Your reply rates just dropped 40%. You have a client call tomorrow. You have no idea what changed.

Sound familiar? This happens to every outbound team at some point. The frustrating part is not the drop itself - it is that cold email campaigns fail silently. An account disconnects but still looks active. Warmup stops running but shows no alert. A bounce rate creeps past 5% before anyone runs a report. By the time the problem shows up in your dashboard, you have already burned a week of sends.

Smartlead MCP changes that. When Claude is connected to your Smartlead account via MCP, it can diagnose your entire outbound infrastructure in real time - account health, warmup status, bounce rates, deliverability signals - the same way a systems engineer would, except in a conversation instead of an afternoon of manual work.

Here is exactly how to use it when your campaign is on fire.

TLDR

What this covers: How to use Smartlead MCP and Claude to diagnose why a cold email campaign stopped performing - in about 90 seconds.

Start with this prompt when performance drops:

My reply rates dropped about 40% this week. Run a full diagnostic across which
sending accounts changed status in the last 7 days, which have bounce rates above 4%,
which warmup scores dropped more than 10 points, and which accounts disconnected at
any point in the last 7 days.


Claude correlates events across data sources and tells you the most likely cause. If accounts disconnected on Tuesday and reply rates dropped on Wednesday, Claude connects those dots.

The four root causes to know: disconnected accounts, warmup degradation, bounce rate spike, and copy or subject line fatigue.

Key thresholds to act on:

• Bounce rate 4%+ means act immediately. Above 5% means pause to avoid domain reputation damage
• Warmup below 70 means pause live sending. Between 70 and 80 means monitor daily. Above 85 means healthy

What To Do If Your Campaign Is in Crisis

Picture this. Your weekly numbers are in and reply rates are down 40% across three active campaigns. You have no idea what changed because nothing was touched this week.

The old way of handling this was painful - open five browser tabs, check account statuses one by one, export stats to a spreadsheet, compare last week's send volume by hand, make a guess, fix one thing, then wait 48 hours to see if it helped.

The new way is to open Claude and ask.

Smartlead MCP exposes 116+ tools to Claude - including a deliverability diagnostics category that covers account connection status, warmup health scores, bounce rate history, and domain signals. This category does not exist in other cold email MCP integrations. It was built specifically for situations like this one, where you need to know why your campaign broke and you need to know fast.

The First Prompt - Getting the Full Picture

When you do not know where the problem is, start broad. Do not ask Claude to check one thing - ask it to check everything at once.

Here is the diagnostic prompt that surfaces most campaign crises in a single pass:

My reply rates dropped about 40% this week compared to last week. Run a full
diagnostic across which sending accounts changed status in the last 7 days, which
have bounce rates above 4%, which warmup scores dropped more than 10 points, and
which accounts disconnected at any point in the last 7 days.


When Claude runs this against your Smartlead account, it calls the account health tools, checks warmup data for each connected account, pulls bounce rate history, and identifies which accounts lost their connection - even temporarily - in the window you specified.

What comes back is not a raw data dump. Claude returns a structured summary: which accounts are affected, what specifically changed, and which of those changes are most likely responsible for the performance drop. If three accounts disconnected on Tuesday and your reply rate dropped starting Wednesday, Claude will say so directly.

This gives you a short list of suspects instead of an entire infrastructure to inspect.

Stop Flying Blind on Campaign Health

Start your free Smartlead trial and connect MCP today - your first diagnostic runs in under 5 minutes.

Interpreting What You Find - The Four Most Common Root Causes

The diagnostic prompt above will typically surface one of four problems. Each has a different fix, so it matters which one you are dealing with.

Disconnected accounts

An account can lose its connection silently. OAuth tokens expire, passwords change, Google or Microsoft revokes access after a security event or policy update - and the account continues appearing "active" in your campaign setup while no emails are actually leaving.

If you sent 500 emails last week and two of your six accounts silently disconnected on day three, you effectively sent 167 of them. Your reply rate looks like it collapsed when really your send volume did.

