Ad testing prices by approach

Four ways to test an ad, from cheapest to most involved, with the prices providers publish.

Ad testing prices by approach
ApproachPublished priceExampleWhat drives the cost
Predictive AIFrom $500Kantar LINK AINumber of assets screened. No viewers are recruited, so it is the cheapest and fastest.
Self-serve survey or webcam platform$325–$990 a month (annual billing $3,900–$9,900 a year), plus $15 per recruited panelistRealEye; SurveyMonkey ad testing (pay per study, price shown at setup)Plan tier, number of participants, study length. You design and analyse the study yourself.
Managed webcam facial coding$1,695–$1,995 per creativeElement Human (credit-based)Number of creatives and volume of credits bought. Discounts for quarterly or annual commitment.
Full-service agency pre-test$11,000–$30,000 per studyDrive Research (ad concept testing)Method, sample size, audience, survey length and depth of reporting.
Enterprise norm-based pre-testQuote onlySystem1 Test Your Ad, Kantar LINK+, ZappiMarkets, sample, norms and subscription or credit bundles.

Prices in US dollars as published on 17 October 2026: Kantar Marketplace (LINK AI “starting from $500”), RealEye pricing, SurveyMonkey ad testing, Element Human pricing ($1,995 per credit for 1–50 credits, $1,895 for 51–150, $1,695 for 151+; one creative test uses one credit), Drive Research and System1 and Zappi (both quote on request). Vendors change prices; check their pages before you budget.

What drives the price of an ad test

Five things account for most of the difference between a cheap test and an expensive one.

Sample sizeViewers per ad
EchoDepth recommends 15 or more viewers per audience segment for each ad or edit. Every group you want to report on separately, such as under-35s and over-35s, is its own segment and needs its own 15. Doubling the sample roughly doubles recruitment cost.
Number of editsAds tested
Comparing edits fairly means a separate sample for each (a monadic design), so three edits cost close to three tests.
Audience incidenceHow hard to reach
Adults in general are cheap to recruit. Recent buyers of a niche product, business decision-makers or parents of toddlers cost more per completed viewer.
MarketsCountries and languages
Each market needs its own sample and often translated questions, and norms do not carry across cultures.
TurnaroundHow fast
Faster fieldwork usually costs more, especially for hard-to-reach audiences.

What EchoDepth costs

We quote per study, with a fixed price agreed before fieldwork starts. Send a brief and we reply within one working day.

The quote depends on the same five drivers: how many ads or edits, how many viewers, how hard the audience is to reach, how many markets and the deadline. Teams that test regularly can take an annual plan with a set number of studies at a lower rate per test.

Every study measures facial response from consenting viewers on their own webcams, coded against 44 FACS Action Units, and reports a second-by-second Emotional Response Curve, plus the Cavefish Emotional Index from what viewers say afterwards. See the methodology for how it works, or pricing and plans.

Ad testing cost questions

Planning a test? Read how to pre-test a video ad in the four weeks before launch.

Published prices in October 2026 range from about $500 for an AI-predicted test (Kantar LINK AI) to $1,695–$1,995 per creative for managed webcam facial coding (Element Human) and $11,000–$30,000 for a full-service ad concept test from a research agency (Drive Research). Self-serve webcam platforms such as RealEye charge $325–$990 a month plus participants. System1 and Zappi quote on request. The price depends mainly on sample size, the number of edits, how hard the audience is to reach, the number of markets and how fast you need results.

Compare the cost of the test with the media budget it protects. If an ad will carry a large TV or online video spend, a test costing a small fraction of that budget is cheap insurance against running a weak edit, and it often pays back by showing which cut-downs to make. If the spend is small, or you can test cheaply in-market and iterate, as with many social ads, a formal pre-test may not be worth it.

Not with a recruited audience. Some self-serve tools let you build a study for free and only charge for responses, and you can show an ad to your own customers or colleagues at no cost, but they are not a representative sample. Free predictive tools give a model's guess rather than measured reactions from real viewers.

A facial coding test needs viewers with working webcams who consent to camera use, a setup check before viewing, and processing of every video frame, on top of the survey. That adds recruitment and quality-control cost, in exchange for a second-by-second record of how viewers reacted rather than only what they say afterwards.

EchoDepth quotes per study, based on the number of ads or edits, the number of viewers and how hard the audience is to reach. Send a brief and we reply within one working day with a fixed quote. Teams that test regularly can take an annual plan with a set number of studies.