Using the Custom Researcher to Elevate your GTM Motion

Overview

Autobound’s Custom Researcher is your own programmable AI assistant that runs inside campaign tables. It can read and process data from your columns, reason across Autobound’s proprietary signal library, and pull verified insights from the wider web.
​

Think of it as an LLM you can embed in your workflow — one that can research, analyze, score, validate, or summarize data dynamically for every contact or company in your table.

Common use cases include:

• Prospect and company research for hyper-personalization

• Lead and account scoring based on firmographics, intent, and signals

• Competitor identification and market mapping

• Compliance and tone checks on outbound content

• Data validation and summarization across multiple sources

• Trend detection or custom enrichment tasks
​

The Custom Researcher combines your structured inputs with Autobound’s proprietary data sources — including LinkedIn, G2, Reddit, Twitter, earnings calls, and investor reports — to uncover information that standard enrichment tools can’t.

Check out this video on how to leverage the Custom Researcher Demo. Read below for more details:

Step 1: Create a Custom Researcher in Your Campaign Table

Once created, the Custom Researcher becomes a reusable column that can power research, analysis, and AI workflows across your table.

Step 2: Define What You Want the Researcher to Do
​

After mapping your inputs, you can either:

• Write your own Custom Research Prompt, or

• Choose from pre-built templates such as “Company Overview” or “Competitor Research.”
​

Each prompt defines the Researcher’s role — whether it’s gathering external data, scoring leads, analyzing messages, or identifying patterns.
​

Be descriptive and specific. For example:
​

• “Analyze {{prospectCompanyName}}’s competitors using LinkedIn, G2, and recent press.”

• “Score this contact based on company size, job title, and outbound signal activity.”

• “Check if this email contains any spam-trigger or non-compliant phrases. Here’s a list of common phrases and words.”
​

For a detailed research use case, scroll to the Templates & Examples section at the end of this article for pre-built frameworks that you can adapt to your own product or workflow.

Step 3: Referencing Data from Your Table
​

When building your prompt, you can dynamically reference any column in your table to give the Researcher more context.
​

To reference a column:

For example, if you want to reference the company name stored in the “Company” column, type “/” and select that column. Autobound will automatically insert it as {{prospectCompanyName}} or a similar variable.
​

This ensures each Researcher output is personalized to the exact data in that row.

Step 4: How It Works Behind the Scenes
​

The Custom Researcher uses Autobound’s Relevance Model to combine public web data with Autobound’s proprietary signals, surfacing richer, more contextually relevant insights.
​

It searches across:

• LinkedIn posts and company updates

• G2 reviews and product comparisons

• Reddit and X (Twitter) threads

• Job listings, investor reports, and transcripts
​

Each result is scored and summarized based on relevance, giving you verifiable, structured outputs for personalization, scoring, or analysis.

Step 5: Review and Use Your Researcher Results
​

Once the Researcher finishes running for a given row, click View Result in the research cell for that row.
​

You’ll see fields such as Final Answer, which contains the summarized output. Click Add as Column to save that data in your table, where you can easily view it by clicking into that cell.
​

If you want the research to influence your content generation, make sure the data is referenced in one of the cells mapped to Content Inputs.
​

By default, when you create a campaign from Autobound’s starter templates (such as “Multi-Step Cold Email Sequence”), the Additional Context column will already be mapped to your content generation column under the “Additional Context” content input.
​

Note:If you created a content column manually, you’ll need to ensure the Additional Context content input is properly mapped to a column so the context can be passed in.
​

In your Additional Context column, add a formula that explains how you want the content generation column to use the research. Here’s an example you can paste into the formula field to apply to every row:
​

“Use the following information to personalize your outreach messages only if it’s relevant and adds value. If the detail doesn’t fit naturally, leave it out.

When referencing research or insights, always:
​

Explain where you found it in a human-readable way (e.g., “I saw on your company’s LinkedIn page that…”). Tie it directly to your reason for reaching out — it should reinforce the “why you” and “why now” story.
​

Each reference should feel intentional and connected to the prospect’s current priorities, changes, or challenges — not just a mention for the sake of personalization.
​

Research: [Insert your research variable here using the “/” key on your keyboard]”
​

If you’re using the Researcher for other purposes like scoring, analysis, or compliance checks, you can map it to its own dedicated column instead. There’s no need to pass it to Additional Context unless you want to use the insights for personalization.

Step 6: Expanding Your Use Cases

The Custom Researcher is designed for flexibility. You can use it to automate a variety of workflows, including:

• Lead Scoring: Build scoring models that analyze company, contact, and signal data to determine outreach priority.

• Competitor Identification: Detect mentions of competitor technologies or related vendors in job posts, reviews, or company updates.

• Safety & Compliance Checks: Analyze message content for spam, legal, or tone issues before sending.

• Data Analysis & Summarization: Aggregate insights or trends across accounts to identify buying patterns or opportunities.