Fix: reconnect via the OAuth flow in Smartlead's account settings. Check whether the account missed any scheduled sends during the disconnection window and factor that into your reporting.

Warmup degradation

Warmup scores are not static. An account sitting at 85/100 three weeks ago can drop to 40/100 if warmup email engagement rates fall - which happens when warmup pool members start marking warmup mail as spam, or when your sending domain picks up a reputation signal that warmup alone cannot offset.

When a warmup score drops below the threshold needed to maintain inbox placement, your live campaign emails start going to spam. Reply rates fall but the account still appears connected and healthy in your dashboard because technically it is. The problem is in the warmup layer underneath.

Fix: pause live sending from that account immediately and let warmup rebuild over 7 to 14 days before resuming. Pushing volume through an account with degraded warmup accelerates the damage.

Bounce rate spike

Bounce rates above 4 to 5% are a hard signal to inbox providers. Once a sending domain consistently returns that level of hard bounces, mail from that domain starts being routed to bulk folders - sometimes across all recipients, not just the ones who bounced.

A bounce rate spike usually comes from one of three sources - a list quality problem with invalid or stale addresses, a list source problem with purchased or scraped data, or an account being used across too many unrelated audiences at the same time.

Fix: pause the affected account and do not continue sending while the bounce rate is elevated. Audit the list that account was sending to, and when you resume, start at 20 to 30% of your previous daily volume and increase gradually over 10 days.

Copy or subject line fatigue

Not every performance drop is an infrastructure problem. If you run the account diagnostic and everything looks healthy - connections intact, warmup scores stable, bounce rates under 2% - then the problem is the message, not the tool.

Copy fatigue is real. Subject lines that worked in January stop working in March. Competitor campaigns using similar hooks dilute your open rates. Recipients are choosing not to respond, and no amount of infrastructure debugging will fix that.

How to tell the difference: after running the full diagnostic, if Claude reports all accounts healthy, ask it to pull open rate and reply rate trends by campaign and by sequence step. A healthy open rate with a collapsing reply rate points to body copy. A collapsing open rate points to subject lines or deliverability.

The Follow-Up Diagnostic Prompts

Once the first diagnostic gives you suspects, these follow-up prompts scope the damage and identify the timeline.

To understand when accounts disconnected and what was affected:

For the accounts that disconnected in the last 7 days - when exactly did each one
disconnect, and which campaigns were actively sending through them during that window?


To quantify the campaign-level impact:

Which campaigns were assigned to the accounts that disconnected, and what was the
average reply rate on each campaign in the 7 days before the disconnection compared
to the 7 days after?


This gives you a before and after for each affected campaign. If reply rate on Campaign A dropped from 4.2% to 1.1% on the exact day Account 3 disconnected, you have a clear, documentable root cause.

To trace a bounce rate spike to its source:

For the accounts with the highest bounce rates in the last 14 days - which lead
lists were they sending to, and what was the list upload date for each?


Old lists have high bounce rates. Lists from certain sources have high bounce rates. This prompt identifies whether the bounce problem is isolated to one list or spread across your entire lead pool - and that distinction changes the fix completely.

The Difference Between This and Checking Your Dashboard

The Smartlead dashboard shows you the current state of your campaigns and accounts. It is great for monitoring and for periodic reviews.

MCP diagnostics is different because Claude does not just retrieve data - it reasons about it in context. It sees that Account 4 disconnected on Tuesday, that Campaign B's reply rate fell on Wednesday, and that Campaign B was running 60% of its sends through Account 4. It connects those dots without you having to.

The dashboard tells you what the numbers are. Claude tells you what the numbers mean and why they changed when they did.

If you want to go deeper on what MCP can do beyond diagnostics, the MCP for RevOps guide covers the full reporting and monitoring workflow, and the AI agents for cold email post explains how to move from reactive diagnosis to proactive monitoring.

