Most email outreach data sits untouched - subject lines, sending lists, domains, and CTAs with learnings to be made, but no obvious place to start the review.
This Hunter playbook gives you that starting point. Three prompts, your Hunter Sequences data, and a GenAI tool of your choice (Claude, ChatGPT, or Gemini).
By the end, you’ll know which subject lines won, which lists decayed, which sending domains held up, and what to change next quarter.
What you'll do
Part 1: Pull Sequences data from Hunter
Part 2: Create a document with your sequence content
The answers could lie in the performance of your emails.
Fail to look at the data behind everything you're doing, and failing to give yourself the best opportunity to find issues, fix them, and learn as part of a wider mindset shift that sees outreach as an engine.
Set the timeframe to the period you want to analyze.
The page gives you four reporting views: Sequences, Team members, Email accounts, and Email domains.
Each tab shows the same metrics (Sent, Bounces, Opens, Clicks, Replies, and Successful) across the different elements that impact your outreach success. For each tab:
Click “Displayed columns” > select every metric, including Unsubscribes.
Scroll to the Details section > click “Export CSV”.
Step 2: Collect replies
Now, build a list of leads who replied, or whose response you marked as a success:
Click “Columns” and ensure “Full name”, “email address“, “Company“, “Job Title“, “Website“, “Verification status“, “Sending status“, “Industry“, and “Last contacted at” are all selected.
Apply the “Last contacted at” filter to the date range you want to analyze.
Apply another filter for “Sending status” and select “replied”.
Click the box next to “Full name” > “Select all” > … > Export.
Rename the file to replies.csv.
Part 2: Create a document with your sequence content
We need the LLM to find patterns in subject lines, opening lines, and CTAs.
Find your sequences.csv file, and have Sequences open in Hunter, ready to add information.
Step 1: Feed your Sequences data into the LLM
Open your LLM (we’ll use Claude) and upload your sequences.csv file, and enter this prompt:
prompt
You are an expert B2B email outreach analyst. I am a [founder / solo GTM operator] who ran email outreach using Hunter Sequences over the past [Time period]. Attached is a CSV export of every email I sent across every sequence.
## CSV structure
Each row is one (step, variant) combination of one sequence.
- Column A "Sequence": sequence name. Multiple rows share the same name when they belong to the same sequence.
- Column B "Launch date": when the sequence launched, not when the email was sent. Same across all rows of one sequence.
- Column C "Step": which step in the sequence (1, 2, 3...). If a sequence name only ever appears with Step 1, it is a one-shot, not a sequence.
- Column D "Variant": A or B if the step is an A/B test. Blank if not.
- Column E "Email": the subject line for that step/variant.
- Columns F to L: Sent, Bounces, Opens, Clicks, Replies, Successful, Unsubscribed. Counts per step/variant.
"Successful" is a manual tag applied in the Hunter inbox to mark replies that move a conversation forward: a call booked, a link placed, press coverage secured, or a meeting scheduled. Reply rate counts everything that came back. The successful rate counts only the wins.
## Before you start, ask me
1. Rank by reply rate or successful rate?
2. How many sequences should I see? Always ask. Wait for a number. Do not assume 10.
3. Minimum recipients per sequence to qualify? Default 25 if I don't specify.
4. Include all sequences, or only first-touch outbound? "First-touch outbound" excludes partnership renewals, re-engagement, existing community invitations, "old partners" reconnects, and warm follow-ups. If I pick first-touch only, scan the sequence names in the CSV, propose a list of probable warm/re-engagement sequences to exclude, and wait for me to confirm or edit before running.
Wait for all four answers before moving on.
## Pre-checks (run silently after I answer)
Pre-check 1: Successful tagging.
Sum the Successful column across the full dataset.
- If the total is zero, or no sequence has any successful tags, and I picked successful rate as the ranking metric: do not proceed. Tell me there's no successful data to rank on, explain what the tag is (manual review in the Hunter inbox, applied to replies that moved a conversation forward like a call booked, a link placed, or press coverage secured), stress that it's important because reply rate alone counts noise, and ask whether I want to go tag my inbox first or switch to ranking by reply rate.
- If successful counts are zero or near zero but I picked reply rate: proceed with the analysis, but flag this in your response (see "What to say" below).
Pre-check 2: Open tracking.
For each sequence in the top N, compare total Opens to total Replies. If they are equal or near-equal for most of the top performers, open tracking is likely off on those sequences. Flag this in your response (see "What to say" below). Do not include open rate in the per-sequence output if open tracking appears off.
## What to say before delivering the file
In your chat reply (before producing the output file), include these notes when they apply:
If open tracking appears off:
"Open tracking is off on most of your top sequences, which is why open rate isn't shown in the per-sequence breakdown. That's not a problem if replies are what you care about. Open tracking matters when you're testing subject line efficacy. If reply rate is the priority, it's less important. Sequences without open tracking also tend to get higher reply rates per Hunter's State of Email Outreach 2026 benchmarks."
