TheCFO'sGuidetoQuestioningYourPPCAgency'sPOASReporting

Most CFOs don't know how to read a PPC agency report critically. The reports are designed for marketing teams, not finance teams — they speak in attribution, conversion windows, blended ROAS and impression share, not contribution margin, cash impact and P&L line items. This article is a 12-question diagnostic any CFO can apply to any agency report in 20 minutes. If your agency can't answer cleanly, you have your answer about whether they understand your business — and whether their reporting is built to surface commercial truth or to obscure it.

For context on POAS as a metric, start with our POAS framework explainer. For the operational framework underneath good POAS reporting, see BOI™ explained.

Why agency reports are designed for marketing teams, not finance teams

The disconnect isn't accidental. PPC agencies report in the language they're paid to deliver against — usually a marketing KPI like ROAS, impression share or conversion volume. Their commercial incentive is to make these numbers look good, defend them in QBRs, and use them to justify retention.

CFOs care about different things. Whether the marketing spend is contributing to profit. Whether ad budget is generating positive cash impact this quarter. Whether the campaign mix is supporting or undermining the contribution margin trajectory. These questions live in the P&L. The agency report rarely touches them.

The result: in many ecommerce businesses, the marketing team and the finance team look at completely different dashboards and reach opposite conclusions about the same campaigns. The marketing team sees ROAS up YoY and declares victory. The CFO sees contribution margin compressing and asks where the disconnect is. Nobody has the diagnostic framework to bridge the two views.

This article gives the CFO the diagnostic framework. Twelve questions you can take into your next agency review meeting. If the answers don't hold up, you have a structural reporting problem to solve before any other commercial decision can be made cleanly.

The 12 questions — and what good answers look like

1. What was our POAS last month — net of all variable costs including payment fees, shipping, returns and discounts?

A good answer is a specific number with the calculation shown. "POAS was 1.8:1 — that's £180,000 of true contribution margin divided by £100,000 of ad spend. Contribution margin is calculated as revenue minus COGS, shipping, payment fees, returns processing and discount value."

A bad answer is "we don't report POAS, we report ROAS — that was 5.2x." This means they are not factoring in the variable costs that determine whether the campaign is actually profitable.

A worse answer is "POAS was 3.1x" without showing the calculation. This usually means they are using Shopify's gross profit number, which excludes payment fees, shipping and return costs. The "POAS" is overstated by 15–40% in this case.

2. Can you show POAS by campaign? Which campaigns were profitable and which were not?

A good answer is a table with one row per campaign and POAS in a column. The agency should be able to point to specific campaigns that delivered above-target POAS and specific campaigns that delivered below.

A bad answer is "POAS isn't reliable at campaign level because of attribution issues." Attribution is real, but it's not an excuse for not having the data. There are ways to estimate. If they can't or won't, they don't have the operational infrastructure to manage commercially.

3. What percentage of our Performance Max spend last month went to branded search terms?

A good answer is a specific percentage, ideally below 15%. The agency should have visibility into where PMax is spending via scripts, search term insights, or audit tooling.

A bad answer is "we don't have that data" or "PMax doesn't break it out." Both are factually wrong. Industry tools exist to surface this. The typical answer from agencies running PMax on full automation without brand exclusions is 30–50%, which represents money the brand is spending to acquire customers it already has.

4. Are our conversion values revenue-based or profit-adjusted?

A good answer is "profit-adjusted via offline conversion imports" or "margin-adjusted via conversion value rules." The agency should be able to explain how the margin signal flows into Smart Bidding.

A bad answer is "they're revenue-based — that's the standard setup." This is technically true but commercially wrong. Smart Bidding optimising against revenue systematically over-rewards low-margin SKUs. The agency hasn't done the work to feed margin data into the bidding signal.

If the answer is "we report on POAS but bid on revenue-based ROAS," the reporting is decorative. The bid decisions are still being made on the wrong signal.

5. What's our true contribution margin per order, and how does it compare to what Shopify reports?

A good answer is two specific numbers and a percentage gap. "True contribution margin per order is £18, Shopify reports £24 — Shopify overstates by 33% because it excludes payment fees, shipping and return costs."

A bad answer is "we use Shopify's reported gross profit." This means they haven't calculated true contribution margin — and all their POAS analysis is built on overstated profit figures.

6. Which of our SKUs have negative POAS, and what's our plan for them?

A good answer is a list of specific SKUs with POAS values and a stated action — suppress from paid promotion, reassign to Liquidation, investigate root cause (margin, return rate, conversion rate).

A bad answer is "we don't see anything below break-even" or "we manage at campaign level, not SKU level." The first is unlikely to be true. The second is a structural reporting choice that hides commercial reality.

7. What's our average CPC trend over the last six months, and how does it compare to category benchmarks?

A good answer is a number with a trend and context. "Average CPC is £1.42, up 18% from this time last year, broadly tracking the category increase per WordStream's benchmarks."

A bad answer is the number without trend or context. Or worse, no answer — meaning the agency isn't tracking competitive dynamics affecting your business.

8. What's the cannibalisation rate on our branded search? How much of our PMax / Shopping conversions would have happened organically?

