Stop Overengineering Your Sales Process
I just watched a sales team spend a week building a workflow that got them zero leads.
Not zero qualified leads. Zero leads. A week of work, a 47-step automation, and nothing to show for it.
This keeps happening. Sales leaders see a demo of some complex Clay or Make workflow on LinkedIn, get FOMO, and immediately try to replicate it. They spend days wiring up enrichment steps, routing logic, and automated sequences before they’ve answered the most basic question: do we know what to say to who?
Automations are powerful. We’ve spent every day for the last four years building them. But an automation is an amplifier. If you amplify a message nobody cares about, you just fail faster and at greater expense.
The 3-Question Diagnostic
Before you build anything, answer these three questions honestly:
1. Can you identify in one sentence what your prospect’s biggest problem is?
Not your product’s value prop. Not your feature list. The actual problem keeping your buyer up at night. If you can’t say it in one sentence, you don’t know it well enough to write an email about it, let alone automate 10,000 of them.
I’ve seen teams automate personalized outreach at scale where the “personalization” was surface-level metadata — the prospect’s job title, their company name, maybe a recent LinkedIn post. None of that matters if the underlying message doesn’t connect to a real pain point. You just sent 10,000 emails that say “I noticed you’re the VP of Sales at Company X” and then pitch a product that doesn’t map to anything the VP of Sales actually worries about.
2. Have you nailed down the right audience yet?
“Series A SaaS companies” is not an audience. “Series A vertical SaaS companies in logistics who just hired their first sales team and are running outbound for the first time” is an audience. The difference between those two statements is the difference between a 0.5% reply rate and a 4% reply rate.
When you automate against a vague ICP, you burn through your addressable market with bad messaging. Those prospects are now trained to ignore you. You don’t get a second first impression with most of them.
3. Do you have an offer that actually resonates?
Your offer is not your product. It’s the reason someone should care right now. “We help companies with sales” is not an offer. “We build the outbound infrastructure for Series A teams so their first two SDR hires don’t waste their first quarter figuring out email deliverability” is an offer.
If your offer doesn’t make the right person stop scrolling, no amount of automation will fix that.
What Happens When Teams Skip This
I see the same failure pattern play out every few months:
The premature automation spiral. A team invests two weeks building an elaborate workflow. Enrichment, scoring, routing, multi-channel sequences. They launch it, get terrible results, and conclude that “outbound doesn’t work for us.” Outbound works fine. They just automated before they had anything worth automating.
The data quality illusion. A team sources 50,000 contacts from three different databases, deduplicates them, enriches them with firmographic data, scores them with a custom model, and routes them into sequenced campaigns. Beautiful engineering. Terrible results. Because the underlying audience definition was wrong, so they built a precision instrument aimed at the wrong target.
The “we need more volume” trap. Results are bad, so the team decides the problem is scale. They add more prospects, more sending domains, more sequences. Reply rates stay flat because the problem was never volume — it was relevance. More volume with bad messaging just gets you blacklisted faster.
In every case, the team would have gotten better results from 50 manually written emails to well-researched prospects than from 5,000 automated emails to a poorly defined list.
The Manual-First Rule
We have a rule at Zevenue: before anything gets automated, it gets done manually first.
This isn’t about being slow. It’s about learning. When you manually research 20 prospects, write 20 personalized emails, and send them yourself, you learn things that no amount of workflow design will teach you:
- Which signals actually correlate with interest
- Which pain points get responses and which get ignored
- What level of personalization matters versus what’s just noise
- Whether your offer lands at all
Manual outreach is your testing lab. You’re running micro-experiments with every email. The replies (and the non-replies) tell you what works before you scale it.
Once you have that — once you can reliably get replies from a specific audience with a specific message about a specific problem — then you automate. Then the 47-step workflow makes sense, because you’re amplifying something that already works.
The sequence matters: understand first, then build. Not the other way around.
How to Actually Do This
If you failed one or more of the three questions above, here’s what to do this week:
Pick up the phone. Call your five most loyal customers and ask them three things: What was happening in your business when you decided to buy? What almost stopped you from buying? What would you tell a peer who’s considering something like this?
Do this every quarter. Not just when a new sales leader joins. Markets shift. Pain points evolve. The language your customers use changes. If your last customer conversation was six months ago, your messaging is stale.
Write 20 emails by hand. No templates. Research each prospect individually. Write each email from scratch. Send them. See what happens. The patterns that emerge from those 20 emails will be worth more than any workflow you could build blind.
Then, and only then, automate. Take what worked manually and encode it. The enrichment steps that surfaced real signals — automate those. The message structure that got replies — templatize that. The audience criteria that produced engaged prospects — build a sourcing workflow around that.
Automation is the last step, not the first. The teams that understand this build systems that actually produce pipeline. The teams that don’t build expensive machines that produce nothing.