Choosing the Right Audit Timeframe

For · Operators about to run an audit, deciding on the date range
Time · 5 min read
Next · Testing timelines and evaluation windows

When you run a Catalyst Audit, one of the first decisions is the date range. Should you look at the last 30 days? 90? Something custom?

The answer depends on what you’re trying to learn. Each timeframe reveals different signals, and choosing the right one is the difference between a snapshot and a story.

The Short Version

• Choose your window and your rhythm from your conversion volume, not from the calendar. What decides whether an audit can tell you anything is how many conversions sit in the period it reads.
• Run the next audit once enough new conversions have accumulated to support a call. For a high-volume account that arrives quickly; for a lower-volume one it takes longer, and running early simply returns held findings.
• The history gate advances by audit count, not by days in the window or by how long you have been a customer. Three audits open projections; six open the full picture. How fast you get there is set by how quickly you accumulate conversions.
• Run 7 or 14 days only to check that a recent change did not break something. A short window cannot tell you whether a bid strategy is working.
• Reach for 90 days in two specific cases: a low-volume account that needs to accumulate conversions, or a review timed to a seasonal boundary.
• The cap is 90 days on purpose. Every audit already layers in trend summaries, a year-over-year baseline, and changelog context, so a longer raw window adds noise rather than range.

Key Terms

How Trellis Approaches Your Data

Think of your ad account like a patient’s medical chart. A doctor doing a check-up doesn’t request every record from the last decade. They pull the last three months of vitals, then reference a one-page summary of your history. That’s how a Catalyst Audit works.

Trellis pulls detailed, campaign-level performance data for the period you select. We analyze that data (spend, conversions, CPA, ROAS, search terms, Quality Scores, and more) alongside your actual gross margin, attribution data, and a historical baseline. Statistical models and rule-based methods run first, producing structured evidence and constraint flags. A language model then synthesizes these pre-computed findings into the narrative report, writing within the constraints the methodology has already established, not guesswork.

The result is an audit tuned to your business, not a generic report.

Available Timeframes

Last 7 Days and Custom 14-Day: Early Warning Checks

Short windows serve one purpose: monitoring recent changes. If you adjusted a bid strategy, launched new ad copy, or paused a campaign, a 7- or 14-day audit tells you whether the change is moving metrics in the right direction, or breaking something.

Google documents the Smart Bidding learning period as 7-14 days after a significant change. A 7-day check captures the initial response. A 14-day custom window (select “Custom Range” and set your dates) covers two full weekly cycles, which removes day-of-week bias and spans the entire learning period.

What short windows can tell you:

• A sudden CPA spike or conversion drop after a change, an early signal to investigate
• Whether budget pacing shifted after a bid strategy switch
• If a new ad variant is getting impressions and clicks at expected rates

What short windows cannot tell you:

• Whether a trend is real or just noise; trend detection requires multiple audit cycles, not a longer single window
• Whether a bid strategy change is “working”: Google and Microsoft both recommend waiting at least 6 weeks before judging automated bidding performance
• Profitability conclusions. Conversion samples this small typically fall into the INSUFFICIENT estimation tier, where COGS-adjusted recommendations are gated

Think of it like checking your temperature the day after starting a new medication. A fever tells you something needs attention. A normal reading tells you nothing went wrong. Neither tells you whether the medication is working; that takes weeks.

When to use: Run a 7-day audit within the first week of a significant account change. If something looks off, investigate immediately. If metrics look stable, wait for your next 30- or 90-day audit for the full picture.

Last 30 Days: The Common Middle

Your routine health check. A 30-day audit catches:

• Wasted search terms draining budget without conversions
• Attribution gaps: how platform-reported conversions compare to actual orders
• COGS-adjusted profitability: which campaigns earn real profit after COGS, not just revenue

A 30-day window suits an account that accumulates a comfortable number of conversions inside a month. That is the test to apply, rather than the calendar: if 30 days reliably carries your campaigns above the evidence floor, it is the right window for you. If it does not, reach for 90 days instead, and if 30 days puts you well clear of the floor you can read a shorter window and audit more often.

Catalyst Audit analysis compounds with each audit, and the history gate advances on audit count rather than on how long each window is or how long you have been a customer. Three audits open projections. Six open the full picture, including confidence-graded projections, control charts, and budget impact modeling. Your conversion volume is what sets the pace along that path, because an audit run before enough conversions have accumulated returns held findings and does not advance anything worth having.

A single audit gives you one data point. Six give you a trend, a calibrated baseline, and models tuned to your account.

Last 90 Days: Quarterly Deep Dive

A 90-day window is the right choice in two situations: accounts that need the extra time to accumulate conversions above the evidence floor, and reviews timed to a seasonal boundary.

