Campaign Ops Skills for AI Agents

Purpose

Turn a raw bounce log into clean actions so you stop burning sender reputation on bad addresses.

Inputs

• CSV export of bounces with email, bounce type, and provider message
• Your current sending domain list
• Suppression list you already maintain

Steps

1. Parse each bounce row and pull the raw SMTP response text.
2. Classify each into hard invalid, mailbox full, greylist, policy block, or unknown.
3. Route hard invalid and no such user to a permanent suppression file.
4. Flag mailbox full and greylist as retry after 72 hours with a capped retry count.
5. Group policy block reasons by receiving domain to spot a domain reputation problem.
6. Cross check flagged addresses against your existing suppression list to avoid duplicates.
7. Summarize bounce rate by sending domain and note any domain above 4 percent.

Output

A suppression file, a timed retry queue, and a short reputation note per sending domain with any domain that needs a pause.

Purpose

Verify every launch condition is met before a campaign goes live so no broken send slips through.

Inputs

• Campaign name and target segment size
• Sending mailbox list and daily cap settings
• Sequence copy with merge fields
• Suppression and prior-contact lists

Steps

1. Confirm the target list is deduped against active and past campaigns and against the suppression file.
2. Check every merge field resolves for at least 95 percent of rows and flag rows missing critical fields.
3. Validate mailbox health, warmup status, and combined daily cap against list size to estimate days to complete.
4. Scan copy for missing unsubscribe language, broken links, and unfilled placeholder tokens.
5. Confirm send window, timezone alignment, and step delays match the intended cadence.
6. Roll all checks into a pass, warn, or fail status per item.

Output

A readiness report with per-item pass, warn, or fail, a blocking summary, and a clear go or no-go verdict with the reasons any launch is held.

Purpose

Turn a thin one or two email sequence into a full follow up cadence that stays useful and not annoying.

Inputs

• The existing emails in the sequence
• The offer and the target persona
• Any case studies, data points, or resources you can cite

Steps

1. Map what each current email already says so new steps do not repeat an angle.
2. Draft follow up steps that each add one new thing: a proof point, a different pain, a short question, or a resource.
3. Keep every new email under 90 words and end with a single clear ask.
4. Space steps 3 to 4 days apart and cap the sequence at 5 to 6 touches total.
5. Write a final breakup email that makes it easy to say not now without burning the relationship.
6. Check the full set reads like one person wrote it, with no repeated openers or phrases.

Output

The complete numbered sequence with send delays, ready to paste into your sending tool.

Purpose

Catch leads sitting in two or more active campaigns before they get hit with conflicting messages on the same day.

Inputs

• CSV exports from each active campaign with email, first name, company, campaign name
• List of currently running campaign names
• Your rule for which campaign wins a conflict (for example priority order or newest enrollment)

Steps

1. Load every campaign export and normalize the email column to lowercase, trimmed.
2. Group all rows by email address and count how many distinct campaigns each address appears in.
3. Flag any email present in two or more campaigns and list which campaigns hold it.
4. Apply your win rule to decide which campaign keeps the lead and which should pause or remove it.
5. Note timing overlaps where two campaigns would send within the same 48 hour window.
6. Build a remove list per campaign so each address stays in exactly one.

Output

A table of overlapping leads with their campaigns, the chosen keeper, and a per-campaign removal list ready to paste back into your sending tool.

Purpose

Split a required daily send volume across many sending mailboxes so no single inbox breaks its safe limit.

Inputs

• List of active inboxes with per-mailbox daily cap and current warmup stage
• Target total sends needed today across all campaigns
• Any inboxes flagged as paused or under review

Steps

1. Pull each inbox, its cap, and its warmup stage into one table.
2. Remove paused or flagged inboxes from the eligible pool.
3. Sum the caps of remaining inboxes to confirm total capacity covers the target.
4. If capacity is short, flag the gap and cap the plan at safe max instead of overloading.
5. Allocate sends proportionally to each inbox cap, rounding down to whole emails.
6. Reserve a 10 percent buffer on newer warmup inboxes to protect reputation.
7. List any leftover volume that must roll to tomorrow.

Output

A per-inbox send allocation table for today, a total planned versus target line, and a flagged list of paused inboxes and rollover volume.

Purpose

Catch the settings that quietly kill a cold email launch before the first send goes out.

Inputs

• Export or screenshot of the sequence steps and delays
• Sending domain and mailbox list
• Daily send cap per mailbox

Steps

1. Confirm SPF, DKIM, and DMARC pass for every sending domain, and flag any mailbox missing a record.
2. Check that each mailbox has completed at least two weeks of warmup and stays under 40 sends per day.
3. Scan every email body for spam trigger words, link count, and image weight, and list each hit with the step it appears in.
4. Verify delays between steps are 2 days or more and that no step fires on a weekend by default.
5. Confirm the unsubscribe path and a plain text signature are present in step one.
6. Return a pass or fail per check with the exact fix for each failure.

