The Isenberg Workflow Audit: Find What to Automate Before You Touch a Tool

Greg Isenberg posted a thread in 2026 that quietly became the most passed-around automation framework in founder circles. It wasn't about a tool. It wasn't about an AI model. It was a list.

Step one: map your niche's workflow end-to-end, every single step, from lead to payment. Step two: find where money actually changes hands. Step three: find what someone does the same way every single day. That's your automation target.

The thread spread because it named the mistake almost every operator makes: they start with the tool. They see a shiny workflow builder, they connect some apps, they automate something, and three months later it's either broken or irrelevant because it never attached to a real business problem.

The operators who win aren't faster at building. They're better at knowing what to build. Here's how to run Isenberg's framework inside your own business before you touch a single node.

Woman examining workflow diagram drawn on green floor in minimalist room

Why Most Automations Fail Before They're Built

There's a specific pattern that kills automations before they ever deliver value. It looks like this: operator discovers a platform, gets excited, builds a workflow that automates something, ships it, and then watches it gather dust because nobody on the team trusts it, the output doesn't connect to revenue, or the edge cases pile up until someone just does it manually again.

This is tool-first thinking. You picked the hammer, then went looking for a nail.

Isenberg's framework flips the sequence. You don't open a workflow builder until you've done the diagnostic work: on paper, with a pen, away from your screen. The goal isn't to automate. The goal is to understand your operation so well that the automation target becomes obvious and undeniable.

This sounds simple. It is simple. Almost nobody does it.

Step 1: Map the Whole Workflow, Every Step, No Shortcuts

Open a blank document or grab a legal pad. Write down every single step in your core business process, from the moment a potential customer first touches your world to the moment they pay you, and then through delivery and renewal.

Every step. Not the idealized version. The actual version. The one with the awkward manual handoffs, the Slack messages that trigger the next action, the spreadsheet someone updates every morning that nobody talks about.

Be specific:

"Lead fills out Typeform," not "lead comes in"
"VA copies lead info into Google Sheet and sends Slack message to sales," not "lead gets routed"
"Founder sends proposal PDF via email, follows up twice manually," not "proposal sent"

You're building a map, not a marketing funnel. Include the ugly parts. Include the steps that feel embarrassing. Those are usually where the biggest opportunities hide.

If you run a client-facing operation, map the client's experience and your internal process side by side. The gaps between what the client sees and what actually happens backstage are where automations create the most leverage.

Step 2: Find Where Money Changes Hands

This is Isenberg's sharpest insight and the one most operators skip. Once your map is complete, circle every point where money moves, or where a decision directly causes money to move.

That includes:

Deposits and first payments: the conversion moment
Invoicing and payment collection: where cash actually arrives
Pricing negotiations: where margin gets won or lost
Upsells and renewals: where LTV compounds
Refund requests and churn triggers: where money walks out the door

These are your highest-leverage automation targets, period. Not because they're the easiest to automate (they're often not) but because any improvement here directly changes your revenue. A 15% improvement in proposal-to-close conversion rate dwarfs saving 20 minutes on a report nobody reads.

If your first automation idea doesn't touch one of these money-movement points, question whether it's worth building at all.

Step 3: Identify the Mechanical Repetition

Now go through your map a third time. Highlight every step where someone does essentially the same thing, the same way, every time it occurs. Not judgment calls. Not creative decisions. Pattern execution.

Common examples:

Copying data from one app to another
Sending the same follow-up email with minor personalization
Generating a report from the same data sources in the same format
Updating a CRM after a call based on the same criteria
Routing a request to the right team member based on simple rules

These tasks look like "work" on a timesheet, but they're actually just unautomated processes disguised as human labor. The person doing them isn't applying expertise. They're being a slow, expensive, error-prone API connector.

The overlap is where you build. When a mechanically repetitive step also sits near a money-movement point, you've found your automation target. That's not a guess. That's a business case.

Step 4: Quantify the Pain Before You Build Anything

Before you open StickyFlow or any other tool, run this formula on each candidate automation:

(Hours spent per week) x (Hourly cost of the person doing it) x 52 = Annual cost of the manual process

Then estimate:

Error rate: How often does this step break, cause a customer complaint, or require a redo?
Revenue drag: Does slowness here delay payments, lose deals, or increase churn?
Opportunity cost: What would the person doing this task produce if they weren't doing it?

Stack-rank your candidates by total pain, not by how fun they'd be to build. The automation that saves your ops manager 6 hours a week on invoice reconciliation near a payment trigger is almost always worth more than the one that auto-generates social posts.

Write down the number. You'll need it later to measure whether the automation actually delivered.

Step 5: Run It Manually First

This is the step that separates operators who build durable systems from those who rebuild the same automation every quarter.

Before you automate your top-ranked workflow, run it manually, with intention, for at least one full cycle. Document every decision point, every exception, every moment where you had to improvise.

Why? Because the edge cases that break automations are invisible until you do the work yourself:

What happens when a client sends the deposit but the invoice doesn't match?
What happens when the form submission is missing a required field?
What happens when two steps that usually happen sequentially happen simultaneously?

Running it by hand creates the spec your automation actually needs. Skip this, and you'll build for the happy path, then spend weeks patching failures you could have predicted in an afternoon.

The StickyFlow Workflow Audit Template

Here's the stripped-down version you can complete in under an hour. Grab a doc and fill in each section:

Full Workflow Map: List every step from first lead touchpoint to payment received to delivery complete. Include internal handoffs, tools used, and who owns each step. Aim for 15 to 30 line items.
Money Movement Points: Circle every step where money changes hands or a decision directly impacts revenue. Star the top three.
Mechanical Repetition Log: Highlight every step that's done the same way every time. Note frequency (daily, per-client, weekly).
Overlap Targets: Identify where money-movement points and mechanical repetition intersect. These are your build candidates.
Pain Quantification: For each overlap target, calculate annual cost, error rate, and revenue drag. Rank by total impact.
Manual Run Notes: Execute your top candidate manually for one full cycle. Document every exception, decision branch, and edge case.
Automation Brief: Write a one-paragraph description of what the automation should do, what triggers it, what output it produces, and what success looks like in dollars.

When you finish step seven, you don't have a vague automation idea. You have a build spec with a business case attached. That's when you open StickyFlow, not before.

The operators who treat this audit as a recurring ritual, monthly or quarterly, don't just build better automations. They stop building the wrong ones entirely. And in a landscape where everyone has access to the same AI tools, that discipline is the actual edge.