Stock footage is why your local ads do not work
A generic van on a generic street tells a local buyer you are not local. We watched this happen across enough campaigns to stop blaming the targeting.
When a small-business campaign underperforms, the first instinct is to blame targeting. Widen the radius, adjust the age bracket, try a new interest stack. We have spent a lot of hours doing that.
Often the targeting was fine. The creative was telling people the business was not real.
What the buyer is actually checking
Someone searching for emergency AC repair at ten at night is not evaluating production quality. They are running one check, mostly unconsciously: is this a real company near me, or is this a lead-generation middleman who will sell my number on?
That second category is genuinely everywhere in home services. Buyers have been burned and they pattern-match hard.
Stock footage trips every one of those patterns:
- The van is a model or livery that does not exist in their state
- The house in the background is architecturally from somewhere else entirely
- The crew is a stock crew — too well lit, too evenly matched, too many of them
- The same clip appears in a competitor’s ad, because there are only so many plumbing clips in the library
None of that registers consciously. It registers as something is off, and the thumb keeps moving.
The tell we kept seeing
The clearest version of this: a client’s static photo ads outperformed their stock-video ads. Same offer, same audience, same budget.
A grainy phone photo of an actual technician at an actual job beat a polished stock clip. Which makes no sense if you think creative quality drives performance, and perfect sense if you think evidence of being real drives performance in local services.
Motion should beat static on paid social. It generally does. It did not here, because the motion was lying and the photo was not.
Which is the whole reason we care about reference images
This is why the generation approach matters to us beyond cost. Wan3.0 takes the client’s own photos as reference input. Their van, their livery, their uniform, their street.
The output is generated, and we are not going to pretend otherwise — it is not documentary footage of a specific job. But it is generated from their actual visual identity rather than from a library of somebody else’s. The van is the right van. The regional architecture is right. The uniform colour matches the one on their truck and their website.
That clears the “is this real” check in a way stock footage structurally cannot.
The rule we work to now
Two questions before any spot goes out:
- Would a local recognise this as their area? If the roofline, the plates or the plant life is wrong, it fails.
- Could this exact clip appear in a competitor’s ad? If yes, it is doing nothing for the brand and possibly harming it.
Stock footage fails both by definition. That is not a quality judgement about stock libraries — they are made to be reusable, which is exactly the property that breaks local trust.
Still true
Nothing here beats a real camera pointed at a real completed job. If a client will send us phone footage from a site, we use it and it outperforms everything else.
Most of them will not, consistently, forever. The realistic comparison is not generated video against a professional shoot. It is generated video against the stock clip they would otherwise have bought, or against no video at all.