Where the money goes
Spend, share of budget and reported ROAS by subplacement. Select a row to investigate.
Does performance hold up?
Compare each platform and subplacement within your current filters. Start with the six largest by spend, or choose your own.
Applies to this chart and its table only, within the dashboard date range. Intervals start on your selected date; the final interval may be shorter.
Which placements work better — and why?
Default order prioritizes spend. Click a placement for campaign and weekly evidence.
Methodology, sample sizes & data coverage
Volume is evidence, not a guarantee
A placement needs the selected quote minimum and at least 14 days with spend to qualify for an evidence-based comparison. A stronger-volume label requires at least max(500, 5 × the quote minimum) quotes and 28 active days. These are adjustable decision rules, not statistical confidence intervals. Eight quotes never qualify under any available setting.
“Test more budget” also requires ROAS at least 5% above the benchmark, at least two eligible weeks, and outperformance in at least two-thirds of eligible weeks. Each eligible week needs at least 20 quotes in that placement and 3 days with spend. “Review spend” uses a 10% ROAS shortfall and the same weekly consistency rule. These thresholds are heuristics, not profitability targets.
Weighted comparisons
ROAS = reported quote value ÷ spend covered by value reporting. Value per quote = reported value ÷ quotes covered by value reporting. Cost per quote = spend ÷ quote events. We sum the underlying amounts before dividing. Value coverage below 95% blocks recommendations. The benchmark holds the selected dates, insurance product and campaigns constant, but includes every placement, even when you filter or exclude Audience Network.
Optional volume-adjusted ROAS = benchmark + [quotes ÷ (quotes + evidence floor)] × (observed ROAS − benchmark). This transparent ranking heuristic tempers tiny samples; it is not a forecast or a statistical estimate of future returns. Recommendation gates still apply.
Why a placement performs differently
Reasons decompose quote cost and quote value against the benchmark. Campaign, creative, audience and auction mix can explain the differences; these are observed associations, not causal placement effects. Drill into a placement to compare campaigns. Shifting spend requires a controlled test.
Scope and exclusions
Daily campaign × publisher platform × platform position data, over the available date range, account timezone America/Los_Angeles. Boundary months may be partial. Weekly buckets start Monday and are clipped to the selected range. Spend/day divides by all calendar days in the bucket, including zero-spend days. No rows means no reported delivery, not a fabricated zero-value conversion.
Audience Network “Classic” is Meta's returned position bucket; this extract does not separate native, banner and interstitial inventory further. Quote events are Auto 26414218331497189 and Home 527662590217855. Meta quote values are not verified sales revenue or accepted quotes. Attribution uses account settings and impression-time reporting; recent results can revise.
Excluding Audience Network recomputes the historical mix only. It does not simulate where that spend would go or predict incremental returns. Date changes read the last successful report. Automatic refresh runs every six hours. Use Refresh Meta data to request an additional pull.