# Ad Testing

How the Ad Test measures whether a video ad is watched to the end, how much it is liked, how it shifts interest in the brand and which scene sticks in viewers' minds.

An Ad Test shows a video ad to a population of synthetic consumers (personas) and measures how they react. The report tells you whether the ad is watched to the end, how much it is liked, how far it shifts interest in the brand, whether viewers can tell whose ad it is, and which scene sticks in their minds. An example question: "Does our new 30-second ad bring viewers closer to the brand, or is the brand nowhere to be seen in the most-remembered scene?"

## When to use it

- When you want to see whether the ad holds viewers in the first few seconds.
- When you want to find out which scene is liked, where interest drops and which scene is remembered.
- When you want to measure whether the ad still links back to your brand when the brand name is hidden.
- When you want to test several ad options separately on the same population and compare them.

If you want to measure willingness to pay (WTP) for the product, set up a separate [Price Analysis (WTP)](/en/docs/research/price-analysis) study in General Research.

## Before you start

- **You need an Ad Testing population.** Populations are tied to a vertical; a population you created in General Research or Fashion isn't listed here. In the **Reklam Testi Attributes** tab of the new population form, you set the distributions for **Kategori İlgilenimi** (Category Involvement) and **Reklama Açıklık** (Ad receptivity). Each persona is assigned an archetype that describes how they view ads (e.g. "Şüpheci" (Sceptic), "Fırsatçı" (Deal Seeker), "Reklam Atlayan" (Ad Skipper)). For details, see [Creating a population](/en/docs/populations/creating-a-population).
- **The population is built with a segmentation.** You can't create a new Ad Testing population without finding and accepting a segmentation; this is what lets the report show what the ad changes in each market segment. Ad Testing populations created before 14 September 2026 have no segments.
- **The whole population is used.** The study runs with every persona in the population; you can't select a subset. A population has between 10 and 2,001 personas.
- **Prepare the ad.** A public YouTube link to an ad of up to 3 minutes is all you need.

## Setting up the study

1. In the sidebar, click **New Research**.
2. On the **Choose a vertical** screen, select the **Ad Testing** card. The vertical has only one method, so the method step is skipped.
3. On the **Choose a population** screen, pick a population with **Use existing population**, or create one with **Create a new population**.
4. If you picked an existing population, click **Continue with selected population**. When you submit the new population form, the next page opens automatically.
5. On the **Reklam Testi** (Ad Test) page, wait until the badge reads **Popülasyon hazır** (Population ready). Until then, the **Next Step** button reads **Waiting...**.
6. Paste a public YouTube ad link into the **Reklam videosu veya görseli** (Ad video or image) field. If **Yükle** (Upload) and **YouTube** tabs appear above the field, select the **YouTube** tab first; you can use the **Yükle** tab to upload a video file from your computer instead. If the link is valid, the player loads and the duration appears after **Video süresi okunuyor…** (Reading video duration…).
7. Optionally, under **Reklam nerede yayınlanacak? (isteğe bağlı)** (Where will the ad run? (optional)), choose the placement, that is, the setting where the ad will run. To clear your choice, click the card again.
8. If the ad isn't meant to sell, describe its purpose in the **Bağlam (opsiyonel)** (Context (optional)) field, e.g. "10 November Atatürk commemoration — no sales goal".
9. Check the **Tahmini kredi** (Estimated credits) card on the right and click **Next Step**.
10. On the **Research Preview** page, check the **Credit Usage** and **Person Count** values on the **Experiment Summary** card, then click **Start**.

There are no separate fields for the brand, product or category; the system reads them from the ad itself.

The bar on the progress screen (**Experiment Processing**) shows the share of personas who have finished. When **Your analysis is ready** appears, you're taken to the report; later, you can reach it from the **Reports** screen with the **Ad Test** filter. If **Experiment Failed** appears, click **Retry**; the study restarts with the same settings.

You can also have Argus prepare an Ad Test study by giving it a public YouTube link, but you have to upload a video file yourself. For details, see [Working with Argus](/en/docs/argus/working-with-argus).

