This is the first tutorial in the Getting Better Results series. You've set up your first Interest — now let's make it actually work.
Why Refining Your Interest Matters
Your Interest has two free-text fields that directly control how Engaggit's AI behaves:
- "What I'm looking for" — the AI reads this to decide whether a Reddit post is relevant to you
- "What to show in the draft" — the AI uses this to decide whether it can genuinely help and how to write a helpful reply
These aren't optional descriptions buried in settings. They're the core of how Engaggit filters the noise and generates drafts. The better you write them, the better everything downstream works:
Better descriptions → better relevance scores → better matches → more leads
The search rules (keywords, subreddits, must-have/must-not filters) cast a wide net. The descriptions tell the AI which catches are actually worth your time.
Anatomy of a Great "What I'm Looking For" Description
This field is the single most important part of your Interest. The AI uses it to evaluate every post it encounters. A strong description includes four elements:
| Element | What It Does | Example |
|---|---|---|
| The Problem | Describes the pain your audience feels | "People frustrated with complex project management tools" |
| The Search | Names what they're actively looking for | "Looking for alternatives to Asana or Monday.com" |
| The Context | Narrows the audience to your ideal buyer | "Small teams of 3–10 people" |
| The Triggers | Flags specific phrases or complaints | "Complaining about too many features, expensive pricing" |
You don't need all four in every Interest, but the more you include, the more precisely the AI can score relevance.
Before and After Examples
The difference between a vague description and a precise one is dramatic.
Example 1: Productivity App
Before (Poor):
"People who might like my productivity app"
After (Great):
"Small team leaders looking for a simpler alternative to Asana or Trello. They're frustrated with too many features, expensive per-user pricing, and complex onboarding. They're asking for 'lightweight project management' or 'simple task tracker' in their posts."
Example 2: CRM Product
Before (Poor):
"Anyone using a CRM"
After (Great):
"Sales teams and founders looking for CRM recommendations. They're comparing HubSpot, Salesforce, or Pipedrive. They need something affordable and easy to set up. They're asking about integration with Slack or Gmail."
Example 3: Developer Tool
Before (Poor):
"Developers who need tools"
After (Great):
"Frontend developers frustrated with slow build times in Vite or Webpack. They're asking for faster alternatives, comparing bundler performance, or complaining about cold start times. They care about DX and want hot reload that actually works."
The pattern: vague descriptions produce vague scores; specific descriptions produce matches you can act on.
Anatomy of a Great "What to Show in the Draft" Description
This field serves double duty. The AI uses it both to decide whether it can genuinely help the person who posted, and to write the actual draft reply. The more specific you are here, the less editing your drafts need.
| Element | What to Include | Example |
|---|---|---|
| What It Is | Your product or service in plain language | "A lightweight CRM built for small teams" |
| Who It's For | Your ideal user | "Teams of 2–15 people who use Slack daily" |
| Key Features | The top 2–3 capabilities | "Lead management, pipeline tracking, Slack integration" |
| Pricing | Cost or free tier | "Free for teams under 5, $15/seat for larger" |
| Differentiator | What makes you different | "You never leave Slack to update your pipeline" |
The AI doesn't just paste this into the draft — it uses it as context to write something relevant to the specific post. The more it knows about you, the better it connects your solution to the poster's exact problem.
How These Descriptions Reach the AI
Under the hood, Engaggit sends both fields to your AI provider during every scan. Here's what happens:
- The AI reads the Reddit post (title, body, and context)
- It reads your "What I'm looking for" description and evaluates relevance — scoring the post from 0 to 10
- If the post scores above your threshold, the AI reads your "What to show in the draft" description and generates a draft reply
- You review and edit the draft before posting
If the "What to show in the draft" field is empty, Engaggit still records the match — but no draft is generated. You'll see the post in your feed with a relevance score, but you'll need to write the reply yourself.
Tuning the Relevance Threshold
The relevance threshold is the minimum score a post needs to appear in your feed. You can adjust it in Settings → AI Settings.
| Goal | Recommended Threshold | What You Get |
|---|---|---|
| Quality over quantity | 7–8 | Only high-intent posts; fewer matches but nearly every one is worth acting on |
| Balanced approach | 5–6 | A good mix of high and medium intent; recommended starting point |
| Quantity over quality | 3–4 | More posts, more noise; useful when you're testing a new Interest |
The default threshold is 6. Here's how to tune it:
- Start at 6 — this is the sweet spot for most users
- Too many irrelevant posts? Raise it to 7 or 8. You'll see fewer matches, but the ones that appear will be stronger.