The 5-Minute Morning Check That Prevents Most Crises

The diagnostic workflow above is for when something has already gone wrong. But most of these problems are detectable before they affect performance.

Running this prompt every morning before campaigns go out catches roughly 80% of infrastructure issues before they reach your reply rates:

Check all sending accounts for any disconnections, warmup health below 70%, or
bounce rates above 4% in the last 48 hours. Summarize anything that needs attention
before today's sends go out.


Make it as automatic as checking your inbox. The time investment is 90 seconds. The cost of skipping it is a week of burned sends and a difficult client conversation.

Getting Started With Smartlead MCP

If you have not connected Smartlead MCP yet, setup takes about 10 minutes. Once Claude has access to your Smartlead account, every diagnostic prompt in this post works immediately.

The full MCP setup guide walks you through the configuration step by step. And if you want the complete reference of available tools, the Smartlead API documentation covers all 116+ endpoints.

The Bottom Line

Campaign performance drops are not always obvious and they are almost never caused by the one thing you guessed first. The real cause is usually a combination of events across multiple accounts and data sources that is hard to see when you are checking things one tab at a time.

Smartlead MCP gives Claude the access to correlate those events automatically. A Tuesday disconnection plus a Wednesday reply rate drop is not a coincidence - it is a root cause. And finding it in 90 seconds instead of 45 minutes means you fix it before it compounds into two weeks of degraded performance and a client asking why results dropped.

Connect MCP once, run the morning check daily, and you will probably never spend an afternoon debugging a campaign crisis again.

Get started with Smartlead for free and run your first campaign diagnostic today.

FAQs

How do I diagnose why my Smartlead campaign performance dropped?

Start with a broad diagnostic prompt that covers the four most common causes at once - account disconnections, bounce rate spikes, warmup score drops, and account status changes. Ask Claude to check all of these across the same 7-day window as your performance drop. This surfaces the root cause in a single pass rather than checking each factor manually across multiple tabs. From there, use follow-up prompts to scope the timeline and campaign-level impact.

What is Smartlead MCP diagnostics?

It is a category of tools within the Smartlead MCP integration that gives Claude real-time access to your account health data - including warmup scores, bounce rates, account connection status, and domain signals. Unlike a static dashboard, MCP diagnostics lets Claude reason across multiple data sources simultaneously and surface the specific combination of factors causing a performance problem.

Can Claude diagnose deliverability issues in Smartlead?

Yes. When connected to Smartlead via MCP, Claude can check warmup health scores, identify accounts with bounce rates above your thresholds, flag accounts that lost their connection, and correlate those account-level events with campaign-level performance changes. It does not just retrieve the data - it interprets it in context and identifies which combination of factors most likely caused a specific drop during a specific time window.

What is the difference between the Smartlead dashboard and MCP diagnostics?

The Smartlead dashboard shows you the current state of your campaigns and accounts. MCP diagnostics lets Claude reason about that data in context - correlating a Wednesday reply rate drop with a Tuesday account disconnection, or tracing a bounce rate spike to a specific list uploaded on a specific date. The dashboard surfaces data. MCP diagnostics surfaces answers.

How do I check warmup health using Smartlead MCP?

Ask Claude directly: "Check the warmup health score for all connected accounts and flag any that have dropped more than 10 points in the last 7 days or are currently below 70." Claude pulls the warmup data for each account via MCP and returns a prioritised list. Accounts below 70 warmup health should be paused for live sending immediately. Accounts between 70 and 80 should be monitored daily. Above 85 is a healthy range for normal send volumes.

What bounce rate should I be worried about in Smartlead?

Anything above 4% is a serious signal. At 4 to 5%, inbox providers begin treating mail from that sending domain as bulk. Above 5%, you risk domain-level reputation damage that persists even after you fix the underlying list quality issue. The standard practice is to pause any account that hits 4%, audit the list, and when you resume, start at 20 to 30% of your previous daily volume and scale up over 10 days while monitoring deliverability daily.

‍