Watering Hole Analysis

Watering Hole Analysis

The persona is defined. Now we find every place they live.

The goal is a comprehensive map of every place the target persona gathers, learns, consumes content, and gets influenced — online and offline. This means all channels: the ones the client is already using, the ones they know about but haven't activated, and the ones they haven't considered. Channels they're already using are not excluded — they belong on the list and will often rise to the top when scored. The research should leave no stone unturned.

Two outputs:

1. Narrative Watering Hole Analysis report — a written summary organized by channel category, with context, observations, and recommended focus areas
2. Channel Map spreadsheet — every channel extracted from the narrative, ready for scrubbing and scoring

Three-phase process:

1. Research — run the AI workflow, write the narrative, extract all channels
2. Scrub — remove channels that are clearly not candidates before scoring
3. Score — Quality and Cost fields completed with the client after the scrub

Tone: Researcher and strategist. Comprehensive first, then opinionated. Cast the widest possible net — the scrub and scoring process handles prioritization.

Inputs Required

Always ask for inputs before running the research — even if some were provided in the trigger prompt. Confirm all four fields are in hand before proceeding. Do not start the research workflow until they are.

Present this intake request clearly:

Before I run the Watering Hole Analysis, I need a few details:

1. Persona / Target Customer — Who are we trying to reach? (job title, role, or description — e.g., "VP of Operations at a mid-size manufacturer")
2. Product / Service — What are we marketing? (e.g., "equipment financing")
3. Location — What geography? (e.g., "Utah", "Pacific Northwest", "nationwide")
4. Additional Context (optional) — Industry, company size, B2B vs B2C, or anything else that would help target the research

If a Buyer Persona has already been developed earlier in this conversation or in project context, pull the relevant details from it — especially the job title, industry, and Priority Initiative — and confirm them with the user before proceeding rather than asking from scratch.

The Three-Prompt Research Workflow

This skill uses a two-model cross-interrogation approach that consistently produces more comprehensive and reliable results than a single-model run.

Prompt 1 — Primary Research (run in both Claude and ChatGPT)

Run this prompt yourself first. Then instruct the client or Fractional CMO to run the identical prompt in ChatGPT and share the response.

Act as a world-class marketing researcher, demand-gen expert, and customer finder.
Research this topic deeply and take your time.

In [LOCATION], where do [TARGET CUSTOMERS] learn about their options for
[PRODUCT/SERVICE]?

Please give me as comprehensive a list as possible. I'm looking for any:

ads, events, community groups, directories, stores, showrooms, trade shows,
conferences, associations, newsletters, podcasts, webinars, YouTube channels,
blogs, Discords, subreddits, Facebook Groups, LinkedIn Groups, websites, TikToks,
etc.

...where a [TARGET CUSTOMER] in [LOCATION] would go to get information about
[PRODUCT/SERVICE].

Prompt 2 — Interrogation (apply to each model's response independently)

After receiving each response, run this follow-up against it:

Interrogate this response. Is it reliable? Are the keywords you suggest really
attached to search volume? Is this as deeply researched as it needs to be?
What's missing? What should be added or removed?

Prompt 3 — Synthesis (run in Claude, feeding in ChatGPT's answer)

Once both models have been interrogated, run this in Claude:

Here is ChatGPT's answer to the same question. Review and compare it with yours
and give me the best final version you possibly can. Eliminate duplicates, resolve
conflicts in favor of the more specific or better-supported entry, and add anything
that appears in one list but not the other that deserves inclusion.

Output of the workflow: A single comprehensive, validated channel list organized by category. This is the source material for the narrative report and Channel Map.

Writing the Narrative Report

Once the synthesis is complete, organize the findings into a written Watering Hole Analysis report. Structure it as follows:

Report Structure

Opening paragraph: Frame the persona and the research objective. One short paragraph — who we're trying to reach and what we set out to find.

Section per channel category — cover each category that produced meaningful results. Suggested categories (use only those with actual findings):

• Associations & Trade Groups
• Conferences & Events
• Publications, Blogs & Trade Media
• Newsletters & Email
• Podcasts
• YouTube & Video
• LinkedIn — Groups, Pages, Influencers
• Facebook Groups
• Reddit & Online Forums
• Discord & Slack Communities
• Search & SEO (keyword themes, review sites, directories)
• Influencers & Individual Voices
• Local / Regional Channels (if location-specific)

For each section:

• Name specific channels found, with URLs where available
• Note audience size, activity level, or quality signals where known
• Flag any channels with sponsorship or advertising opportunities
• Note channels the client may already be using (to be confirmed in session)

Closing section — Observations & Recommended Focus:

• 3–5 highest-opportunity channels based on persona fit and accessibility
• Channels the client is already active in — note these explicitly, as they are strong candidates for increased investment
• Gaps — channel categories where the persona is present but the client has no footprint
• Channels worth monitoring but not activating immediately

Length: Aim for substance over brevity. A thorough Watering Hole Analysis for a well-defined persona in a specific geography typically runs 1,000–2,000 words. Sparse reports underserve the client.