## Settings and limits

| Setting | Limit |
| --- | --- |
| Ad source | One ad per study: a public YouTube link or (if the **Yükle** tab appears) a video file |
| Duration | Up to 3 minutes (180 seconds) |
| Video file | mp4, mov, webm; up to 100 MB |
| Placement | Optional; one of four cards |
| Context | Optional; up to 2,000 characters |
| Population | A completed Ad Testing population; all of it is used |

The placement determines how the personas watch the ad:

| Placement card | How the persona watches | Is reach measured? |
| --- | --- | --- |
| **Sosyal dikey video** (Social vertical video) | In a phone feed, muted; decides in the first 2 seconds whether to scroll past | Yes |
| **YouTube video reklamı** (YouTube video ad) | With sound; decides after watching for 5 seconds whether to skip | Yes |
| **TV / CTV** | An ad break with sound; can't be skipped | No |
| **Açıkhava (billboard)** (Outdoor (billboard)) | A single, silent glance; sees only the opening frame and the text on it | No |
| No selection | Watches the whole ad without skipping | No |

## Credit calculation

Credits = video duration (in minutes, rounded up) × number of personas × 10. There is no fixed fee.

- 30-second ad, 50 personas: 1 × 50 × 10 = 500 credits.
- 2-minute 10-second ad, 100 personas: 3 × 100 × 10 = 3,000 credits.

The **Tahmini kredi** card fills in once the duration has been read; the **Birim** (Unit) row shows the calculation. If your balance is lower than the estimate, **Start** is disabled on the **Research Preview** page; hover over the warning icon next to the **Credit Usage** value to see the **Buy Credits** button.

Credits are deducted from your balance only when the study completes successfully; this also applies to a study restarted with **Retry**. A failed study doesn't use any credits. For details, see [Credits](/en/docs/account/credits).

> [!WARNING]
> Credits are calculated on the whole population and on every started minute. A 61-second ad uses the same number of credits as a 120-second ad.

## Reading the results

The report opens under the heading **Reklam Testi Analizi** (Ad Test Analysis); sections with no data aren't shown.

### How personas evaluate the ad

Each persona goes through these steps:

1. Before watching the ad, the persona scores the brand in the ad and up to three competitors chosen from the same category on a scale of 0 to 100; the score is the likelihood of choosing that brand next time. For ads that aren't meant to sell, they score their view of the brand instead.
2. On a skippable placement, they decide whether to scroll past or skip the ad. Personas who drop the ad then watch it in full as well, so every metric except **Erişim** (Reach) covers the whole population.
3. They watch the ad scene by scene with the brand name hidden and describe it in their own words: their first impression, what they liked and disliked, the best and weakest scenes, the scene they remember, the main message and whose ad it is.
4. Once told the ad's brand, they score the brands again and write down what draws them to the brand and what holds them back.
5. A neutral evaluator uses this account to derive liking, sentiment and clarity scores from 1 to 10.

The ad is split into at most 12 scenes. The system also works out the ad's objective: sales, brand image, commemoration/celebration or awareness. If the objective isn't sales, the note "Bu reklam ... amaçlı değerlendirildi" (This ad was assessed as a ... ad) appears at the top of the report. If the objective was misread, write the correct one in the **Bağlam** field and start a new study.

### Test Edilen Reklam (Tested Ad)

This section shows the ad, the placement you chose and a one-line summary, e.g. "32/40 sonuna kadar izledi · 18/40 markaya ısındı · en çok hatırlanan sahnede marka yok" (32/40 watched to the end · 18/40 warmed to the brand · brand absent from the most-recalled scene). The five boxes below it are the report's main metrics:

