What Should Revenue Operations Teams Evaluate in Sales Cadence? A 7-Step Checklist
2026-08-17 · Julian Hartwell
-
1. Audit the Contact Data Before You Critique Anything Else
-
2. Make Sure Every Touch Has a Job
-
3. Apply the "Key Account" Test to AI-Written Emails
-
4. Read the Documentation — Including the Verification API Docs
-
5. Check Whether the Cadence Reacts to Intent
-
6. Audit the Research Layer — Are Triggers Actually Wired?
-
7. Test the Reply-Handling Workflow — the Step Nobody Checks
- Common Errors and Final Notes
In my role, I review outbound sales sequences before they're allowed to reach prospects — roughly 40 to 50 cadences a quarter. In 2025 so far, I've rejected about 30% of the first drafts that crossed my desk. The reasons are almost always the same, and they've turned into a working checklist.
If you're in revenue operations — a RevOps lead, a sales ops manager, or a founder who suddenly owns outbound quality — this list is for you. It's not a strategy essay. It's the exact sequence of checks I run before any cadence ships: the boring ones, the technical ones, and the one step that almost everyone skips until it hurts.
There are 7 steps. Go in order. Most teams do steps 1 through 3, stumble on 4, and never get to 5, 6, or 7 consciously. That last part is the expensive mistake.
1. Audit the Contact Data Before You Critique Anything Else
The best email in the world fails if it lands in a dead inbox. I can't tell you how many times a team has blamed a cadence for "poor performance" when the real problem was a stale list. So this is where every evaluation starts.
- Historical bounce rate: if recent sends are bouncing at 3-4% or higher, the data is the problem, not your copy.
- Data age: a list from 8 months ago can easily be 30-40% stale in B2B. People change jobs fast.
- Verification status: are addresses checked at the point of upload, or was it a one-time cleanup that happened months ago?
I use Wiza's email checker on every new contact before they enter a sequence. I'm not a data scientist, so I can't speak to the internal mechanics of how verification works. What I can tell you from a quality perspective: if you don't verify at the point of data entry, everything downstream is built on sand. The best AI-written email in the world won't fix a 10% bounce rate.
Checkpoint: before a cadence goes live, project your first-send bounce rate below 3%. If you can't, stop and clean the data. Do not ship it anyway and hope it improves. That's a hope-based strategy, not a quality plan.
2. Make Sure Every Touch Has a Job
Here's a pattern that shows up constantly: "Day 1 email, Day 3 email, Day 7 LinkedIn, Day 10 email" — with nothing connecting the touches except time. It's tempting to think cadence quality is about timing and frequency. That was more or less true in the early days of outbound automation. But what was best practice in 2020 does not automatically work in 2025. Buyers have more noise than ever, and they can feel a generic sequence within two sentences.
When I evaluate a sequence, I map every touch to a job:
- Touch 1: establish relevance — why me, why now.
- Touch 2: deliver an insight or data point the prospect hasn't seen yet.
- Touch 3: create a decision point — a question, an offer, an objection-handling move.
- Any touch that doesn't have a defined job gets cut. Fewer touches, each doing more work, is the target.
This is where I'd challenge most cadences: the problem isn't that they send too many emails. It's that large chunks of those emails say the same thing in different words. "Quality over volume" is a cliché, but in this case it's also a spec.
Checkpoint: write down the full sequence map on one page, and next to each touch, note its job in a single sentence. If you can't fill that in, the touch shouldn't ship.
3. Apply the "Key Account" Test to AI-Written Emails
AI email writers are standard now. Wiza's AI email writer, for instance, generates and personalizes large volumes of outbound email every month. That's a genuinely good development — it moves the work from writing to reviewing. But the reviewing has to be rigorous, and I see a lot of teams treating it as a quick skim.
I did that myself once, early on. We shipped a sequence with an obviously generic AI opening line to a high-value account. The reply started with "Do you actually know who I am?" — that's a brand-reputation problem, not a one-off miss.
My review rubric for AI-written emails has three checks:
- Does it sound like a human wrote it about this specific person — not "a person in this role," but this person?
- Is the personalization based on current, verified information? If it references something from last year, that section gets deleted.
- Are there filler phrases — "I know you're busy," "I hope this finds you well"? Two of those in the first two lines, and it's a reject.
Honestly, I'm not 100% sure why AI tools still default to those filler patterns. My best guess is that the prompts are too generic. From my perspective, the best AI-written email is the one you can't tell was AI-written. That's the only acceptable standard.
Checkpoint: read the email out loud. If you stumble, so will a prospect. Fix it or reject it.
4. Read the Documentation — Including the Verification API Docs
We're at the boring step now. But every quarter, I see at least one cadence underperform because of a technical issue that was documented in plain sight.
The list of "boring stuff" isn't long: an email verification API that runs automatically on new contacts (not a one-time batch from last quarter); authentication records like SPF, DKIM, and DMARC that quietly break deliverability when misconfigured; reply detection that actually ends the sequence when a prospect responds. Any one of these failing is enough to sink a good cadence.
