Your first growth hire is rarely a “run ads” specialist. At pre-PMF and early traction, they sit between product, content, and founder-led sales - reading what buyers actually say in public and turning that into experiments you can measure.
Signal literacy is the skill that separates a hire who compounds your GTM from one who burns community goodwill. This guide gives you a 90-minute take-home, live probes, and a rubric you can reuse.
What signal literacy actually means
A literate growth candidate can:
- Read a thread in context (who posted, why now, what they tried)
- Cluster pains without forcing everything into your positioning
- Propose experiments with falsifiable success metrics
- Name ethical risks before you approve outreach
- Reject spray tactics even when pressured for pipeline
They do not need to know your product cold on day one. They need to show judgment on messy public data.
The 90-minute take-home (template)
Send this verbatim; swap in your own URLs.
Task: You have 90 minutes.
Input: 10 public URLs (mix Reddit, HN, a forum thread, one YouTube comment section excerpt - include 2–3 low-signal noise threads on purpose).
Deliverable: One-page memo (max 600 words) with:
- Three pain themes with 2–3 supporting quotes each (paraphrase; link sources)
- Two experiments you’d run in the next 14 days (channel, copy angle, success metric, kill criterion)
- Ethical risks (spam, selection bias, vendor shilling) and how you’d mitigate
- One headline you’d A/B on a landing page - and what evidence supports it
Scoring tip: Weight reasoning over polish. A candidate who discards 4 of 10 URLs as irrelevant is often stronger than one who mines “intent” from every mention.
Rubric (1–5 per dimension)
| Dimension | 5 (strong) | 1 (pass) |
|---|---|---|
| Clustering | Themes emerge from quotes; outliers noted | Everything tagged “our ICP” |
| Experiment design | Metrics + kill criteria | “Post more on LinkedIn” |
| Ethics | Names community norms; when not to reply | “DM all keyword matches” |
| Bias awareness | Calls out survivorship / loud minority | Treats Reddit as census |
| Writing | Clear memo a founder could act on | Buzzwords, no quotes |
Hire bar: No dimension below 3; clustering and ethics both ≥ 4.
Live interview probes (15 minutes)
Use the memo as fuel - don’t re-ask what they wrote.
- “Pick your weakest theme. What would change your mind?”
- “When would you not reply in-thread?”
- “How do you detect selection bias in community research?”
- “Translate thread #7 into one homepage headline - defend it in 60 seconds.”
- “Your CEO wants 50 outbound DMs this week from Search results. What do you say?”
Green flags: They ask for ICP definition, cite subreddit rules, propose a smaller high-intent batch first.
Red flags: Guaranteed reply rates, scraping emails from profiles, “growth hack” language without metrics.
Role variants (adjust expectations)
| Profile | Emphasize in task |
|---|---|
| Content-heavy | Headline + theme clustering |
| Lifecycle / PLG | Activation metrics tied to pains |
| Light outbound | Ethical reply templates + CRM handoff |
You’re not hiring a community manager or an SDR alone - but the task reveals which muscle is strongest.
Sample thread pack (build your own)
Include at least:
- One “alternatives to X” thread (switching intent)
- One rant with no buying signal (noise discipline)
- One technical Stack Overflow or GitHub issue (implementation pain)
- One founder thread asking for advice (support vs sell judgment)
Pull URLs from your category with multi-platform search or manual browsing - do not use your stealth roadmap or unreleased features as the subject.
On-the-job tooling (optional)
Candidates may use multi-community search and monitors within plan limits on the job. That’s fine - you’re testing whether they interpret results, not whether they’ve memorized a UI.
Pair tooling with your founder outbound SOP so experiments stay compliant.
Related reading
- Find first 100 customers
- Customer discovery marketing early stage
- Founder outbound after community research
- Intent signals before Apollo outbound stack