| Metric | What it shows |
| --- | --- |
| **Erişim** | The share of personas who watched the ad to the end. Appears only on the two skippable placements. |
| **Beğeni** (Liking) | The share who strongly liked the ad (8 or more out of 10). If your organisation has at least 8 other completed Ad Test studies, it shows a 0–100 rank instead: "Önceki N reklam arasında üst %X" (Top X% among the previous N ads). |
| **Alınabilir talep** (Reachable demand) | The average likelihood of choosing the brand after the ad. The sentence in the box gives the number of personas who warmed to the brand (whose score rose, however slightly) and who cooled, the rises of 10 points or more ("belirgin biçimde" (markedly)) and the before → after value. For non-sales ads, the box is called **Marka yakınlığı** (Brand affinity). |
| **Ayrışma** (Distinctiveness) | The share who named the brand correctly while the brand name was hidden, and the chance level. |
| **Kalıcılık** (Memorability) | The share who got the main message; the most-remembered scene and whether the brand is in it. |

The liking average stays within a narrow band in the middle of the scale, whatever the ad. That's why the 1–10 average appears in small print; what gives liking its meaning is the comparison with your other ads.

### Sahne Sahne (Scene by scene)

In scenes where the brand is on screen, the scene card carries a **marka** (brand) tag. The chart gives persona counts for each scene: **İzliyor** (Watching: still watching when the scene starts), **Zirve** (Peak: called it the best moment), **Dip** (Low point: got bored or tuned out), **Hatırlanan** (Recalled: remembered that scene) and **Hikâye** (Story: the ad's own story arc, 1–10). Click a scene to open its "neden zirve" (why it peaked) and "neden dip" (why the dip) comments.

### Erişim (Reach)

Appears only on the **Sosyal dikey video** and **YouTube video reklamı** placements. It shows how many personas scrolled past or skipped, how many kept watching second by second, and, under the heading **Neden bıraktılar?** (Why did they drop off?), the most common reasons with sample comments.

### Alınabilir Talep (Reachable demand)

**Markayı alabilecek kesim** (Share who could buy the brand) gives the share before and after the ad, the difference between them, and the number of personas who warmed, cooled or didn't move. The short comment below it states the direction of the effect ("Tek yönlü etki." (One-way effect.), "Kazanç var, ama iki yönlü." (There is a gain, but it cuts both ways.)) and who the ad resonated with.

Some reports also show a **Reklamdan önce ve sonra** (Before and after the ad) price chart. This chart is drawn against an estimated shelf price found automatically and needs at least 12 personas to answer a short price question. The chart doesn't give an exact price; for pricing decisions, run a separate Price Analysis study.

### Ayrışma (Distinctiveness)

Compares the **Marka adı gizliyken doğru bilen** (Correctly identified with the brand name hidden) share with the chance level, that is, the probability of getting it right by guessing (25% in a four-brand set), and gives the split of answers between correct, another brand and couldn't say. **Markaya kilitlenme** (Brand lock-in, 1–10) shows how far the ad could only belong to this brand. Results are also given separately for **Markayı kullananlar** (Brand users), **Rakip kullananlar** (Competitor users) and **Markası olmayanlar** (No brand). Competitor names aren't shown in the report.

### Kalıcılık (Memorability)

Shows the most-remembered scene and whether the brand is in it. **Markalı hatırlama** (Branded recall) gives the share of remembered scenes in which the brand is on screen; **Mesajı alan** (Got the message) gives the number of personas who put the ad's main message into their own words. **Akılda kalan mesaj** (Message that stuck) lists the ad's main messages and sample comments that carry them.

### Persona Sesleri (Persona Voices), Çekenler ve Engelleyenler (Draws and Barriers), Neler İşe Yaradı? (What Worked?)

- **Persona Sesleri**: The two warmest and two coldest personas by liking, from different archetypes where possible.
- **Çekenler ve Engelleyenler**: The reasons that draw personas to the brand and the reservations that stop them trying it, with persona counts. The sentence at the top says which side outweighs the other. Personas who already use the brand and those not currently considering a purchase aren't counted. For non-sales ads, the heading becomes **Dokunan ve Sahte Gelen** (What landed and what rang false).
- In the **Neler İşe Yaradı?** section, **En çok ne işe yaradı?** (What worked best?) shows for how many personas each persuasion element (e.g. **Mizah** (Humour), **Fiyat/Fırsat** (Price/Offer), **Duygu/Hikâye** (Emotion/Story)) was the main driver. **Tüketiciler neyi eksik buldu?** (What did consumers find missing?) lists the elements that viewers of this ad found missing more often than viewers usually do.

