Views tell you a creative was displayed, not that it worked. To judge a campaign you need a hierarchy of measures that runs from attention through engagement quality to business actions, plus at least one method that asks the harder question: would these results have happened anyway?
This article sets out that hierarchy, explains brand lift and incrementality in plain terms, and gives you the tagging and reporting habits that make the numbers trustworthy.
Why views mislead
A view count is a delivery metric. It depends on platform definitions, placement, autoplay and how many people were shown the ad. Two campaigns with identical views can differ enormously in what viewers did next. Treat views as context, then move down the hierarchy.
The metrics hierarchy
| Level | Question | Example measures | Caution |
|---|---|---|---|
| 1. Delivery | Did it reach the right people? | Reach, frequency, impressions | Check audience and placement, not just totals |
| 2. Attention | Did people stay? | View-through rate, average watch time, hold rate past the opening seconds | Platform definitions differ, so compare like with like |
| 3. Engagement quality | Did they care? | Saves, shares, meaningful comments, profile or site visits | Likes alone are weak |
| 4. Brand effect | Did perception change? | Ad recall, awareness, consideration from a lift study | Needs a control group |
| 5. Business action | Did they act? | Enquiries, bookings, purchases (key events) | Attribution can overstate |
| 6. Incremental impact | Would it have happened anyway? | Lift tests with holdout groups | Needs scale and design |
Choose two or three measures per level you can influence, and agree them before launch. The planning logic sits in our digital marketing strategy framework.
Attention: using view-through and hold rate
Short-form video lives or dies in the opening seconds, so look at how many people stay past the first few seconds and how far through they watch. Use these to compare creative versions against each other within one platform. Do not compare a hold rate on one platform with another’s, because definitions vary. For craft and testing ideas, see short-form video ads for TikTok, Instagram and YouTube.
Engagement quality
Rank engagement by effort and intent: a save or share shows more commitment than a like, and a question in comments shows interest you can act on. Read a sample of comments to check sentiment, because a spike in comments can be complaints.
Brand lift studies
Brand lift is survey-based measurement of how ads shift brand perception. Google describes Brand Lift as a measurement tool for video ad campaigns that tracks brand outcomes such as ad recall, brand association, awareness and consideration, rather than clicks or views. It surveys people who saw the ads and people who were eligible but were not shown the ad, and the difference between them shows the ads’ influence. It supports Video and Demand Gen campaigns and is not available to all accounts, as access is through a Google account representative (Google Ads Help).
When to consider it: when your goal is awareness or consideration and you have enough media budget to produce usable survey responses. Smaller advertisers can approximate the idea by asking new customers how they heard of you, while recognising that is self-reported and imperfect.
Conversions: key events in GA4
In Google Analytics, a key event measures an action particularly important to your business. Any collected event can be marked as a key event, and the results appear in reports such as Landing pages and User acquisition (Analytics Help). Define one to three: form submission, booking, purchase, click-to-call.
Link them to your destination page. A weak page hides a strong creative, which is why landing pages that turn campaign traffic into enquiries is part of measurement, not separate from it.
Remember that key events in analytics record actions, not their quality. Add an offline step: did the enquiry become a sale?
Incrementality: what would have happened anyway
Attribution assigns credit to the ads a user saw or clicked. Incrementality asks how much of the result the campaign actually caused. Google’s measurement guidance describes incrementality testing as a randomised, controlled experiment comparing people exposed to a campaign with people who are not, showing the extra revenue, profit or action beyond what would have happened without it. It notes that attribution suits day-to-day optimisation, while privacy changes make attribution alone harder to rely on for budget decisions (Think with Google, Australia and New Zealand).
Google Ads’ Conversion Lift runs this kind of experiment by splitting an audience into a group that sees the ads and a control group that does not, and is not available for all accounts (Google Ads Help).
You can run a low-tech version without those tools: pause a campaign in one region or period while keeping another running, then compare enquiries. It is imperfect, because seasonality and promotions confuse the picture, but it is better than assuming every conversion was caused by an ad.
UTM discipline
Without tagging, traffic from your campaigns blends into “direct” or “referral”. Required UTM parameters are source, medium and campaign; term and content are optional. Analytics is case sensitive, so utm_source=google and utm_source=Google are separate sources (Analytics Help). A workable convention:
- All lower case, no spaces, hyphens between words
- Source is the platform, medium is the type (paid-social, email, video)
- Campaign is a stable name used by everyone
- Content identifies the creative version, so you can compare them
- One shared register, owned by one person
Reporting cadence
- Weekly (operational): delivery, spend, attention, errors. Used to fix problems.
- Monthly (tactical): engagement quality, key events by channel and creative, cost per result. Used to decide what to keep.
- Quarterly (strategic): brand measures, incrementality results, budget reweighting.
Every report should end with decisions: what to stop, start and continue. A report with numbers and no decision is a diary.
Measurement checklist
- Goal and level-by-level measures agreed before launch
- Key events configured and tested
- UTM convention documented and applied to every link
- Creative versions labelled for comparison
- One lift or holdout approach planned for any large spend
- Offline outcomes (sales, no-shows) fed back to the report
- Review dates scheduled
Where Radiance Impact fits
Radiance Impact’s Growth & Marketing division covers strategy, campaigns and search visibility, and the Studio produces the creative those campaigns run on. To plan measurement alongside production, get in touch at /contact or see /growth.
Sources
- About Brand Lift, Google Ads Help
- Incrementality testing, Think with Google (Australia and New Zealand)
- About Conversion Lift, Google Ads Help
- Key events in Google Analytics, Google Analytics Help
- Collect campaign data with custom URLs, Google Analytics Help
