How I Audit a DV360 Account in Under Thirty Minutes

When a client hands over a stalled account, they want two things in the first meeting: what is wrong, and how quickly it can be fixed. Not a strategy deck. Not a quarterly plan. A diagnosis.

Over the last few years I have compressed that diagnostic into a repeatable thirty-minute pass. It fits neatly into the gap between one client call and the next, and it tells me — with maybe eighty percent confidence — where the account is actually leaking.

Here is the audit, in the order I run it.

1. Delivery pacing (5 minutes)

Open every active insertion order and check the Delivery column against the Days remaining column. Anything under ninety percent pacing at the mid-flight mark is a red flag, and anything over one hundred and ten percent is worth investigating too — over-pacing often means bids are set too aggressively and the account will run out of budget early.

Over-delivery is not a win. It is a budgeting failure with a friendlier name.

At this stage I am not trying to fix anything. I am just building a list of IOs that need attention.

2. Frequency caps and audience reach (7 minutes)

The two most common causes of quiet under-delivery are:

• Frequency caps too tight — a cap of two per twenty-four hours can filter out huge chunks of the addressable pool by mid-morning.
• Audience segments too narrow — a first-party segment intersected with a third-party segment intersected with a device filter often leaves a delivery pool of five hundred users a day.

For each flagged IO, I check the Reach & frequency report in DV360. If unique users delivered is far below the estimated reach, the caps or the audience logic is the culprit — not the bid.

3. Creative rotation (5 minutes)

In Creatives → Performance, sort by impressions descending. If one creative is taking more than sixty percent of impressions in a multi-creative line item, DV360’s optimiser has quietly picked a winner and is starving the others. Sometimes that is fine. Often, the “winner” is fatigued and its click-through rate is trending down week over week.

Force even rotation for a control period, and let the numbers tell the truth.

4. Floodlight and conversion attribution (8 minutes)

This is where most audits find real money.

• Are Floodlight tags firing on the right pages?
• Are custom variables passing through correctly, or arriving as undefined?
• Is the conversion window still set to the default thirty days, when the client’s own CRM says the median lead-to-close is closer to nine?

I keep a small snippet handy for quick sanity checks in the browser console:

// Check what Floodlight sees on this page
console.log(window.__floodlight_variables__ || 'no floodlight object found');

Bad tags do not announce themselves. They just quietly attribute the wrong conversions to the wrong campaigns until someone gets fired.

5. Bid strategy sanity check (5 minutes)

Only now — with pacing, audience, creative, and tracking checked — do I look at bids. Specifically:

1. Is the bidding strategy aligned to the KPI the client actually cares about? A CPM target when the goal is CPA leads to predictable disappointment.
2. Is the Bid multiplier stack doing anything useful, or is it a decade of well-meaning tweaks that cancel each other out?
3. Is the Optimise for viewability toggle on, when it does not need to be — and paying a twenty percent premium for it?

Wrapping up

At the end of thirty minutes I have a one-page diagnostic that reads something like:

• Two IOs pacing under sixty percent — frequency caps and audience narrowing.
• One creative dominating impressions across three IOs — pause and rotate.
• Floodlight custom variable order_value returning undefined on the checkout page — engineering fix.
• Bid multiplier stack has redundant device and geo layers — simplify.

None of this is glamorous. It is not the deck the client was expecting. But it is what the account actually needs. And the good clients — the ones worth working with — always prefer a diagnosis to a strategy.