Opening a local sale with a spec video

A cold email that says "I make short-form video for local businesses" asks the owner to imagine something. A link to a 20-second clip of their own storefront, built from three photos already sitting on their Google Business Profile, asks them to react to something. The second one gets replies, and not because the video is especially good. It's because a local owner has no framework for evaluating "I make video" and a complete framework for evaluating a video of their shop.

That difference is the whole tactic. Everything below is about making it repeatable without the render spend eating the channel.

Why spec beats a pitch on a local prospect

A portfolio reel of other people's businesses reads as "this could be anyone." It asks the owner to do the translation work — to picture their bakery where your client's gym is. Most of them won't. A clip of their own signage skips the translation entirely and replaces "can you do this" with "you already did this before I paid you."

It also self-qualifies. The owners who reply to a spec clip are, near enough, the owners who care what their business looks like from outside. That is the same trait that makes someone renew a content retainer in month four, which is why the reply rate matters less than who replies.

The catch is that the cost of being ignored is now a real number instead of the price of an email. Most of these will be ignored. Budget for that up front or the channel quietly turns into a credit sink.

Source the photos from what they already published

Pull stills from the business's own Google Business Profile gallery, its Instagram grid, or its website. Do not pull from Street View, do not use a competitor's interior, and do not substitute stock that vaguely resembles the place.

Two reasons, and the second one matters more than the first. The rights position on someone's own published photos used to demonstrate work for them is defensible; a Street View grab is not yours to animate. But the tactic dies anyway if the owner doesn't recognise the room. The entire effect is recognition. Animate the wrong storefront and you've sent a stranger a video of a stranger's shop.

Three or four stills is the working set:

• Exterior with legible signage. This is the recognition beat and it has to be first.
• One interior wide. Establishes the space.
• One detail or product shot. A cortado, a haircut, a finished install.
• Optionally one shot with a person working in it. Skip it if the faces are recognisable and you have no permission.

The build: three shots, one export

1. Animate each still with a bounded move. Slow push-in, slight parallax, nothing that reinterprets the frame. The turn a photo into a video task and the image-to-video tool both start from the still you attach, which is exactly the constraint you want here — the output has to stay obviously their place.

   "Animate this storefront photo: slow push-in toward the entrance, no camera shake, keep the signage sharp and legible. 5 seconds, 9:16."

2. Keep it short. Three clips at five seconds each plus a title card lands near twenty seconds. That's the length an owner will actually watch on a phone between customers.

3. Assemble and preview before you export. The editor runs preview: true at 480p with no credit charge and a short per-user cooldown between passes, so pacing decisions cost nothing. The single charge lands on the final export, regardless of how many clips are on the timeline.

4. Do not add a voiceover. A voice you invented for their brand is the single most likely thing to feel wrong, and it drags the conversation from "nice" to "that's not our tone." A text card with the business name and one line is enough.

Default frame rate is 25 fps, which matters only if you're cutting these against footage the client already has.

The per-prospect render budget

Outreach is a volume activity, so the ceiling has to be per prospect and it has to be set before you start, not discovered afterwards.

Step	What it is	How to cap it
Hero clip	One image-to-video pass on the signage still	One reroll, maximum
Supporting clips	Two more image-to-video passes	Zero rerolls — fix it in the prompt or ship the take
Preview passes	480p editor previews	Free, short per-user cooldown
Final export	One editor export	One charge, one time

The discipline that makes this work is the reroll cap, not the model choice. A spec clip does not need to be your best work; it needs to be recognisably their shop, moving, in under twenty seconds. The instinct to reroll until it's beautiful is the instinct that turns a 30-prospect week into a month of credits, and the beauty is not what generates the reply. Reroll rate is the variable that actually sets the cost of a batch.

Build them in batches of ten in one sitting rather than one a day. Same prompt scaffold, same edit template, different stills. The per-prospect time cost falls hard after the third one and the credit cost stays flat.

The message that goes with it

Short, link-based, no pricing.

Subject: made this for [business name]

Hi [name] — I make short videos for [town] businesses. Rather than describe it I made you one: 20 seconds, built from three photos off your Google profile. [link]

Nothing attached, it opens in a browser. Yours to use either way. If it's useful I'll tell you what a month of these looks like. If not, ignore this and you'll hear from me twice more at most.

Four rules that carry most of the reply rate:

• Link, not attachment. Attachments get filtered, don't preview, and don't open cleanly on a phone. A share link opens with a preview card and no login, which means the owner sees the first frame before deciding whether to tap.
• Say where the photos came from inside the first three lines. Unexplained footage of someone's premises reads as surveillance for about four seconds. Naming the source defuses it.
• No price in message one. You are selling a reaction, not a package.
• Name the follow-up limit and honour it. It costs you nothing and it is the only part of a cold email that's verifiable.

The follow-up sequence to a paid setup fee

The common mistake on a reply is jumping straight to a monthly retainer. That's a decision the owner can't make from an inbox. Sell a paid setup instead.

Day	Move	Ask
0	Spec clip and link	A reaction
3	A second cut — different angle, or the 1:1 version for their feed	"Want the file?" Hand it over, no conditions
8	One line tied to something real: a seasonal push, an event, a slow month	Fifteen minutes
15	Close the loop, name the paid pilot	The setup fee

The paid pilot is the actual product being sold. Price it against a month of the retainer you intend to sell next, not against what the clips cost you to make: a setup fee that reads as a rounding error beside the retainer quietly tells the owner the retainer is the rounding error. Agency rate cards for local social work are easy to find and worth reading for structure rather than for numbers — nobody publishes fill rates, so a rate card only ever tells you what is asked for, never what is paid.

Three reasons the setup fee goes first:

1. It converts a maybe into a customer at a price the owner can approve without convening anyone.
2. It pays for the genuinely expensive part — collecting the photo library, agreeing a brand look, identifying the one person who approves things, fixing a cadence — which is invisible work that a monthly retainer silently absorbs.
3. It filters. An owner who won't pay a setup fee will not survive month three of a retainer, and finding that out in week one is cheap.

From there the structure is the ordinary one: a local business content retainer with a fixed clip count, a standing brief, and a cadence that suits the vertical. If the pilot is aimed at physical traffic rather than bookings, the get local walk-ins angle is the one to build the first batch around, and the clips have a second home on the client's Google Business Profile rather than just social.

FAQ

Is it a problem to use their photos without asking?

You're animating images the business published itself, for that business, and handing over the result unconditionally. That's a defensible position and it's also why the message says where the photos came from. Where it stops being defensible is Street View captures, photos of identifiable staff or customers, and anything from a third-party review site whose terms you haven't read. If in doubt, use the exterior and the detail shot and skip the people.

How many of these before the channel pays for itself?

That depends on your reply rate and your close rate, neither of which anyone can hand you. What you can control is the denominator: fix a hard credit ceiling and reroll count per prospect, batch ten at a time, and track replies against clips sent for the first fifty. After fifty you have your own number, which is worth more than any benchmark.

Should I tell them it was made with AI?

Yes, and it's easier than it sounds because the tactic already tells them — they know you didn't shoot their storefront yesterday. One clause in the follow-up covering how the work is produced is enough at this stage. It also avoids the far worse conversation six months later when a platform label appears on a post and the client learns it from the label.

What if they take the free clip and never reply?

Then they've done you a small favour by disqualifying themselves early, and a business that posts your clip has your work on their feed. Price the giveaway into the budget rather than trying to prevent it. Withholding the file to force a conversation converts a warm prospect into an annoyed one and saves you one render.