> ## Documentation Index
> Fetch the complete documentation index at: https://docs.superscale.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Research best practices

> Get sharper insights from competitor research, ad account data, and market signals.

Research is only useful if it changes what you create next. Do not use Superscale research as a screenshot gallery. Use it to find hypotheses: hooks, formats, offers, audiences, and objections worth testing.

## Start with the decision

Before running research, decide what you need to learn.

<CardGroup cols={2}>
  <Card title="Find a format">
    Ask which ad structures repeat across winners: testimonial, demo, comparison, founder story, problem-solution, offer-led, or UGC.
  </Card>

  <Card title="Find a message">
    Ask which pains, desires, objections, and proof points show up across competitors and your own reviews.
  </Card>

  <Card title="Find a benchmark">
    Ask what good looks like for your market: pacing, offer clarity, visual density, hook style, and production quality.
  </Card>

  <Card title="Find the next test">
    Ask which variable to test next, not just which ad looks good.
  </Card>
</CardGroup>

## Use three signal types

<Steps>
  <Step title="Competitor signal" icon="binoculars">
    What other brands are running now, which formats keep repeating, and which ads have stayed live long enough to be worth studying.
  </Step>

  <Step title="Customer-language signal" icon="message-circle">
    Reviews, Reddit, comments, support tickets, and sales calls. This is where hooks and objections usually come from.
  </Step>

  <Step title="Your account signal" icon="chart-no-axes-combined">
    Your own spend, CTR, CPA, ROAS, fatigue, audience, and creative history once your ad accounts are connected.
  </Step>
</Steps>

## Do not over-trust one source

Public ad-library data is directional. Your own ad account data is more specific but only reflects what you have tested. Reviews are emotionally rich but not always representative. The best research brief combines all three.

<Note>
  If the agent sounds too certain, ask it to separate facts, public signals, and assumptions. Good strategy is honest about confidence.
</Note>

## Better prompts

* "Find same-league competitors, not category giants. I care about brands with similar production quality and budget."
* "Analyze these ads for hook patterns, not visual style. I want the messaging logic."
* "Compare competitor winners with our own best ads. Which variable should we test next?"
* "Extract the exact customer phrases from reviews that could become hooks."

## Research output should become context

When research produces a durable learning, save it. Examples:

* Winning competitor formats.
* Objections that keep coming up.
* Claims that work or should be avoided.
* Benchmarks for creative performance.
* Audience-specific language.

<Card title="Competitor research" href="/features/competitor-research" horizontal>
  Learn how Superscale finds and interprets competitor ads.
</Card>