence first. You can search them, switch between a table and cards, and filter by business model and by the part of the page they touch.
This page exists because "the AI suggested it" is not a reason a marketer can defend in a review. Being able to point at the play, its mechanism, and its evidence tier is.
Reading A Play
Each play carries the same five attributes, which together tell you whether it is worth your traffic.
| Attribute | What it tells you |
|---|---|
| Change type | What the play physically does: copy, element addition, element removal, flow simplification, offer and framing. |
| Works by | The mechanism it relies on: removing friction, directing attention, improving relevance, or adding motivation. |
| Evidence | How well established the play is. See below. |
| Effort | Low, medium, or high, meaning what it takes to build and approve rather than to run. |
| Fits | Which business models the play is written for. |
The pairing worth using is evidence against effort. A proven, low-effort play is where to start on a page nobody has tested; a promising, high-effort play is a bet you should only take on a page with the traffic to read it.
Evidence Tiers
Proven
Repeatedly demonstrated. Safe to run without a strong prior belief, and the right default for a page with no test history.
Strong
Well supported, though more sensitive to context than a proven play. Worth running where the mechanism plausibly applies to your page.
Promising
Real but less established. Treat these as genuine experiments rather than expected wins, and give them enough traffic to actually answer.
Some plays are also marked as rising, meaning they are gaining evidence recently rather than being long-settled. Those are the ones worth reading when you want an edge rather than a safe increment.
Star And Ignore
Two controls on every play, and they are the reason to spend time here at all.
- Star a play that fits your strategy, and Tailor leans toward proposing it.
- Ignore a play and Tailor stops proposing it entirely.
Ignore is the more valuable of the two. Every team has changes it will not ship for reasons no model can infer: brand rules, legal review, a checkout nobody is allowed to touch this quarter. Ignoring those once is faster than declining the same suggestion every month, and it stops the queue filling with tests that were never going to launch.
For steering the shape of the plan rather than individual plays, see Test Ideas, which takes a standing brief and can be scoped to specific domains.
Business Model
Plays are filtered by business model, so an ecommerce account sees cart and checkout plays that a B2B account does not. Alongside those sit the plays that apply to every business, and Tailor draws on both when it proposes tests regardless of which filter you are looking at.
So the filter changes what you are reading, not what Tailor is allowed to use. If you want to genuinely rule a play out, ignore it.