### Persona Yorumları (Persona Comments)

The **Persona Kanıtları** (Persona Evidence) card lists each persona's reaction one by one. Search the comments with the **Ara...** (Search...) box and narrow the list with the **Tüm arketipler** (All archetypes) filter. When you expand a row, you see **Duygu** (Sentiment), **Beğeni**, **Netlik** (Clarity), **Değişim** (Change: the difference between the scores after and before the ad), **Söylediği marka** (Brand named), the scenes, and the **Beğendiği** (Liked), **Beğenmediği** (Disliked), **Ana mesaj** (Main message) and **Gerekçe** (Reasoning) texts.

> [!NOTE]
> Reports for studies analysed before mid-September 2026 appear in the old layout: purchase intent, brand recall, a "Yayına hazır / İyileştir / Yeniden ele al" (Ready to launch / Improve / Rework) verdict, **Skor Dökümü** (Score Breakdown) and **Satışa dönme** (Conversion to sales). New studies don't produce these metrics.

## Analysing by segment

- **Hangi Kitle Ne Düşünüyor?** (What Does Each Audience Think?) shows the change in intent after the ad in the **Marka ilişkisine göre** (By brand relationship), **Yaşa göre** (By age), **Cinsiyete göre** (By gender) and **SES'e göre** (By SES) panels. A green bar marks a group whose intent rose by more than half a point; a red bar marks one whose intent fell by more than half a point.
- **Segmentte ne değişti** (What changed in each segment), inside **Alınabilir Talep**, plots the before and after values for each market segment. This needs a population with a segmentation and at least 4 personas per segment; otherwise, the warning "Bu panelde segment yok" (No segments in this panel) appears.

For details, see [Segmentation](/en/docs/results/segmentation).

## Exporting

- **Download Excel**: Downloads a single-sheet Excel file with one row per persona. The columns are archetype, age band, gender, income group, the 1–10 scores, placement behaviour, before and after scores, the brand named, scenes and main message. The column headers are partly in English and partly in Turkish; the "Brand Recall" and "Purchase Intent" columns are left over from an earlier version and may be empty in new studies.
- **Download PDF**: Prepares a five-page PDF report in the background. The pages are the cover, the main metrics, draws and barriers, the persuasion elements that worked and those found missing, and the audience breakdown.

For details, see [Exporting and downloading](/en/docs/results/exporting).

## Tips and common mistakes

- **Choose the placement where the ad will actually run.** If you don't choose one, the ad is watched from start to finish and **Erişim** isn't measured.
- **Choose Açıkhava (billboard) only for an ad made for outdoor display.** On this placement, personas see only the opening frame and the text on it.
- **Fill in the Bağlam field for non-sales ads.** Otherwise the system may treat the ad as a sales ad and ask personas about buying.
- **Read Ayrışma carefully.** If the note "Marka N sahnede ekranda görünüyor." (The brand is on screen in N scenes.) appears, a high share mostly means the logo was seen; compare the result with the chance level, not with 100%.
- **Upload a video file if you can.** With a YouTube link, the reactions rely only on a text description of the ad. If the **Yükle** tab appears and you upload a file, personas also see frames from the ad, and the **Sahne Sahne** strip shows frames instead of scene numbers.
- **Read the results as directional.** The footnote in the report's **Alınabilir Talep** section also reminds you that the results come from synthetic consumers; see [Accuracy and limitations](/en/docs/getting-started/accuracy-and-limitations).

## Related pages

- [Verticals: General Research, Fashion, Ad Testing](/en/docs/research/verticals)
- [Creating a population](/en/docs/populations/creating-a-population)
- [Segmentation](/en/docs/results/segmentation)
- [Exporting and downloading](/en/docs/results/exporting)
- [Credits](/en/docs/account/credits)