• Trend Detection: Combine your CRM data with Autobound’s signals to surface new opportunities based on emerging activity.

Think of the Custom Researcher as a customizable AI engine that can analyze any column, reason across your data, and enrich your campaigns with intelligence that updates as your tables do.

Step 7: Templates & Examples

Below are templates to help you build effective research prompts and use them for outbound personalization or scoring.

Paste the following prompt into ChatGPT (or something similar) to create your detailed custom research prompt. Make sure to fill in the info about your company:

"Create a custom research prompt I can use at my company: ENTER YOUR COMPANY WEBSITE

PURPOSE

This framework teaches the LLM (or teammate) how to create a deep research prompt customized to a specific company, so it can later surface relevant, evidence-based insights for personalized outreach.

The output of this process is a research prompt, not the research itself.

STEP 1 — DEFINE CONTEXT & OBJECTIVE

Identify the company you’re building for (e.g., “J2 Interactive,” “Informa TechTarget,” “Autobound”).

Define the core research goal — what type of insight will best equip a seller to reach out now?

Examples: uncovering friction in GTM operations, marketing efficiency, data quality, compliance, or pipeline acceleration.

Write a single sentence that clarifies why the research is needed (e.g., “to generate three insights that explain why now is the right time to engage {{contactFirstName}} at {{contactCompanyName}}”).

Explain how the research connects to the company’s solution (“the research should surface problems that {{VendorName}} directly helps solve”).

STEP 2 — DESCRIBE THE COMPANY’S SOLUTION SPACE

Summarize what the vendor does in one short paragraph (from their site or deck).

Translate that into 3–5 pain-to-value statements.

Example: fragmented intent data → unified first-party insights; wasted spend → verified buyer engagement.

Add one line defining what qualifies as a ‘relevant insight’ for this company (e.g., “anything showing a need for better audience targeting, lead quality, or pipeline attribution”).

STEP 3 — MAP THE TARGET PERSONAS

List the personas the company sells to and the outcomes they care about.

Include 2–3 examples like this:

CMOs – want measurable pipeline contribution

RevOps – need clean data and attribution

SDR Leaders – want faster lead activation

Tell the model to prioritize insights most relevant to these roles.

STEP 4 — DEFINE RESEARCH LEVELS

Create guidance for three research tiers:

Prospect-level: find individual activity such as posts, interviews, promotions, or role changes revealing focus areas or pain points.

Company-level: surface organizational changes — leadership hires, product launches, GTM shifts, or tech-stack updates — that indicate friction or transformation.

Industry-level: include macro trends, competitor moves, or regulatory shifts that add urgency or context.

Always require recent (≤ 90 days) and verifiable sources with links.

STEP 5 — BUILD SCORING LOGIC

Define how to score each insight for relevance to the vendor’s offering:

10 = Directly mentions the core pain the solution fixes

9 = Leadership change or investment in that domain

8 = New project or partnership overlapping the solution area

7 = General alignment (growth, modernization, efficiency)

Anything ≤ 6 should be excluded.

STEP 6 — DEFINE THE OUTPUT STRUCTURE

Tell the model to present the research in a clean, structured format that includes:

A numbered ranking of each insight by strength (1 = most relevant).

A short headline or title summarizing the problem or opportunity, followed by a relevance score (e.g., “(Relevance: 9/10)”).

A Fact sentence describing what happened, where, and when — including the company or individual involved.

A Why This Matters sentence explaining the friction or opportunity this creates relative to the vendor’s solution space.

A How {{VendorName}} Helps sentence connecting the issue back to one or two capabilities or outcomes the company provides.

A Source line with the exact URL.

Limit the total output to ≤ 250 words, ≤ 75 words per insight, and no more than three insights.

STEP 7 — ADD QUALITY & SAFETY GUARDS

✅ Only use verifiable public sources (LinkedIn, press, filings, etc.)

✅ No speculation or opinion

✅ Prefer content published within 90 days

✅ Rank insights by strength > quantity

✅ If no strong data exists, output fewer insights and explain why

STEP 8 — WRITE THE FINAL PROMPT SHELL

Combine everything above into a final, ready-to-use prompt with these variables:

{{VendorName}}

{{TargetPersona}}

{{PrimaryPainAreas}}

{{CoreOutcomes}}

{{ProspectFirstName}} / {{ProspectCompany}} / {{CompanyURL}}"

​

Based on the above prompt template, this is what we created at Autobound. Feel free to pass this into your chat as well so it can understand what you’re looking for and how to create it for your own company:
​

"Generate the top 3 most compelling, evidence-based reasons for a sales rep at Autobound to reach out to the contact: {{contactFirstName}} at their company: {{contactCompanyURL}}, RIGHT NOW.

Each insight must uncover a specific trigger, friction, or opportunity indicating that the company could benefit from signal-based personalization or outbound workflow automation.