This gets into engineering territory, which isn't my core expertise. I've read enough email verification API documentation to spot config gaps, but I always bring in a developer for anything deep. The quality question is simpler: is the infrastructure solid enough that your messages actually land?
A caution from my own work: I approved a sequence once that passed every copy and data check. Two months in, it was underperforming badly. The cause? An outdated verification API key that had failed silently, so the system had been collecting unverified contacts for weeks. The documentation spelled out the requirements clearly. Nobody re-read it after setup. The fix took ten minutes; the frustration lasted a month.
Checkpoint: open the verification API logs for the last 30 days. If there are failures, don't launch anything until they're resolved. And make sure every email includes a valid physical postal address — that's a legal requirement for commercial email under the CAN-SPAM Act (ftc.gov).
5. Check Whether the Cadence Reacts to Intent
A cadence that sends the same steps to every prospect, regardless of buying signals, is basically a broadcast. And broadcasts are easy to ignore.
Intent data changes the equation. If your sales intelligence platform shows that a target account is actively researching the category your product serves, the sequence should adapt — different message, different urgency, sometimes a different channel.
The fundamentals of good outreach haven't changed: clear value, respect for the buyer's time. But the execution has transformed. In 2025, treating an intent-qualified lead exactly the same as a completely cold contact leaves most of your pipeline potential sitting on the table.
Checkpoint: for any prospect showing active intent — recently visited pricing, searched your category, engaged with your content — does your sequence change? If the answer is no, this is your biggest gap.
6. Audit the Research Layer — Are Triggers Actually Wired?
The best outbound sequences use LinkedIn Sales Navigator as a research layer. Job changes, company news, leadership moves — all of these should trigger timely, human touches. Most cadences I review just use Sales Navigator as a place to fish for connection requests.
When I evaluate a cadence, I ask the sales rep: what do you actually know about this prospect before you send? Does the system prompt you to check recent Sales Navigator activity? Or is the whole sequence running on full automation with no human judgment points?
This might be controversial: fully automated cadences have a performance ceiling. Not because automation is bad — it's essential for consistency and scale. But the sequences that consistently win blend automated structure with human judgment. The automation handles the structural emails; the human handles the moments that follow a trigger. A prospect who just changed jobs is the easiest meeting you'll book all week. A prospect whose company just announced layoffs needs a different, more respectful message. Automation can't tell those apart — a human with Sales Navigator can.
Checkpoint: for a trigger event like a job change, is there a defined handoff to a human touch within 48 hours? If not, you're losing your best conversation starters.
7. Test the Reply-Handling Workflow — the Step Nobody Checks
This is the one that surprises people. It's the most commonly ignored part of cadence quality, and it's a genuine competitive advantage if you handle it well.
When a prospect replies — positively, negatively, or with a question — what happens next?
- Does the sequence auto-exit when a reply is detected?
- Is there a response-time SLA for the rep? Same-day should be the minimum.
- Are different reply types routed differently? "Take me off your list" is not the same as "I'm interested" or "circle back next quarter."
If you've ever had a prospect say "yes, let's talk" and then wait two weeks for a meeting to get booked because nobody handled the reply — you know that sinking feeling. There's something genuinely satisfying about a reply-handling workflow that works: the notification fires, the rep responds within the hour, and the meeting lands before the prospect's interest fades. That's what quality looks like end to end.
Checkpoint: send a test reply to your own sequence. Time how long it takes for the sequence to stop and for a human to get pinged. Most teams find out the hard way that none of this works the way they assumed.
Common Errors and Final Notes
A few things I've learned the expensive way:
Don't check these in the wrong order
Data before copy. Copy before infrastructure. Infrastructure before triggers. If you start by critiquing the message, you'll fix the wrong layer and wonder why nothing improves. I've made that mistake; nearly every vendor I've worked with has made it too.
Don't over-optimize deliverability while ignoring the message
I've met teams that obsess over SPF, DKIM, and bounce rates while sending content that no one wants to read. Reachability without relevance is still spam. Both dimensions matter, and you can't compensate for one by perfecting the other.
Don't treat all prospects as one segment
If you're running one "perfect" cadence for every prospect, it's a broadcast with extra steps. Build at least two or three variants — by intent, by company size, by engagement level. The tools support it. Your results will show it.
Write down your specs before you launch
This is the quality auditor in me talking: a cadence without documented standards isn't really evaluable. Define your thresholds — bounce targets, response SLAs, personalization requirements — before launch, not after three months of "it's underperforming and nobody knows why." When I implemented this discipline for our own outbound process, the difference was night and day.
That's the full checklist. It won't cover every edge case — every business has its quirks. But if a cadence genuinely passes all seven checks, it's in the top tier of what I've seen in four years of reviewing sequence quality. If you're evaluating your own sequence today, start with step 1. It's the least glamorous and the most important.