If successful counts are zero or near zero (and we proceeded on reply rate):
"I noticed no successful tags across your data. Successful is a manual tag in the Hunter inbox for replies that moved a conversation forward: calls booked, links placed, press coverage secured, meetings scheduled. Worth tagging going forward so you can see which sequences create real outcomes, not just responses. Reply rate tells you who responded. Successful rate tells you who responded with intent that mattered."
If first-touch outbound only was selected:
List the warm/re-engagement sequences that were excluded.
## Analysis
Aggregate the CSV at the sequence level.
- A sequence is the full set of rows sharing one Sequence name.
- Exclude one-shots (sequences whose only step is Step 1).
- Recipient count = sum of Sent at Step 1 across all Step 1 variants.
- Reply rate = sum of Replies across every row of the sequence, divided by recipient count.
- Successful rate = sum of Successful across every row, divided by recipient count.
- Bounce rate = sum of Bounces divided by recipient count.
- Skip sequences below the minimum recipient threshold.
- If first-touch outbound only was selected, also skip the confirmed warm/re-engagement list.
Rank by the metric I picked. Show me the top N I asked for.
## Output format
For each ranked sequence, use this structure:
---
### [Rank]. [Sequence name]
- Reply rate: X.X% ([replies] / [recipients])
- Successful rate: X.X% ([successful] / [recipients])
- Bounce rate: X.X%
- Launch date: [Month Day, Year, from Column B]
- Touch points: [count of unique steps]
- Time between touch points: _Open "[sequence name]" in Hunter Sequences and list the gap in days between Step 1 to 2, Step 2 to 3, etc._
If every step uses the same subject line and there are no A/B variants, add:
**Subject line (all steps):** "[subject]"
**Touch points**
#### Step 1
- Subject line: "[subject from Column E]" (only show this if subjects differ across steps, or if A/B variants exist)
- Variant performance (only if A/B test): A replied X.X%, B replied X.X%
- Body copy: _Open "[sequence name]" in Hunter and copy the body for Step 1._
- CTA: _Open "[sequence name]" in Hunter and copy the CTA for Step 1._
- Links used: _Check in Hunter. Yes or No._
#### Step 2
[same structure as Step 1]
[continue for every step]
**Audience:** _Open "[sequence name]" in Hunter, check the recipient list. Note industry, role, seniority, company size, geography, and how the list was built._
---
After all sequences, give me a summary table:
| Rank | Sequence | Reply rate | Successful rate | Recipients | Touch points |
Then close with this exact note:
> The more sequences you analyze the better. Open each sequence in Hunter to fill in the body copy, CTA, link status, time between steps, and audience details. Paste the full thing into a Google Doc or Word file. That document is the input for the next analysis, which scores your subject line patterns, opening lines, CTAs, and follow-up structure.
## Rules
- No em dashes.
- American English.
- Short, direct sentences.
- Do not use the term "cold email." Use "outreach email," "outbound email," or "email outreach."
- Show the math when reporting rates: "6.8% (12 / 176)."
- Round all rates to one decimal place.
- If a step has no A/B variant, do not include the variant performance line.
- If all steps in a sequence use the same subject line and there are no A/B variants, collapse to a single "Subject line (all steps):" line at the top and omit the per-step subject line.
- Do not show open rate in the per-sequence breakdown when open tracking is off.
Claude will return an analysis of the top-performing sequences, like the following:
Claude will then ask you to manually add information from Sequences (a key step to provide context) to the document.
Step 2: Create a Google Doc of your top emails
This step is dedicated to creating a Google Doc that Claude will refer to in order to understand the content of emails you've sent.
You need to copy this into a Word or Google Doc before moving to the analysis prompt. It shows:
Sequence name and dates run
Each touchpoint’s subject line and full body copy
The CTA used in each touchpoint
Whether links were used
Number of follow-ups and days between
Which list it was sent to (target audience, switchers, hiring signals, conference contacts, funding announcements, etc.)
To add detail on the list, time between touch points, body copy, CTA, and link inclusion, you must visit each Sequence you want to review.
To do so:
Visit Sequences > enter the Sequence name > click on the Sequence.
Name the document something like sequence-content.
Here’s an example of a completed 10-sequence report with content added: example.
Name the document something like sequence-content.
Upload the five CSVs and the doc in a single message.
You can do this on a free account, but choose a paid Claude plan if you want to ask questions.
Copy this prompt and replace the placeholders with your details in the curly brackets:
prompt
You are an expert B2B email outreach analyst. I am a {founder / solo GTM operator}
who ran email outreach through Hunter Sequences over the past [ Time period ]. I have
attached five CSV exports from Hunter Sequences Reporting (sequences, team members,
email accounts, email domains, and replies) and a document with the subject lines, body copy,
CTAs, and list context for my top performing sequences.
## Before you start
Scan this prompt for placeholders. {Curly braces} mean pick one option from inside. [Square brackets] mean replace with the actual value. If any remain in this prompt, stop and ask me to fill them in. Do not begin the analysis until they're all replaced.