Extracting the Channel Map

After the narrative is complete, extract every channel mentioned into the Channel Map spreadsheet. Each row = one channel.

Do not pre-populate Quality or Cost. These fields are intentionally left blank at this stage. Pre-populating them before the scrub creates anchoring bias and clutters the scoring conversation. Leave columns E, F, and G empty.

Column Definitions

Standard Channel Types

Association, Blog, Conference/Event, Discord, Facebook Group, Forum, Guru/Influencer, LinkedIn Group, Newsletter, Online Course, Podcast, Reddit/Subreddit, Search/SEO, Social — Instagram, Social — LinkedIn, Social — TikTok, Social — YouTube, Social — X (Twitter), Substack, Trade Publication, Webinar, Website/Directory

OEP Assignment Guidelines

• Owned: Client controls it — their website, email list, social profiles, content
• Earned: Built through relationships and reputation — referrals, press, organic community participation, speaking
• Paid: Requires spend to activate — ads, sponsorships, event booths, paid placements

Cost T-Shirt Sizing Guide

• S (Small): Low or no cost — organic posting, free community membership, basic listing. Under ~$500/month to activate.
• M (Medium): Moderate investment — newsletter sponsorship, event attendance, modest ad spend. ~$500–$2,500/month.
• L (Large): Significant investment — conference sponsorship, major publication placement, substantial ad campaign. ~$2,500–$10,000+.
• XL (Extra-Large): Major commitment — headline conference sponsorship, national media, sustained paid campaign at scale. $10,000+/month or major one-time spend.

Weighted Score Logic

Higher score = better channel given quality and cost together.

Weighted Score = (Quality × 200) + Cost Offset
Cost Offsets: S=150, M=100, L=50, XL=0

Examples:
Quality 5, Cost S → (5×200) + 150 = 1,150  ← Top tier
Quality 5, Cost XL → (5×200) + 0   = 1,000  ← High quality, expensive
Quality 2, Cost S → (2×200) + 150  =   550  ← Low quality even if cheap
Quality 1, Cost XL → (1×200) + 0   =   200  ← Discard

Sort the Channel Map by Weighted Score descending before the prioritization session. The top of the list becomes the conversation; the bottom becomes the backlog.

In-Session Use

The Watering Hole Analysis runs in three phases. Research happens before the session; scrubbing and scoring happen in the session.

Phase 1 — Narrative Walkthrough (15–20 min, pre-session delivery) Walk through the report category by category. The client reads or reviews it before the scoring session. This primes their thinking and surfaces any channels they already know about or are already using.

Phase 2 — Scrub (15–20 min) Open the Channel Map spreadsheet. Go row by row. Remove any channel that is obviously not a candidate — wrong audience, irrelevant geography, defunct, or otherwise not worth considering. Don't debate borderline entries; if there's any plausible case for it, keep it. The scoring step will handle prioritization.

This step is fast and binary: keep or cut. No scoring yet.

Phase 3 — Score (20–30 min) With the scrubbed list in front of you, work through the remaining channels together with the client — or assign the client to complete it independently before the next session. For each channel:

• Quality (1–5): "Is our persona actually here and reachable?" 5 = perfect fit, 1 = poor fit. Client judgment drives this — they know their buyer.
• Cost (S/M/L/XL): Estimate the investment to activate and sustain this channel. See the Cost T-Shirt Sizing Guide below. Either party can fill this in.
• Weighted Score: Calculates automatically once Quality and Cost are entered.

Sort by Weighted Score descending when complete. The top 3–5 channels surface naturally and receive budget allocation from the Zero Square marketing budget number.

Facilitator Notes

• The goal is a comprehensive list, not a discovery list. Channels the client already uses belong on the list. They often score highest because fit is already proven. Don't omit them because they're known.
• Don't pre-populate Quality or Cost. Pre-scoring before the scrub creates anchoring bias and wastes time on channels that will be cut anyway. The spreadsheet ships with those columns blank.
• The scrub comes before the score — always. Scoring a list that still has obvious rejects slows the conversation and muddles the output. Cut first, score second.
• The narrative report is a standalone deliverable. It's not a working document — it ships as a final output and the client keeps it as a reference.
• Don't pre-filter the channel list. Include channels you're skeptical about. The scrub step is where those get removed — but that's the client's call, not yours.
• Location matters. A national association matters differently than a regional chapter with active events the client could actually attend or sponsor.
• The research is only as good as the persona input. A vague persona produces a vague channel list. If the persona is thin, push back before running the research.
• Channels the client already uses are not automatically good channels. They survive the scrub and get scored like everything else. History of use ≠ quality fit — but they deserve a fair score, not a discount.