S3 · Credit Governance — budgets, protocols, and the one-person problem

(paid-course opener: orient → stakes → in) Okay, this video is about credit governance. Who gets to spend, under what rules, and what happens when nobody decides. And remember the company from the credit video, the one sitting on a six hundred thousand credit package that only ever used a hundred and fifty thousand? Here's the part of that story I didn't tell you. Of everything that company did spend, ninety-five percent went through one single seat. Eight people had access. Seven of them never really touched the tool. One person, untrained, running random enrichments with no plan, burned almost all of it.

And here's the detail that should actually worry you: that's not a settings problem you can toggle away. Clay has no per-user credit cap. None. There is no switch in the app that stops one seat from spending the whole balance. Spend is governed by how tables are built and the protocols your team follows. That's it. So the protocols are the product of this video.

(Master map: Strategy Track highlighted.) So if the app won't enforce limits, what does? Three layers, and every one of them is behavior, not settings.

  LAYER 1 · HABITS      how every table gets BUILT (the 8 habits)
  LAYER 2 · BUDGETS     each project gets a number BEFORE it starts
  LAYER 3 · THE RITUAL  big runs get a second pair of eyes BEFORE they fire

Habits govern the tables. Budgets govern the projects. And the ritual governs the moments that actually destroy balances, the big runs. Let's take them in order.

These eight habits come straight from the field. Every one of them exists because somebody paid for the lesson. (the habits stamp on one at a time)

  1 · AUTO-UPDATE OFF while you build     6 · ORDER WATERFALLS BY COST,
  2 · RUN CONDITION on every paid column      not by brand
  3 · FILTER AND DISQUALIFY before        7 · DEDUPE before enriching
      you enrich                          8 · WATCH THE SYNCS,
  4 · LOOK IT UP before you buy it            not just the lookups
  5 · GET THE PROMPT RIGHT before it runs

One, auto-update off while you build. Left on, every row that lands in the table enriches immediately, mistakes included. Switch it on once the table is finished and behaving. This single toggle prevents most accidental overspend, including the exact accident from the last video's cold open.

Two, a run condition on every paid column. Only run if this field is empty. That one line stops you re-buying data you already hold, because without it, a re-run charges the whole table again.

Three, filter and disqualify before you enrich. You know this one from the credit video. Logic is free, enrichment is not, so the cheap decisions happen first in the table.

Four, look it up before you buy it. A lookup pulls values you already own, from your CRM or another table, at no cost. Buying data you already own is the quietest form of waste.

Five, get the prompt right before it runs. Iterating a prompt against a live column re-runs and re-charges on every rewrite. Draft it, test it on ten rows, then release it.

Six, order waterfalls by cost, not by brand. A waterfall stops at the first match, so the sequence decides the bill. Cheapest plausible provider first, the expensive specialist last, where it only fires on what's left.

Seven, dedupe before enriching. Two rows for the same person cost twice and produce one usable record.

Eight, watch the syncs. An export firing on every change is an action every time. Tiny individually, material across a busy table. Sync on a schedule unless continuous is genuinely needed.

Now here's the governance move: these are not tips. In your team, these are policy. The difference between a tip and a policy is that a policy gets checked.

Layer two, budgets. Every project that touches paid columns gets a number before it starts. Not a precise forecast, a ceiling. You already know how to produce it: the calculator from the credit video, sixty seconds, rows times columns times cost. Write the number where the team can see it.

And then watch the actuals weekly. Your usage-by-table view is the whole monitoring stack you need at this stage. Ten minutes every Friday: which tables spent, against which budgets, and is any number moving faster than its project. A spend spike you catch in week one is a conversation. The same spike caught at renewal is a crisis.

Layer three, the ritual, and this is the one that prevents disasters instead of explaining them. Any run above your threshold, and pick a real number, say anything over a thousand rows or a meaningful slice of the monthly budget, gets a second pair of eyes before it fires.

The sign-off is four questions, and they take sixty seconds:

  PRE-FLIGHT SIGN-OFF · before any big run
  1 · What does this cost at FULL row count?
  2 · Is there a run condition on every paid column?
  3 · Is auto-update off, or supposed to be on?
  4 · Has this exact table run before, and what happened?

That's it. Four questions, out loud, with one other person. Both accidental disasters you've heard about in this course die at question one or question three.

And if you're solo, the ritual still applies, just differently: future you is the second pair of eyes. Write the four answers down, and for any genuinely big run, sleep on it. The run will still be there in the morning. The credits won't come back.

One more thing about that company, because the ninety-five percent is only half the tragedy. The other half is the seven seats that sat idle. Governance is not about distrusting the one person who builds. It's about enabling the seven who don't. Budgets make spending legible, habits make it safe, and the ritual makes it shared. Do that, and the one-person problem solves itself from both directions: the spender gets guardrails, and everyone else gets a way in. You bought eight seats of capability. Governance is how you stop renting one.

Here's a quick recap of what we learned in this video.

One: Clay has no per-user credit cap. Your protocols are the cap. Spend is governed by how tables are built and the rules your team follows, nothing else.

Two: three layers. The eight habits govern every table. Budgets govern every project, a ceiling before it starts and ten minutes of actuals every Friday. And the pre-flight sign-off governs the big runs, four questions, sixty seconds, one other person. Solo, future you signs off, and you sleep on the big ones.

Three: governance is enablement, not distrust. It's how eight seats become eight builders instead of one spender and seven spectators.

The Governance Protocol under this video has all of it: the eight habits written as policy, the sign-off card, and the budget-and-alert setup.

Your exercise is under this video, and it's not scored. Run the workspace audit on your own instance: usage by table, auto-update settings, run conditions on paid columns. Name your top three red flags and write one fix for each. Thirty minutes, and you'll know exactly where your balance is exposed.

Next up: auditing anyone's work. How to open any table, a vendor's, a colleague's, or your own from three months ago, and review it properly in fifteen minutes.