Skip to main contentTailor AI LogoTailor AI

    Test Ideas

    A standing, ranked list of AI-recommended experiments built from your ads, traffic, and pages. You review and launch; Tailor does the drafting.

    Overview

    Test Ideas reads your ad spend (Google, Meta, LinkedIn), your traffic, and your landing pages, and returns a ranked plan of A/B tests to lift conversion and ROAS. Each idea comes with the audience it targets, its expected 30-day reach, the key page changes, and a before/after preview built from your live site.

    The plan is standing: new runs append to it rather than replacing it, dismissed ideas stay dismissed (and can be restored in one click), and future runs learn from the results of tests you launch.

    Ranked by expected impact

    Ideas are ordered by expected monthly impact (traffic to the targeted segment times the targeted lift), not by confidence alone. A modest win on a high-traffic page usually beats a big win on a page nobody visits.

    Generate a Plan

    1

    Open Test Ideas

    Find it in the left sidebar at app.tailorhq.ai.

    2

    Pick your inputs

    Choose which ad platforms, data signals, and pages Tailor should consider, and whether you want the best tests across your pages or a page for every paid keyword. A Fine-tune section gives you finer data controls when you want them.

    3

    Set your standing brief (once)

    Tell Tailor who you are and what you optimize for. The brief steers every future run, and you can add a one-off focus line to any single run.

    4

    Let it build

    Plans take a few minutes. You can switch tabs; your browser notifies you when the plan is ready (enable notifications when prompted, or under Settings → Notifications).

    Larger plans arrive organized into launch waves: a first phase of tests you can run in parallel now, with later waves sequenced behind them. Toggle between the phased view and one flat ranked list.

    Keyword Coverage

    Choosing "a page for every paid keyword" runs a keyword coverage sweep: Tailor maps each paid-search keyword to the landing page it actually hits, flags weak message match, and turns the gaps into ready-to-launch test ideas with the copy changes drafted. A sweep covers hundreds of keywords per run.

    Keyword targeting needs the keyword in the URL

    Serving a tailored experience per keyword relies on your URLs carrying the keyword (usually utm_term). If your tracking templates don't pass it, fix that first. See the Targeting Guide.

    Reading an Idea

    The plan reads as a ranked table: value proposition, audience, estimated 30-day reach, and the key page changes. Click any row to open the full detail: targeting, before/after preview screenshots from your live site, and the launch, preview, refine, and dismiss actions.

    • 1Numbers checked badges mean the traffic and reach figures were verified against your real data.
    • 2Runtime estimates come from the targeted segment's actual traffic, so a niche audience shows an honest (longer) time to a result.
    • 3Targeting is editable: adjust the UTM or custom parameters an idea targets and the audience size re-measures automatically.

    Refine & Edit

    Refine one idea

    Tell Tailor what's off ("less formal", "lead with the integration") and it rewrites that test from your feedback.

    Refine all

    One instruction re-voices every headline, subhead, and CTA across the plan. Useful for brand-voice passes.

    You can also edit proposed copy inline before launching, rewrite a single change with AI, remove a change from the test, or jump into the extension to edit the page directly.

    Launch & Learn

    Launching an idea creates a real experiment on your existing URL with the targeting attached. Nothing goes live without your approval. Results feed back into the next round: runs read the outcomes of your live and rolled-out tests, iterate on champions, and stop proposing what you've dismissed.

    From there, the normal experiment workflow applies: ramp, read results, and promote winners. See the Experiments Workflow guide.

    Best Practices

    • 1Connect your ad accounts first so proposals reflect real spend, not guesses.
    • 2Write the standing brief before your first run. Two sentences about your audience and goal noticeably improve the ideas.
    • 3Dismiss freely. Dismissals teach future runs what not to propose, and you can restore anything in one click.
    • 4Launch the top wave in parallel rather than one test at a time. The waves are sequenced so they don't conflict.