Reading Your Audit
Audit Evidence and Citations
Every recommendation cites its evidence. Here is how Catalyst Audit sources, vets, and applies industry references.
For · Operators wanting to verify the numbers their audit cites
Time · 5 min read
Next · Business context and your economics
When a Catalyst Audit recommends a change to your campaigns, it doesn’t just say “trust me.” Every recommendation cites its evidence, labels every claim so you know where it came from, and includes the case against the recommendation so you can make an informed decision.
The Short Version
- Every recommendation must be backed by at least two measured data points from your account. If the data isn’t there, Catalyst Audit says so rather than guessing.
- Every number in the audit is labeled: [FACT] for measured data, [PROJECTED] for estimates with stated assumptions, [INFERRED] for pattern recognition, and [INSUFFICIENT DATA] when the sample is too small to act on.
- Each recommendation names a credible alternative interpretation and the specific condition that would flip the call, along with a risk score and a monitoring plan with specific thresholds.
- Estimation tiers gate what Catalyst Audit is willing to recommend based on how much conversion data is available.
- References come from ad platform documentation, verified industry research, and your own account’s historical performance.
Data Labels: Know What You’re Looking At
Every number in a Catalyst Audit carries a label that tells you how much weight to put behind it:
[FACT]: This number came directly from your ad platform or your order data. It’s measured, not estimated. CPCs, conversion counts, spend totals, and revenue figures are facts. You can act on these with confidence.
[PROJECTED]: This is a forward-looking estimate. “Pausing this keyword could save approximately $200/month.” The audit spells out the assumptions behind the projection and applies a conservative discount to account for real-world conditions that don’t match the model. Projections always show three scenarios: optimistic, expected, and conservative.
[INFERRED]: A pattern Catalyst Audit identified based on your data combined with known industry behavior. These always use ranges (“15-25%”) rather than exact figures because the precision isn’t there for a point estimate. Treat these as hypotheses worth testing.
[INSUFFICIENT DATA]: There isn’t enough information to draw a reliable conclusion. The recommendation is always the same: wait, collect more data, then revisit. Catalyst Audit will never invent a recommendation when the data doesn’t support one.
A good rule of thumb: act on [FACT]-tagged findings first, verify the assumptions behind [PROJECTED] estimates, and treat [INFERRED] insights as areas to watch on the next audit.
The Evidence Requirement
Every recommendation in a Catalyst Audit must pass a series of evidence checks before it appears in the report. These checks are built into the methodology, not applied after the fact.
At least two data points. A recommendation citing only one metric can be misleading. “Your CPA is $40” is a fact, but it’s not a recommendation. Catalyst Audit requires at least two measured data points to support any action. For example: “Your CPA is $40 [FACT] and your break-even CPA is $32.50 [FACT], placing this campaign in the marginal profitability tier.”
A credible alternative and a flip condition. For every recommendation, Catalyst Audit names the most defensible alternative read of the same evidence and the specific observation that would flip the call. This isn’t hedging; it’s intellectual honesty. If the recommendation is to pause a keyword, the credible alternative might be that the keyword is in a learning phase after a recent match type change; the flip condition might be that conversion volume recovers above 5 per week within the next 14 days. You decide whether the alternative interpretation outweighs the recommendation, and you have a named threshold to watch.
A risk score. Each recommendation is scored 1-5 based on how difficult it would be to reverse and what could go wrong:
| Score | Level | What It Means |
|---|---|---|
| 1 | Minimal | Easily reversible within minutes. Adding a negative keyword, minor bid adjustment. |
| 2 | Low | Reversible within 24 hours. Budget changes, match type adjustments. |
| 3 | Moderate | Requires a monitoring plan. Bid strategy changes, campaign restructuring. |
| 4 | High | Cite past precedent, specify safeguards. Multi-campaign changes. |
| 5 | Critical | Comparable to known incidents that caused significant performance drops. |
Recommendations scored 4 or 5 include specific safeguards and past account precedent where available.
A monitoring plan. Not “watch your CPA” (vague), but a specific, measurable plan: “If CPA exceeds $35 after 14 days, pause the campaign and revert to the previous bid strategy.” Every monitoring plan includes the metric to track, the threshold that triggers action, the date to evaluate, and the specific action to take if the threshold is breached.
Executable instructions. “Optimize your keywords” is not a recommendation. “In the Google Ads UI, navigate to Campaign X > Keywords tab > select keywords Y and Z > set bid to $0.90” is. Catalyst Audit recommendations include step-by-step actions you can execute directly.
Where References Come From
Catalyst Audit draws from three categories of reference material:
Your account’s historical performance. Baseline metrics from prior periods, recent account changes and their measured impact, and longitudinal trends all inform the analysis. A recommendation to change a bid strategy carries more weight when the audit can cite how similar changes performed in your account previously.
Ad platform documentation. Google Ads and Microsoft Ads publish guidance on bid strategy behavior, learning periods, conversion thresholds, and match type dynamics. Catalyst Audit references this documentation when evaluating whether a recommendation aligns with how the platform actually works. For example, automated bidding learning periods vary by strategy type, and the audit respects those documented constraints.
Verified industry research. Sources are logged with their publication date, the date they were referenced, and a summary of how they informed the analysis. Each source must meet one of three criteria: it contradicted an assumption, it helped choose between competing approaches, or it revealed a platform limitation that constrained the recommendation. Routine lookups and generic best-practice articles don’t qualify.
Estimation Tiers
Catalyst Audit gates its recommendations by the estimation tier the data supports. More conversion data unlocks tighter projections; less data narrows what the audit is willing to claim.
| Estimation Tier | What Catalyst Audit Will Do |
|---|---|
| Very High | Full analysis with statistical projections and narrow confidence ranges |
| High | Full analysis with probability statements and wider intervals |
| Medium | Analysis with conservative projections (optimistic/expected/conservative scenarios) |
| Low | Observation-only callouts for extreme signals |
| Insufficient | No recommendations. Data summary only. Everything tagged [INSUFFICIENT DATA]. |
As your conversion volume grows, the audit moves up this ladder: more data unlocks tighter projections; less data narrows what the audit is willing to claim. This gating exists because small sample sizes produce unreliable patterns. A campaign with only a handful of conversions might show a large CPA swing that is really just normal variance rather than a real problem. Catalyst Audit won’t tell you to act on noise.
What Catalyst Audit Won’t Do
Catalyst Audit does not make guaranteed claims. You will never see “this change will improve ROAS to 3.5x” or “budget increase will generate $5,000 more revenue.” Guaranteed outcomes violate the estimation methodology.
The audit does not extrapolate trends into the future. “Based on this trajectory, Q2 revenue will be $50,000” is the kind of statement that sounds precise but is built on assumptions that rarely hold. It sticks to what the data actually shows.
The audit does not silently proceed when conversion tracking accuracy is below 80%. It still produces its full set of recommendations, but the framing depends on which band the accuracy falls into. Between 60-80% (Warning), the audit raises a tracking_accuracy_degraded banner and attaches a [TRACKING CAUTION X%] caveat to each affected recommendation. Below 60% (Unvalidated), the banner escalates to error severity and the caveat becomes [TRACKING UNVALIDATED X%]. In both bands the recommendations still ship so you can apply judgment item by item while fixing the tracking gap at the source.
What’s Next
- Business Context and Your Economics: how your margins, seasonality, and CLV shape every recommendation
- Catalyst Audit vs. Platform Dashboards: what the audit adds beyond what you see in Google Ads or Microsoft Ads
- Reading Your Audit Report: how to read the analysis and act on recommendations