A good answer is a specific percentage with the methodology. "Branded cannibalisation is approximately 22% — we measure by holdout testing two weekends per quarter and comparing organic conversions in test vs control regions."

A bad answer is "all our branded conversions are incremental" or "we don't measure that." The first is improbable. The second is choosing not to know.

9. What's our customer lifetime value, and how is it factored into our bidding signals?

A good answer covers both halves. "Our LTV is approximately £140 for new customers acquired through paid search, weighted across cohorts. We feed predicted-LTV into Google Ads conversion values via offline imports, which lets us bid up to higher first-order CPC for customers who will deliver lifetime profit."

A bad answer is "we don't track LTV" (the brand has chosen not to know its own unit economics) or "LTV isn't relevant to bid signals" (it's the most important signal for any subscription or repeat-purchase business).

10. When did you last audit our feed health, and what's the current disapproval rate?

A good answer is a recent date (within the last 30 days) and a low percentage (below 5% disapproval). The agency should be doing this weekly.

A bad answer is anything over 60 days old or any disapproval rate above 10%. Feed health drift is the single most overlooked source of Shopping performance erosion.

11. What changed in our account last month — what was paused, what was scaled, what was restructured?

A good answer is a specific list of changes with rationale for each. "Paused asset group A in PMax because it was 40% branded; bid up 12% on premium SKUs based on margin analysis; restructured Shopping into three priority levels."

A bad answer is "nothing significant" or "the algorithm makes most decisions." The first means they aren't actively managing. The second is true to a point, but the strategic-level changes (campaign structure, audience signals, bid targets) should be ongoing.

12. If we wanted to leave your agency tomorrow, what would the transition look like, and what would we own?

A good answer is comprehensive and reassuring. "You own your Google Ads account and MCC access. We'd hand back full admin within 24 hours. All scripts and historical data stay with you. The 30-day notice period is unconditional. We've documented everything in the shared drive you have access to."

A bad answer is anything evasive. "The account is held in our MCC for performance reasons." Or "there'd be a longer notice period given the strategic dependencies." Or "we can't transfer the bidding strategies, they're proprietary." Each of these is a red flag — the agency is making themselves harder to leave, which means they expect to need to.

Red flags in agency language

Specific phrases that indicate the agency is not running your account commercially:

"Industry-standard ROAS target." ROAS targets are category-specific and margin-specific. Any agency citing "industry standard" without that context is using a default rather than thinking about your business.

"The algorithm is making most of the decisions now." True in a sense — Smart Bidding has improved — but said in this tone usually means the agency has stopped doing the strategic work.

"We'll get the data on that for next month's report." If a CFO question requires another month of data preparation, the reporting infrastructure is too thin.

"That's hard to measure with all the attribution changes." Attribution is genuinely hard. But it's not a complete blocker. Agencies using attribution challenges as a reason to not measure are choosing not to measure.

"POAS is one of many metrics we look at." Translation: we don't optimise against it, but we'll show it on the report when asked.

"We work backwards from your revenue targets." Revenue is not profit. Working backwards from revenue targets is the structural cause of "ROAS up, margin down."

What weekly Loom updates should actually contain

If your agency does weekly written updates (or Loom video updates, which we recommend), here's what should be in them:

• POAS performance vs target — by campaign, with explanation of any variance
• Top 3 SKUs by margin contribution this week
• Bottom 3 SKUs by POAS this week with action taken
• Spend allocation changes vs the previous week
• Bid signal changes if any (always with rationale)
• Feed health — disapprovals fixed, new issues raised
• One thing we're watching for next week (forward-looking)

If the weekly update doesn't contain at least items 1, 4, 5 and 6 — it's a status update, not a management report.

When to switch agencies

Three patterns that indicate a switch is overdue:

The CFO has asked the same question three times across three months and not gotten a clear answer. Means the reporting infrastructure is structurally unable to surface the answer. The agency would need to rebuild their data layer to fix it — which they usually won't.

POAS has been flat or declining for two quarters while spend has been rising. Means the agency is comfortable scaling spend without scaling commercial outcome. Their incentives are different to yours.

The agency cannot articulate a clear, specific plan for the next quarter that's distinct from the last quarter. Means they're maintaining rather than managing. Maintenance is fine if the account is performing exceptionally. If POAS isn't where you want it, maintenance is not enough.

How to manage the transition without losing data

If you decide to switch:

Get full admin access to your Google Ads MCC before you announce the switch. The account should be in your ownership, not the agency's.

Export all historical data — campaigns, scripts, custom reports, Looker Studio dashboards — before changing access.

Document the current setup in writing — campaign structures, bid targets, conversion tracking implementation. Useful for the next agency's onboarding.

Notice period management — keep the outgoing agency running operationally while the new agency rebuilds in parallel. Don't switch with a hard cutover.

30-day parallel audit — have the new agency audit the account while the old one is still running. Captures opportunities and mistakes before changeover.

We've documented our own approach to agency transitions in our Spend Governance service page.

Next steps

Want a finance-grade audit of your current PPC agency's reporting and operational discipline? We run independent audits for ecommerce brands considering an agency switch, and provide a written report you can use to make the decision either way.