Rank insights by relevance and strength of evidence.

Score each insight’s relevance to Autobound’s platform from 1–10.

CORE PRINCIPLE

Find proof of outbound inefficiency or missed personalization opportunity — not generic sales or marketing activity.

Look for evidence that the company or contact is:

• Struggling to scale outbound or inbound follow-up effectively.

• Manually researching prospects or building sequences in tools like Outreach, Salesloft, or HubSpot.

• Expanding SDR or marketing teams (growth = automation need).

• Running fragmented workflows between CRM, enrichment, and sequencing tools.

• Launching AI or personalization initiatives but lacking orchestration.

• Evaluating or referencing tools like Clay, Demandbase, 6sense, Apollo, or Lavender.

• Discussing conversion rates, pipeline efficiency, or signal-based selling.

If no meaningful or recent signals exist, don’t speculate — return fewer insights and explain why.

WHAT AUTOBOUND SOLVES (GUIDE YOUR RELEVANCE)

Autobound automates hyper-personalized outbound by combining:

• Signal Engine: Detects the right time to engage (job changes, social activity, funding, etc.).

• Insight Engine: Surfaces context across 350+ data categories (LinkedIn, 10-K filings, news, etc.).

• Persona Messaging Engine: Maps pain points, use cases, and value props to each persona.

• Content Engine (AI Studio): Generates multi-channel messaging (email, LinkedIn, call scripts) and routes it to sending systems.

Common pains solved:

• Manual research slowing SDRs and AEs.

• Inconsistent or low-quality personalization.

• Missed buying signals due to data silos.

• Low reply or conversion rates from generic outreach.

• Disconnected tech stacks that require heavy setup.

RESEARCH BREAKDOWN

RELEVANCE SCORING (1–10)

10 – Company explicitly mentions outbound automation, AI personalization, or signal-based GTM.

9 – Rapid SDR/marketing team growth or new market entry requiring scale.

8 – Hiring or posts related to personalization, sequencing, or messaging challenges.

7 – Indirect signal of sales inefficiency, data fragmentation, or low reply rates.

(Include only ≥7. Provide fewer if fewer qualify.)

PRIORITY RESEARCH SOURCES (prefer last 90 days

Prospect-Level Sources

• LinkedIn posts, comments, and articles.

• YouTube or podcast appearances (sales, AI, GTM themes).

• Speaking panels or webinar features.
​

Company-Level Sources

• Company LinkedIn page, press releases, and blog.

• Job postings (SDR, AE, RevOps, Growth, Demand Gen).

• Glassdoor, G2, or Reddit threads referencing outbound process.

• Funding or product launch coverage (TechCrunch, Crunchbase).

• Mentions of Salesloft, Outreach, Clay, HubSpot, or AI tools.
​

Market-Level Sources

• Industry reports on AI in sales (Forrester, Gartner, GTM Partners).

• LinkedIn or X discussions around sales automation or signal-based outreach.

• Press or research about outbound reply rate trends or pipeline efficiency.

OUTPUT FORMAT (≤250 words total)

Provide up to 3 insights, ranked by strength.

[#] – [Specific Signal or Friction] (Relevance: X/10)

Fact: [Describe what happened, where, and when, with source link.]

Why This Matters: [Explain how this limits outbound performance or personalization.]

How Autobound Helps: [Tie directly to Signal → Insight → Persona → Content orchestration.]

Source: [Insert direct URL]

QUALITY CHECKLIST

✅ Recent (≤90 days)

✅ Verifiable source (with URL)

✅ Reveals outbound, personalization, or workflow friction

✅ Connects directly to Autobound’s automation and orchestration capabilities

✅ Includes both prospect- and company-level signals

✅ Relevance ≥7

✅ Each insight ≤75 words"

Here's the template you can pass into the Additional Context formula:
​

Additional Context Template

"Once your Custom Researcher is producing results, use this guidance in your Additional Context column to help Autobound’s content engine personalize messages naturally.
​

Use the following information to personalize outreach only if it’s relevant and adds value. If the detail doesn’t fit naturally, leave it out.
​

When referencing research or insights, always:

• Explain where you found it in a human-readable way (for example, “I saw on your company’s LinkedIn page that…”).

• Tie it directly to your reason for reaching out — it should reinforce the “why you” and “why now” story.

• Ensure each reference feels intentional and connected to the prospect’s current priorities, changes, or challenges.
​

Research: [Insert your research variable here using the “/” key to reference your Researcher column]"

Summary

The Custom Researcher is Autobound’s most adaptable AI tool — capable of connecting your first-party data, company context, and 350+ proprietary signals to uncover insights, generate scores, validate content, and power hyper-personalized outreach.

To get the most out of it:

• Type “/” in your prompt to reference data from your table.

• Save and reuse templates for repeatable workflows.

• Map your research output to Additional Context for personalization.

• Or connect it to dedicated columns for scoring, compliance, or analysis.