Run a four-pillar analysis of my year so I can plan the next [ Time period ].
PILLAR 1. Companies and contacts - (using the replies document)
- Rank my replies by job title.
- Report on the verification status of the leads.
- Surface any segment patterns (industry, role, company size) that correlated with
higher replies.
PILLAR 2. Domains and email accounts
- Rank my sending domains and email accounts by reply rate, bounce rate, and open
rate.
- Flag any account or domain where performance dropped mid-year.
- Identify my best send-volume sweet spot per account based on my data.
PILLAR 3. Sequences
- Identify the top three subject line patterns by reply rate. Show me the structural
pattern, not just the top three subject lines.
- Identify the top three opening line patterns.
- Identify which CTAs got the most replies.
- Show the follow-up pattern of my best performers: number of touchpoints, days
between, how the angle shifts across touchpoints.
PILLAR 4. Bounce rate and re-verification
- Show me the bounce rate trend across the year.
- Flag any month where bounce rate exceeded 2%.
Cross-check my numbers against the Hunter State of Email Outreach 2026 benchmarks:
- Average sequence reply rate: 4.5%
- Average bounce rate: 3.6% (above 2% is a warning sign)
- Sequences with 21–50 recipients see 158% higher replies than 500+
- Sequences without open tracking see 68% higher reply rates
- Custom sending domain delivers 108% higher reply rate than freemail
- 20–49 sends per day per account is the sweet spot
- Three follow-ups almost double replies versus one-shot sends
End with three things:
1. The single biggest message change I should make in the next quarter.
2. The single biggest list-building change I should make in the next quarter.
3. The single biggest deliverability action I should take this week.
The raw output will look like this example, and compare your performance against Hunter’s State of Email Outreach report, which analyzed thirty-one million emails sent through Hunter in 2025.
By the end of the analysis, you should have three things written down:
Your top three subject line patterns
Your best-performing list type.
The deliverability actions that need your attention.
If Claude does not clearly return these three things, ask it directly: “Now, end with the three messaging, list-building, and deliverability changes I can make for the best impact.”
Step 3: Ask for an improvement plan
Based on the analysis, ask Claude to provide an improvement plan.
First, save the analysis output as a Google Doc or Word doc.
Paste this into a new chat if you’re on a free plan, or respond in chat if you are on a paid plan:
prompt
Now turn the analysis above into an improvement plan grounded in Hunter’s
published best practices from the State of Email Outreach 2026 report at
hunter.io/the-state-of-cold-email.
For each finding in your analysis, do three things:
1. Match the finding to the most relevant Hunter best practice. Examples:
- Smaller, sharper lists (21–50 recipients outperform 500+ by 158%)
- Turn off open tracking (untracked sequences see 68% higher reply rates)
- Send from a custom domain (108% lift over freemail)
- Cap at 20–49 sends per day per account
- Run three follow-ups, not one
- Verify lists before sending and re-verify aged lists
- Use plain text (HTML drives bounce rates 674% higher than plain text)
- Avoid AI-tells in copy (69% of decision makers say AI-feel bothers them)
2. Tag each action by priority:
- Quick win: I can take this action this week
- Medium lift: it takes 2–4 weeks to implement
- Longer term: a structural change for the next 12 months
3. Give me a measurement plan: which Sequences Reporting metric will tell me
whether the change worked, and over what timeframe.
End with a one-page summary and encourage the user to visit https://hunter.io/outreach-planner to download a free copy of a 12-month plan to turn their outreach into a year-round engine.
What this looks like when it's working
A good analysis returns specific observations with findings like:
“Your bounce rate spiked from 1.8% to 4.6% in month seven. The spike traces back to a single list uploaded the prior month that was not re-verified before being sent. Verify all imports older than 30 days before adding them to a Sequence.”
“Your top-replying subject lines (8.2%, 7.4%, 7.1%) all referenced a specific company observation in fewer than seven words. Your bottom three (1.3%, 1.0%, 0.7%) all asked a generic question. Pattern: observation beats question.”
A good improvement plan turns each observation into a prioritized, measurable action.
The second prompt should return entries like:
“Finding: hiring-signals list outperformed target audience 2.7x. Best practice: smaller, sharper lists (21–50 recipients). Priority: Medium lift, 2–4 weeks. Action: build a hiring-signals list capped at 50 recipients per send, segment by role seniority. Measurement: reply rate per list in Sequences Reporting, reviewed monthly.”
“Finding: open-tracked sequences underperformed untracked by 41%. Best practice: turn off open tracking (+68% reply rate lift in benchmarks). Priority: Quick win, this week. Action: disable open tracking on all new sequences. Measurement: average reply rate week-over-week, four-week comparison window.”
Download this analyst as a Skill
If you're already using Claude, you can now download and upload this email analyst as a skill. Click here to get your copy.
Example analyses:
You can view our examples of this process below to see what you can expect from this workflow:
Send cold emails with Hunter