• Statistical confidence for low-volume accounts. Google’s documentation recommends 30 conversions per month for Target CPA and 50 for Target ROAS. Microsoft Ads requires 30 conversions in 30 days for automated bidding to optimize reliably. If your account’s monthly conversion volume falls below the estimation-tier floor where Catalyst Audit can act, a 30-day window may not produce enough data to act on; a 90-day window accumulates enough conversion data to cross that bar.
• Bid strategy evaluation. Google recommends at least 6 weeks to evaluate a Smart Bidding change. A 90-day window captures the full evaluation period plus stabilization, enough to tell you whether that switch from Manual CPC to Target ROAS actually worked.
• Seasonal transitions. A quarter captures at least one seasonal boundary (winter to spring, summer to fall), revealing how demand shifts affect your campaigns.
• Budget pacing patterns. Three monthly cycles in a single window reveal whether you’re systematically under- or over-spending relative to your targets.

A 90-day Catalyst Audit also pulls in historical trend context and a year-over-year baseline comparison from summarized data outside the audit window. This layered approach gives you the depth of a long lookback without the noise.

Month to Date

Covers the current calendar month so far. Helpful for checking pacing mid-month.

Previous Month

The full prior calendar month. A clean 28-31 day window useful for month-over-month comparison.

Custom Range (up to 90 days)

Pick your own start and end dates, capped at 90 days. Use this when you need a specific window: for example, isolating the four weeks before and after a campaign restructure, or measuring performance during a holiday promotion.

Why 90 Days Is the Cap

You might wonder: if 90 days is good, wouldn’t 180 or 365 be better?

Not necessarily. Here’s why.

More data doesn’t always produce better analysis. When you pour too much information into a single analysis pass, critical findings get buried under volume. Research on analytical systems consistently shows that curated, well-scoped data produces higher-quality output than exhaustive datasets. It’s the same principle behind a focused lab panel versus ordering every test in the catalog; precision beats volume.

Old search term data adds noise, not clarity. Search queries from nine months ago reflect a different competitive landscape, different seasonal demand, and potentially different campaign structures. Including them dilutes the findings that actually drive your next decision.

Mixing seasons distorts your averages. Combining July performance with January performance in the same analysis produces misleading numbers. July and January represent fundamentally different demand environments for most businesses. Averaging them tells you nothing useful about either period.

Trellis already delivers long-range context without requiring a long date range. Every Catalyst Audit automatically layers in three additional data sources beyond the raw performance window:

• Historical trend summaries spanning up to 180 days from the datamart (Trellis’s first-party context layer). These show the trajectory of key metrics (CPA, ROAS, conversion volume) month over month, so even a 90-day audit reveals whether your account has been improving or declining over the past half-year.
• Year-over-year baseline comparison using statistical models that calculate the probability that current performance has meaningfully changed from the same period last year.
• Changelog context that tracks what changed in your account (bid strategy switches, budget adjustments, paused campaigns) and when those changes happened. This connects performance shifts to their likely causes.

This layered architecture means a 90-day audit already delivers roughly 80% of the insight a 365-day raw data dump would provide, with significantly higher analytical quality.

A Note on Low-Volume Accounts

This is the clearest case of the rule that governs the whole article: the window should be as long as your conversion volume requires, and no longer. If your campaigns generate fewer conversions per month, the 90-day window becomes more important, not less. Here’s why:

Both Google and Microsoft define practical minimums for their automated bidding to work:

A low-volume campaign produces only a handful of data points in a 30-day audit, often too few to support a call. That same campaign accumulates enough conversions over 90 days to cross the floor where Trellis can apply its models with meaningful confidence. For these accounts a 90-day window is often the only one that produces something you can act on, and running more frequently in the meantime returns held findings rather than earlier answers.

The same logic runs in the other direction. An account clearing the floor comfortably inside a month can read a shorter window and audit more often, and each of those audits advances the history gate exactly as much as a slower one would.

Sources: Google Ads Help: Smart Bidding Learning Period, Google Ads Help: Smart Bidding with Shopping/Performance Max, Microsoft Learn: Budget and Bid Strategies

Choosing Your Timeframe: A Quick Guide

Best Practices

1. Set your rhythm by conversions, not by the calendar. Look at how long your account takes to accumulate enough conversions to clear the evidence floor, and make that interval your audit rhythm. This is how Catalyst Audit compounds: the models sharpen, the history gate advances, and baseline comparisons gain precision with each audit. Auditing faster than your conversions accumulate does not speed that up, because findings below the floor are held rather than reported.
2. Run a 7-day check after any significant account change: bid strategy switches, budget shifts, or campaign restructures. Treat it as a smoke test, not a verdict.
3. Extend the window when the floor is the constraint. When 30 days does not carry your campaigns above the estimation-tier floor, extend to 90 so enough conversions accumulate to support a call. Also appropriate when you want a seasonal transition inside a single window.
4. Use custom ranges when you need to isolate a specific period, like a holiday promotion or the weeks following a major campaign change.
5. Don’t chase longer windows for their own sake. The Catalyst Audit’s layered context (trend summaries, baselines, and changelogs) already provides the long-range perspective. Your selected timeframe controls the detailed, campaign-level analysis. The history gate advances by audit count, not by days in the window.

What’s Next

• Testing Timelines and Evaluation Windows: how long each type of account change needs before you can draw meaningful conclusions
• Reading Your Audit Report: how to read the analysis and act on recommendations
• Business Context and Your Economics: how your margins shape every recommendation