In short
With €500, start by choosing the uncertainty you need to resolve. Software can flag problems in the creative; a specialist can turn those observations into an editing brief; real respondents can explain what they understood. None of these establishes which ad will produce the most sales. This guide gives a practical plan for reviewing three TikTok ads, with dated prices and a clear boundary between pre-launch review and a live experiment.
Research checked 16 October 2026. SaliencyLab publishes this guide and sells creative-review services and software. The comparison is selective, not an independent ranking or a hands-on evaluation of every supplier.
Decide what the €500 must cover
Imagine three 30-second TikTok ads for the same product, audience and offer. One opens with a demonstration, one with a creator's explanation and one with a problem. Your launch is approaching, and the team needs to decide what to revise.
Before choosing a tool, write down whether €500 is a research allowance or your entire remaining budget. If it must also pay for editing, tax and media, a €500 service excluding VAT will not fit an all-inclusive €500 ceiling. Prices in dollars also need conversion at checkout; this guide does not treat a dollar as a euro.
Put one question at the top of the brief:
- Creative diagnosis: Is the product, benefit or next step difficult to understand in the actual cut?
- Audience response: What do eligible viewers understand or misunderstand after seeing it?
- Campaign performance: Does one version change a specified outcome under live delivery?
A team can need all three answers. With a small budget, choose the unresolved question that would most change the next edit. Buying another score is unlikely to help if the actual disagreement is whether the intended audience believes the offer.
What the available approaches buy
| Approach | Useful output | Public commercial basis | Important limit |
|---|---|---|---|
| SaliencyLab software | RoastIQ diagnostics; optional BuyerLens hypotheses after RoastIQ | Free and Pro plans; check current limits on the pricing page | Model output is neither observed gaze nor a real consumer interview |
| SaliencyLab expert review | Prioritized shortlist, annotated editing recommendations and a decision call | €500 per sprint, excluding applicable VAT | A review service, not ad production or a campaign-performance guarantee |
| PickFu survey | Feedback and reasons from recruited human respondents | Surveys start at US$1 per response; configuration changes the total | A starting rate is not a quote for three videos and your target audience |
| Lyssna study | Structured tasks and human responses | Free platform plan available; panel recruitment is charged separately | Study duration, screening and recordings can increase the bill |
| Prolific recruitment | Participants for a study you design | Participant rewards plus a platform fee | You must supply the study, quality controls and analysis |
| TikTok split test | A comparison during actual ad delivery | Media spend and any creative-production costs | It is a live campaign experiment, not a pre-launch review |
Commercial sources: SaliencyLab pricing, PickFu pricing, Lyssna pricing, Prolific pricing and TikTok's split-test instructions. Inclusion here does not establish comparative accuracy or availability for every country and audience.
When a self-directed review makes sense
If someone on the team can inspect the source material and translate feedback into edits, start with a shared review sheet. SaliencyLab's workspace can organize a RoastIQ diagnostic around the creative and brief. Check every important observation against the actual frames and audio before acting on it. Add BuyerLens only when a buyer hypothesis needs exploration; its simulated perspectives are not an independent second sample of people.
The Free software plan currently allows ten RoastIQ analyses per month, capped at three per day. Three initial ads can therefore fit a day's allowance if it is unused, but fresh reruns and revisions consume capacity too. Pro is a separate €99 monthly software subscription. Neither plan should be described as including the expert sprint. Check current terms before allocating a budget.
When the expert sprint fits
SaliencyLab's creative review sprint is €500 excluding applicable VAT. Its scope covers one brand, one campaign, one audience and one platform, with up to six creatives; each image or video of up to 60 seconds counts as a creative. Three 30-second TikTok ads can fit that scope, subject to written agreement.
The offer includes annotated priorities, a shortlist, a 30-minute decision call and one review of revised versions. Initial recommendations are due within ten business days from the agreed start after payment and complete materials. The thirty-calendar-day revision window is the time to submit revisions, not the initial delivery promise. Your editor still makes the ads.
This route suits a team that needs help interpreting evidence and giving its editor a precise brief. Choose a human-response study instead when the missing evidence is what actual eligible viewers understand or believe. Inspect the native fictional QA sample before buying: it illustrates an output and its limitations, not a customer outcome or proof of lift.
When to pay for human responses
PickFu's survey documentation supports video uploads and human feedback. Ask for the final cost of your exact questions, three assets and targeting. A headline per-response price cannot tell you that total.
Lyssna's detailed pricing guide separates platform access from recruitment. Its stated unmoderated panel base is one credit per study minute per response, with a credit normally US$1; screening and recording add fees. Illustrative arithmetic: a five-minute study with 30 responses has a US$150 base before optional fees, tax or currency conversion. This is a planning example, not a confirmed quote or a claim that 30 responses are enough for your decision.