- Missing opportunities? Lower it to 4 or 5. You'll catch more edge cases, but you'll also filter through more noise.
- Monitor for a week, then adjust. Don't change it daily — give the scan enough data to show patterns.
A/B Testing Your Interest
If you're not sure which description works better, test two versions side by side:
- Create two Interests with the same search rules but different descriptions
- Name them clearly — "CRM Leads v1" and "CRM Leads v2"
- Run both for a week
- Compare which one produces higher-scoring matches
- Keep the winner, delete or refine the loser
This works because each Interest runs independently. The AI evaluates the same Reddit posts against different descriptions, so you can see which framing resonates better.
Note: Free tier limits you to 1 Interest, so A/B testing requires a PRO trial or subscription.
The Iteration Process
Writing a great Interest isn't a one-time task. The best results come from a loop:
Write Interest
│
▼
Run Scans for 1 Week
│
▼
Review Matches
│
▼
Identify Patterns
(What's working? What's not?)
│
▼
Refine Description
│
▼
Repeat
What to look for during review:
- High-scoring matches that are actually relevant — your description is working; keep it
- High-scoring matches that miss the mark — the AI is latching onto keywords but missing context; add more specifics to your description
- Low-scoring posts that should be high — the AI isn't picking up on something; add the missing trigger words or context to your description
- Lots of noise at the bottom — raise your threshold or tighten your search rules
Common Refinement Mistakes
| Mistake | Why It Hurts | Fix |
|---|---|---|
| Being too specific | You miss people who use different language | Broaden your description; use multiple phrasings |
| Being too broad | Too many irrelevant posts flood your feed | Add specific triggers, competitor names, or context |
| Not updating based on results | Your first draft is never your best | Review weekly and refine |
| Ignoring what the AI found | The AI's "reason" field tells you why it matched | Read the match reasons — they reveal what the AI latched onto |
| Leaving "What to show in the draft" empty | No drafts are generated for matching posts | Fill it in — even a short description produces better drafts than none |
Quick Reference: Field Checklist
Before you run your next scan, make sure your Interest has:
- [ ] A clear, specific "What I'm looking for" with problem, search terms, and context
- [ ] A detailed "What to show in the draft" with your key features, audience, and differentiator
- [ ] At least 3–5 high-intent search keywords
- [ ] A relevance threshold that matches your goals (start at 6)
- [ ] Must-have and must-not filters if you're getting noise
What's Next
Your Interests are dialed in. Now let's make sure you're using the right AI to power them.
Next tutorial → Choosing the Right AI Provider — comparing providers for cost, speed, and quality.
Frequently Asked Questions
1. Does the "What I'm looking for" description affect draft quality too?
Yes — it's used in two places. First, the AI uses it to score relevance (0–10). Second, it's passed to the draft generator as context for why the post was found. A vague description hurts both scoring and draft quality, so it's worth investing time here.
2. What happens if I leave "What to show in the draft" empty?
Engaggit will still score and filter posts using your "What I'm looking for" description — you'll see matches in your feed with relevance scores. But no drafts will be generated. You'll need to write replies manually for each post.
3. Can I have multiple Interests monitoring the same subreddit?
Yes — each Interest runs independently. Two Interests can target the same subreddit with different keywords, descriptions, and relevance thresholds. This is useful for A/B testing descriptions or monitoring different angles of the same market.
4. Why do some high-scoring posts still feel irrelevant?
The AI might be latching onto keyword matches without understanding the full context. Fix this by making your "What I'm looking for" description more specific — add the exact phrases you want to see, name competitor products, and describe the situation you're targeting, not just the topic.
5. How often should I update my Interest descriptions?
Review after the first week, then monthly. Early on, you'll catch obvious mismatches quickly. After that, descriptions generally hold up well — just watch for new competitor names, shifting language patterns, or changes to your own product.
6. Does the relevance threshold affect which posts get drafts?
Yes — posts below the threshold are saved without drafts. Only posts that score at or above your threshold trigger draft generation (assuming you've filled in the "What to show in the draft" field). Raising the threshold means fewer drafts, but each one is higher-quality.
Engaggit is a privacy-first, AI-powered Reddit lead generation app. It runs on your machine, keeps your data to yourself, and helps you never miss a relevant conversation again.