DEXUN AdWhiz’s core promise is that your savings are visible and verifiable. For a savings number to hold up legally and be statistically falsifiable, it must rest on a randomized control — not a before/after comparison. This page discloses our measurement method so it can be reproduced and audited.
1. The core idea: incremental savings
We do not claim that “you spent less this year than last” is our doing — that could be seasonality, your own campaign changes, or market shifts. We measure the increment: within the same period and the same account, we compare the portion AdWhiz optimizes against a randomized holdout that is never optimized. The difference is the savings AdWhiz actually contributed.
2. Definitions
- Holdout (randomized control group): a randomly reserved portion of your budget / ad sets that AdWhiz never optimizes, kept as a control baseline throughout the Measurement Period.
- Optimized cohort: the portion AdWhiz optimizes.
- Cost-per-result: the cost of each conversion (or your chosen conversion objective). Both cohorts use the same conversion-value definition so they are comparable.
- Measured Savings: the cost-per-result reduction of the optimized cohort relative to the holdout, on the same basis, scaled to your actual spend and shown in your dashboard.
3. Why randomized control, not before/after
A before/after comparison (with vs. without AdWhiz) cannot rule out seasonality, market swings, or your own operational changes, so it is not causally sound. The holdout and the optimized cohort share the same time window and account environment; the only systematic difference is “optimized by AdWhiz or not,” so the gap between them is attributable to AdWhiz — a randomized experiment whose conclusion is falsifiable.
4. Industry precedent: Conversion Lift
This is the same family of method as Meta’s, Google’s, and TikTok’s Conversion Lift / incrementality measurement — the platforms themselves use randomized control to measure the incremental effect of advertising. We follow the industry-standard causal-measurement paradigm rather than inventing our own.
5. Data source & record
- All performance data is ingested directly via the official APIs of the relevant advertising platforms (Google / Meta / TikTok) — not hand-entered or estimated.
- DEXUN computes Measured Savings in good faith from that objective data.
- The dashboard figure is the agreed measure; absent manifest error or technical malfunction, it is binding on the parties.
- You have a 14-day reconciliation right: if you in good faith believe there is a manifest error or technical malfunction, you may dispute in writing within 14 days of the figure being published, and we will review and correct it if confirmed (see ROI Savings Guarantee Terms, Section 5).
6. What counts / does not count as Measured Savings
- Counts: the amount corresponding to the optimized cohort’s cost-per-result reduction relative to the holdout, on the same conversion-value basis.
- Does not count: platform algorithm/policy changes, account suspensions, your own campaign changes outside AdWhiz’s recommendations, seasonal market swings, and other factors outside AdWhiz’s control.
- This methodology measures cost saved — it is not a guarantee of profit, revenue, or ROAS.
7. Data sufficiency (building a baseline)
A statistically valid measurement needs enough conversion data. When conversions are too few or the posterior credible interval is too wide, the dashboard shows “building a baseline” rather than a number we ourselves do not believe. The minimum ad-spend threshold (see the Guarantee Terms) exists precisely to keep the measurement statistically valid.
8. Versioning policy
This methodology is versioned. The version applicable to you is the then-current version published on this page as of the date you purchased the annual subscription (or as of the start of your Measurement Period). We will not unilaterally change it to your detriment during your Measurement Period; any change is prospective.