Output

A go or no-go table listing every check, its status, and the specific change needed before launch.

Purpose

Spread a backlog of queued emails across available mailboxes so no inbox exceeds its safe daily cap while the campaign still finishes on schedule.

Inputs

• List of mailboxes with per-mailbox daily send cap and current warmup stage
• Total queued contacts and target completion date
• Current per-mailbox scheduled load

Steps

1. Sum total capacity by adding each mailbox cap adjusted for warmup stage.
2. Divide queued contacts by days remaining to get required daily throughput.
3. Flag any day where required throughput exceeds total capacity and extend the timeline.
4. Allocate the daily target across mailboxes proportional to each safe cap, never over cap.
5. Reserve 10 to 15 percent headroom on newly warmed mailboxes to avoid spikes.
6. Produce a per-mailbox per-day send schedule and note the projected finish date.

Output

A table of daily send counts per mailbox, a flag list for over-capacity days, and the adjusted campaign completion date.

Purpose

Shift outbound send times so each contact receives email during their own local working hours instead of the sender's.

Inputs

• Contact list with company city, state, or country fields
• Target local send window, for example 8am to 11am
• Sending capacity per hour from your platform

Steps

1. Parse each contact's location field and resolve it to a timezone using city and country lookups.
2. Flag rows where location is missing or ambiguous and route them to a default timezone bucket.
3. Convert your target local window into UTC send slots for every distinct timezone in the list.
4. Group contacts by timezone and count volume per slot against your hourly capacity.
5. Spread overflow into the next valid slot so no single hour exceeds capacity.
6. Write a scheduled_send_utc value and a confidence tag to each row.

Output

The enriched list with resolved timezone, UTC send time, and a summary of contacts per window plus unresolved rows needing manual review.

}}<|entity|> error correction below.</title>title></

Purpose

Spread outbound volume across a pool of sending mailboxes so no single inbox exceeds its safe daily cap.

Inputs

• List of sending mailboxes with per mailbox daily cap
• Total contacts queued for the day
• Current sent counts per mailbox for the day

Steps

1. Pull the mailbox list and note each cap plus how many sends already went out today.
2. Compute remaining capacity per mailbox by subtracting sent count from cap.
3. Sum remaining capacity and compare against the queued contact total to flag overflow.
4. Assign contacts round robin across mailboxes with capacity, filling the lowest utilized first.
5. Reserve a ten percent buffer on each mailbox so no inbox hits its exact ceiling.
6. If demand exceeds total capacity, list the overflow count and the earliest next day slot.
7. Output the final contact to mailbox assignment table.

Output

A per mailbox assignment table showing planned sends, remaining buffer, and any overflow deferred to the next day.

Purpose

Set clear numeric thresholds that automatically pause a running outbound sequence before deliverability or reputation drops.

Inputs

• CSV export of current sequence metrics (sends, bounces, replies, spam complaints, unsubscribes)
• Target thresholds or acceptable ranges if you have them
• Number of active sending mailboxes

Steps

1. Parse the metrics CSV and compute rolling rates for bounce, spam complaint, unsubscribe, and negative reply per mailbox.
2. Flag any mailbox that crosses a hard limit, for example bounce over 3 percent or complaints over 0.1 percent.
3. Rank mailboxes by risk so the worst offenders surface first.
4. Write plain kill switch rules that state the metric, the trigger value, the window, and the action to pause.
5. Add a warmup exception clause so new mailboxes get a lighter first week limit.
6. Produce a short daily check routine the operator can run manually.

Output

A rules table with trigger conditions plus a list of mailboxes to pause now and a daily monitoring checklist.

Purpose

Design a clean subject line test that produces a readable winner instead of noise.

Inputs

• The campaign audience size
• Two to four candidate subject lines
• Your current baseline open or reply rate

Steps

1. Calculate the minimum sample per variant needed to detect a 20 percent lift at your baseline, and warn if the audience is too small to call a winner.
2. Split the list randomly so each variant gets an equal, non overlapping slice of contacts.
3. Hold every other variable steady: same send window, same body, same sender, same day.
4. Define the primary metric up front, reply rate for cold outbound rather than opens, plus one guardrail metric like bounce rate.
5. Set a decision date based on when the sample fills, not a fixed calendar day.
6. On the decision date, compare variants and state whether the gap is large enough to act on or still inside noise.

Output

A test plan with per variant list sizes, the metric to judge on, the decision date, and a rule for declaring the winner.