Prolific's pricing guidance recommends at least £9 or US$12 per hour and lists a standard corporate platform fee of 42.8%. At that recommended dollar reward, 30 ten-minute completions cost US$60 in rewards; adding 42.8% gives US$85.68 before any applicable tax, screening costs or separate study-hosting costs. That calculation buys participation, not a finished research design, representative sample or statistically reliable winner.
Keep the questionnaire short enough to respect the estimated completion time. Pilot it before recruiting the full group. If it takes longer, budget and compensate accordingly.
A practical review sequence for three ads
1. Freeze the inputs
Save the exact three versions, duration, language, offer, intended audience and campaign objective. Label them A, B and C. Record whether captions and sound are part of the intended viewing experience. Preserve an unedited copy so a later review can identify what changed.
A creative brief is context, not evidence that the viewer received the intended message. If the brief explains a benefit that never appears in the ad, a diagnostic should not quietly credit the ad for communicating it.
2. Inspect the creative before buying more research
Review each ad once without pausing, then frame by frame. Record the earliest visible product cue, the first understandable benefit and the call to action. Check mobile text size, subtitles, mismatched voiceover, cropped qualifications and any claim that needs substantiation.
Use the video heatmap worksheet to distinguish an inspected frame from a model's inference. A predicted attention map can suggest where to look; it cannot tell you that a particular person looked there.
Write observations as evidence: “The refill appears at 0:11,” rather than “Viewers will not understand refills.” The second sentence requires a different kind of test.
3. Ask human viewers neutral questions where needed
For an exploratory comprehension study, collect an unaided answer before showing the brand's intended message. Useful questions include:
- What product or service was this ad offering?
- What would you expect to receive or experience after buying it?
- Which part made that expectation clear?
- What, if anything, was difficult to understand or believe?
Avoid asking “Did the clearer opening make this more persuasive?” That supplies both the interpretation and the desired answer. If people see all three versions, vary the order and record it. Repeated exposure changes the task: a participant who has learned the product from A may understand B more easily. A study in which each person sees one version avoids that particular carryover but needs enough participants in each group.
These are design choices, not a promise that either design produces representative results. Record recruitment eligibility, exclusions and incomplete responses. Do not turn a handful of comments into a population percentage.
4. Build one editing decision per ad
The following is a fictional teaching example, not a SaliencyLab customer result:
| Ad | Observable issue | Proposed revision | What remains unknown |
|---|---|---|---|
| A: demonstration | Product name appears only on the closing card | Introduce the name alongside the first demonstration | Whether earlier naming helps recall |
| B: creator explanation | Spoken benefit and subtitle describe different offers | Align the wording with the approved offer | Whether viewers find the offer credible |
| C: problem opening | The first product view is at 0:11 | Test an earlier product reveal while preserving the premise | Whether the revised version changes live performance |
The creative revision brief helps an editor distinguish what to change, what to preserve and how to check completion. Correcting the subtitle mismatch can be verified directly. Establishing an effect on recall or purchases requires new evidence.
5. Choose what deserves a live test
After revision, select the comparison that tests your most consequential remaining hypothesis. Do not combine a new opening, a new offer, a new audience and a new landing page, then attribute the result to the opening alone.
TikTok's current setup guide describes two versions, a selected variable and a key metric, and recommends at least seven days. Three candidates therefore do not automatically become one equally controlled three-way test. Plan the comparison, budget and stopping rule before delivery. A small media allowance may simply be too small to resolve a purchase-rate difference; record an inconclusive result honestly.
A budget decision you can explain to a client
Treat these as alternatives, not a bundle that somehow totals €500:
- Internal diagnosis plus targeted research: use existing staff capacity and, if suitable, a free diagnostic; obtain an exact quote for human responses; retain a stated reserve for tax, editing and follow-up. Staff time is still a cost.
- Expert-led creative review: allocate €500 excluding VAT to SaliencyLab's agreed sprint when interpretation and the editing brief are the bottleneck. Arrange production and later media separately.
- Live performance question: preserve funds for a properly designed campaign comparison. Do not spend the entire allowance on a pre-launch tool and present its score as the answer to a sales question.
Download the three-ad review worksheet (CSV). It records each asset, the evidence, the decision and the next test without manufacturing an overall “winner” score.
Before sending confidential work to any provider, use the AI creative-review confidentiality checklist. Before accepting an accuracy claim, use the creative-testing evidence guide. The useful purchase is the one that resolves the next decision at an acceptable cost and level of uncertainty.