When a video campaign is being set up, the discussion usually circles the video itself: the edit, the length, the music. Yet a decision made much earlier in the panel shapes the outcome more than any of that — the campaign objective. That choice tells the platform who to show the ad to and what to optimise towards. A campaign built on the wrong objective can waste even a good video.
The objective decides who the platform looks for
Ad platforms run on machine learning, and you are the one who tells them what to learn. Choose a views or reach objective and the system goes looking for people who will watch: users who have sat through similar content, who do not skip, who spend time around videos of that kind. Choose a conversion objective and it looks for someone else entirely — people who have filled in forms, called, or bought.
On YouTube ads the split is especially visible: depending on the objective chosen, the same video can reach two audiences that barely overlap.
These two groups are often not the same people. Someone who enjoys watching a video is not necessarily the person closest to buying, and the person closest to buying may skip at the first opportunity and go straight to the site. Choosing an objective is therefore not a matter of preference; it is a description of who you want found.
What cheap views hide
Campaigns built on a views objective usually report well: view counts are high, cost per view is low, completion rates look satisfying. It is easy to read that table and conclude the campaign is working. But a falling cost per view does not mean the business is receiving more enquiries.
What usually happens as the cost drops is this: the system drifts towards whichever audience is cheapest to reach. That audience tends to be made up of people with little interest and no intention to buy, who simply do not close the video. The numbers improve while the buyer gets further away. This is why a video report is never read on its own; it has to be weighed against the real demand the business is seeing on the measurement side.
Asking for conversions before the brand is known
A conversion objective is not always the right call. Where the brand is not yet known, the product needs explaining, or the purchase decision takes time, asking someone to act the moment they first see the video is asking too early. The platform cannot gather enough signal either; the campaign narrows onto a small audience and the budget burns through the learning phase.
This is where video advertising is at its strongest, and it works in two stages. First a broad but relevant audience is reached with an awareness objective; then the people who watched a certain share of that video are collected as their own audience and remarketed to. Watch time matters here: having persuaded someone to stay through half of your video is a measurable signal of interest. That signal produces a far smaller but far better qualified audience for the next step.
The audience definition changes with the objective
When the objective changes, the audience definition has to change with it; building the two independently is a common mistake. In the awareness stage, broad interest and behaviour-based audiences make sense: people engaged with the subject, consuming similar content. In the conversion stage the audience needs to be narrow and its signal strong — people who have visited the site, watched the video, looked at the product.
Pointing a conversion objective at that same broad audience hands the system a job it struggles with: as the pool grows the conversion rate falls and learning slows. The reverse holds too — running an awareness objective at a very narrow audience simply shows the same people the ad more often than necessary.
Why the video's contribution is missing from the report
The injustice video advertising most often suffers is on the measurement side. A viewer sees the video, does not click then, searches for the brand directly a few days later, arrives at the site and fills in a form there. In a report built on last-click logic that conversion is credited to search; the video's share is nowhere to be seen.
The consequence is predictable: the video campaign is switched off as "not working", and some time later search performance drops too, with no obvious reason. That is why video campaigns are judged not only on direct conversions but on the movement in brand searches and the overall demand curve.
Reusing the same video with different objectives
A video does not belong to one campaign. The same footage can go to a broad audience in the awareness stage; a shortened cut can serve as a reminder in remarketing; the section explaining the product can live on the site. What changes is not the video but the objective it carries and the people it reaches.
That approach changes the cost too: rather than filming separately for every stage, the material from one shoot can be recut for different purposes. The decision to make is less about which video to film than about what gets measured at which stage.
The question to answer before choosing
In practice a single question settles most of it: at the end of this campaign, what will we look at to say it worked? If the answer is "people should know the brand", the views and reach side is right, and the measures are watch time and the number of people reached. If the answer is "forms should arrive, the phone should ring", the conversion objective is right — but then conversion tracking has to be in place and the video has to invite the viewer to a next step.
Where that question goes unanswered, the campaign ends up between the two: neither reaching a wide enough audience nor producing demand anyone can measure. Choosing the objective comes before the quality of the video, and decides the outcome more than it does.
Changing the objective mid-flight
When a campaign is not going as hoped, the first reflex is to change the objective. That intervention costs more than it looks: once the objective changes, most of what the platform has learned so far stops applying, because it is now looking for a different behaviour. In practice the campaign starts over, and the cost of the learning phase is paid a second time.
The better move is usually not to change the objective but to build the new one as a separate campaign and run the two side by side for a while. That keeps the data from the original setup and makes a comparison possible. Even where the objective clearly was chosen wrongly, when to change it is a separate decision — results taken while a campaign is still learning say little that can be trusted about the setup itself.