Why are you even tracking this?

#028

Read Time: 3 minutes

In the world of AI everything, just remember…

"You can't vibe code your way out of bad data."

Last week I sat down with Tiffany Gonzalez

Revenue Manager @ Dell
Head of Revenue @ Microsoft
Head of BizOps @ AWS
CRO @ Coupang
And now, CCO @ GraphIQ

She dropped that quote last week and I can’t stop thinking about it.

Here’s 5 more lessons I learned from her worth stealing this week

👇

1. Find your "moment that matters"

Most RevOps teams track activity. Calls logged, emails sent, dashboards refreshed. Tiffany calls this reactive revenue. You're measuring what already happened.

The alternative = find the earliest moment in your customer's journey where they get it. Not where product thinks they should get it. Not where marketing wants them to get it. Where the customer actually says "oh, I see the value."

Slack's early growth metric wasn't signups or team plans. It was… did this person invite 4 people and send 300 messages? That individual behavior predicted retention better than anything on a contract.

The Salesforce translation = If your org can't answer "which early customer behavior predicts renewal," that's the report to build. Before the 14 others in your backlog.

2. Run the "So What" test on every dashboard

Tiffany is famous for asking this 1 question for every team she's ever led:

"So what happens if I stop tracking this?"

Does any decision change?

Does anyone's Tuesday look different?

If the answer is no, kill the report.

Most inherited Salesforce orgs are graveyards of reports someone needed once, in 2021, for a QBR that no longer exists. Every one of them is noise between you and the number that matters.

3. Revenue is iterative. Your reporting should be too.

The dashboard that saved you last quarter might be dead weight this quarter. Once a problem is solved and absorbed into the business, it's no longer the thing blocking revenue. Stop maintaining it out of habit.

4. AI = models + data (and you only have one of them)

Everyone has access to the same models. Claude, GPT, Gemini. The differentiator is whether your data gives those models anything to reason over.

This reframes CRM hygiene entirely. Clean data used to be about accurate forecasts. Now it's about whether AI can actually work inside your business. A Salesforce org full of stale opportunities and 42 required fields nobody fills in honestly isn't helping anyone.

And no, asking an LLM "what fields should go on my Account object" won't save you. You'll get generic bloat you'll pay someone (possibly me) to rip out in 12 months.

5. Do the work manually before you scale it

Tiffany's advice to founders = run your process by hand first. Yes, even in a spreadsheet. Understand exactly what your sales motion needs at a discrete level, then build it into your CRM.

There's no playbook you can copy, because nobody else has your customers or your motion. The teams with the worst Salesforce implementations skipped this step. They bought the tool, imported someone else's blueprint, and wondered why nobody used it.

Some doors are one-way. Pause before you walk through them.