Minimum Viable Test Budget Per Campaign Before You Judge It
How to size a test budget with enough statistical weight to judge a Shopify campaign fairly, worked out from CVR, CPC and conversion volume math.
Written by Mantas Jurgutis — Founder, Adsify — builds the Google & Meta automation merchants use daily
Editorially reviewed by Adsify Editorial on April 1, 2026 — Reviewed against Shopify, Google Ads and Meta official documentation.
The most common budgeting mistake
Merchants routinely judge a new campaign after 3-5 days and $150-$300 spent, then kill it for a 'bad' 1.2x ROAS. The problem isn't necessarily the campaign — it's that 3-5 conversions is nowhere near enough data to know whether true underlying CVR is 1% or 3%. Before setting a kill/keep threshold, you need to size a minimum viable test budget: the spend required to reach a conversion count where the observed CPA is a reasonably reliable estimate of the true CPA, not noise.
Why conversion count, not spend or days, is the right unit
Google's and Meta's own bidding documentation both anchor learning-phase guidance to conversion event counts, not calendar days or dollars — Meta cites roughly 50 optimization events per week as the threshold for exiting the learning phase, and Google's Smart Bidding documentation notes campaigns need a meaningful volume of recent conversions to generate reliable bid predictions. A test budget should therefore be sized backward from a target conversion count, with spend and days as the derived outputs, not the inputs.
Setting the conversion count target
Statistically, 15-20 conversions is a commonly used rough floor for getting a directionally reliable CPA read (still with meaningful variance, but no longer dominated by single-order noise); 30-50 conversions materially tightens that estimate. For a first-pass keep/kill decision, 20 conversions is a reasonable minimum viable target; for a decision about scaling budget significantly, wait for 30-50.
Worked example: converting target conversions into budget
Estimated CPC $0.90, estimated CVR 2.2% (a reasonable starting estimate from category benchmarks or the store's own site-wide average before this specific campaign has data). Estimated CPA = CPC ÷ CVR = $0.90 ÷ 0.022 = $40.91. To reach 20 conversions: minimum viable test budget = 20 × $40.91 = $818.20. Round to a clean $800-$850 test budget before making any keep/kill call on this specific campaign.
What to do if CVR is unknown or highly uncertain
For a genuinely new product or new audience with no CVR history, build a range rather than a point estimate: model a pessimistic CVR (say half your site-wide average) and an optimistic CVR (your site-wide average), then size the test budget off the pessimistic case so you don't underfund the test and kill a campaign prematurely. If site-wide average CVR is 2.4%, model the pessimistic case at 1.2%: CPA = $0.90 ÷ 0.012 = $75, minimum viable budget for 20 conversions = $1,500. This is a larger number than the optimistic case, deliberately — underfunding is the more common and more costly error.
Daily budget pacing within the test window
A minimum viable test budget of $1,500 shouldn't be spent in a single aggressive day — both platforms' delivery algorithms need a spread of impressions across different times, audience segments and (for Meta) placements to properly evaluate performance. A reasonable pacing target is spending the full test budget over 10-21 days, giving daily budgets in the $70-$150 range for the example above, long enough to smooth out day-of-week variance (weekday vs weekend CVR often differs meaningfully in ecommerce) without dragging the test out for months.
Per-platform nuances in test sizing
Google Shopping/PMax campaigns often show a wider early spread in CPA than Meta because the auction includes both branded and non-branded queries with very different intent levels, meaning the effective sample needed to stabilize CPA can be somewhat larger. Meta Advantage+ campaigns typically stabilize faster within the platform's own learning-phase framework but are more sensitive to creative fatigue, meaning the underlying CVR itself can shift mid-test if the same 2-3 ad creatives are shown repeatedly to a capped audience — worth checking frequency metrics during the test window.
Worked example: a genuinely underperforming campaign caught early
A campaign reaches 20 conversions at $58 CPA against a modeled $40.91 estimate and a break-even CPA of $48 (from a 55% contribution margin on a $87 AOV). This is a real signal — the campaign is running 41% above the estimate and above break-even with a reasonably sized sample. This is exactly the situation the minimum viable test budget exists to surface with confidence, versus killing the campaign on day 3 at $58 CPA from just 4 conversions, where the true CPA could easily have been anywhere from $30 to $90.
When to extend the test rather than kill or scale
If a campaign hits its minimum viable conversion count with a CPA landing close to (within roughly 10-15% of) break-even rather than clearly above or below it, that's a genuine 'inconclusive' result, not a failure. The correct response is extending the test to the 30-50 conversion tier rather than making a binary call on ambiguous data — a $48 break-even against an observed $52 CPA at 20 conversions is well within the noise band that a further 15-20 conversions could resolve in either direction.
Budgeting for multiple simultaneous tests
Running 4 new audience or creative tests simultaneously, each needing an $800-$1,500 minimum viable budget, requires $3,200-$6,000 in dedicated test spend — which needs to be planned for explicitly as part of the monthly test reserve discussed in the overall monthly budget framework, rather than squeezed out of core campaign budgets mid-month. Prioritize which tests actually run based on available test budget and expected upside, rather than running every idea simultaneously and underfunding all of them.
A pre-test checklist
Before launching any new campaign, write down: the estimated CPC and CVR (and their source — historical data, category benchmark, or an educated guess), the resulting estimated CPA, the minimum viable conversion target (20 for a first read, 30-50 for a scaling decision), the resulting test budget, and the break-even CPA it will be judged against. Having this written down before launch prevents the common failure mode of moving the goalposts mid-test based on gut feel about early, statistically thin results.
How this connects to automated optimization
Rules-based budget automation, including Adsify's auto budget scaling rules, works best when it's configured with these same conversion-count thresholds rather than acting purely on daily spend caps — for example, holding a campaign's budget flat until it reaches 20 conversions before allowing any scale-up or scale-down adjustment, which mirrors the manual discipline this article describes but removes the temptation to intervene emotionally on day 3.