Pitch a Playable Ad Without Pretending It Is the Full Product
Pitch a Playable Ad Without Pretending It Is the Full Product
A video ad runs on a timeline. A playable ad runs on a state machine. That is the whole difference, and it is where treatments start drifting.
A timeline plays. A state machine waits for input, changes something, and then waits again. The moment your ad does that, a question appears that a video never has to answer: the thing the viewer just did — is that the thing the product does?
Most playable treatments fumble that question quietly. Six boards, a clean match-cut to the end card, and a line that reads "player solves a puzzle." That sentence is carrying weight the boards do not support. Before anyone draws a frame, settle what the interaction is a sample of.
Choose the relationship to the actual product
There are three honest answers and one that gets teams into trouble.
A real mechanic, sampled small. The activity in the ad is the activity in the product — same rules, same kind of decision, same kind of failure — played on a smaller board or an early level. The viewer does less than they would in the app, but everything they learn about the mechanic is true.
A disclosed simplification. You are showing a version of the mechanic with something removed or changed, and the treatment says so in plain words. "This is the drag-and-slide control without the route planning that follows it." The disclosure has to name the specific thing that is missing, not gesture at it. A treatment that admits to a simplification is still an honest treatment.
A separate branded game. The interaction is its own thing — tap-and-collect, a runner, a match game — wearing the product's art. This is common, legitimate, and works best when it is pitched as what it is. The art is the connection. The activity is not, and should not be implied to be.
The one that gets teams into trouble is the separate branded game that gets pitched as a sample. Usually nobody decides to lie. Somebody reuses the artwork because it is already approved, the mechanic gets swapped for something easier to build in three days, and the treatment keeps a sentence about the game's "signature puzzle feel." Nobody in the room is sure who wrote it.
Getting this right starts with a conversation, not a screenshot. Ask the person who owns the mechanic what the game's attraction actually is, in one sentence, and what a player has to do to get it. For a puzzle that might be: you cannot move the piece you care about out of the board until you have rearranged something else. That sentence is the thing you are either sampling or replacing. Everything downstream in the treatment depends on which you chose.
If the answer is not settled yet — whether a later level introduces a timer, whether the control changes on tablet — leave it visibly unresolved in the treatment. "TBD, pending level tuning" is a working note. A guess written in confident phrasing is a commitment someone will build against.
Walk one action through its consequence
A state sequence needs five slots filled for at least one path: the starting state, the available action, what responds, what the player can do next, and what happens when they do something the ad does not allow.
Write them in that order and the treatment becomes checkable.
Starting state. What is on screen when the ad begins, and what can be touched. Name the elements, not the mood.
Action. One specific thing the player does. Not "interacts with the scene."
Response. What changes as a result — position, count, color, sound, nothing. If the answer is that nothing changes, say so, because that tells the team the state machine has a dead branch.
Next. What is now possible that was not possible before. This is the part that makes it a state change rather than an animation.
The unavailable action. The player will try the thing your ad does not support. Say what happens: does the action refuse and return, does it consume a resource, does it count as a mistake? This slot is where the honest treatments separate from the sketchy ones.
A linear movie with a destination button does not fill these slots, no matter how many hotspots it has. Touching a button that moves to the next scene is navigation. Touching something that changes what the next touch can do is a state machine. Getting the path itself right — feedback, return, the general shape of a clickable interaction — is a separate problem, and worth solving before this one.
Identify which simplifications change the promise
There is a difference between showing a small example of a mechanic and removing a condition that made it that mechanic.
Cutting a board from forty blocks to six is a small example. The blocking rule survives; the player still has to plan around obstruction; they just do less of it. The thing they learn about the product is still true.
Removing obstruction entirely is a removed condition. The player's gesture might look identical — a drag is a drag — but there are no decisions left, because there is nothing to plan around. The ad is now advertising a control, not a puzzle. If the product's attraction is route planning under obstruction, the ad no longer contains the attraction.
To find which kind of change you have made, list what is absent from the sample and then ask what each absence costs the player's understanding. Not "the demo is shorter" — that describes length, which nobody is confused about. The useful question is which decision, constraint, resource, or wait is gone, and whether a player who only saw the ad would recognize the real product.
Keep the limit next to the experience it limits. A single line buried in an appendix under a heading about deliverables will not stop anyone in the room from assuming the ad is the game. Put it where the mechanic is described.
If the simplified version teaches the wrong thing, you have two repairs and no third. Put the condition back — reduce the scope instead of the rule — or stop calling it a sample and pitch it as a separate branded activity.
Finish the experience and verify the delivery route separately
A treatment needs an ending and an exit.
Completion. What counts as done, and what triggers the end card. Name the condition precisely enough that a build could not satisfy it two different ways.
Reset. What returns to its starting state and how the player gets there. If the ad has no fail state, say that too, because it changes what reset is for.
Waiting. Any load, transition, or pause, and where it falls relative to the player's action. A wait that lands mid-gesture is a different experience from a wait before the first frame.
Failure. What happens when the player does the unavailable thing, and whether it costs them anything.
Handoff. What continues at the destination and what does not. This gets skipped more than any other line. If the ad's version had no timer and the app's early levels do, that is a difference the viewer will meet within a minute of installing. If nothing carries over — no state, no progress, no board — say so.
Google's help documentation for HTML5/playable ads describes the playable as an interactive advertising asset with its own delivery and interaction rules, rather than as the installed app (page read 18 September 2026). That framing is useful for exactly one purpose: it keeps the format question separate from the product-fidelity question. It is also narrow. It is public help text. It does not say your playable will resemble your product, and it does not predict anything about a campaign.
So confirm the actual delivery route on its own terms, for the placement you intend, before anyone builds: current size and format requirements, technical constraints, account eligibility. None of that is settled by a plausible state diagram, and none of it is settled by this page.
And a storyboard — however complete, however annotated — is not a working playable. It establishes an intention. It does not establish that the drag feels right, that the board reads at thumbnail size, or that anything survives first contact with a real thumb.
The same artwork, two different activities
Here is the comparison that makes the distinction concrete. Keystone Line is a fictional sliding-block puzzle, invented for this exercise so the rules stay fixed. Assume the following are its rules. Nothing below reports a released app.
The attraction, stated by the product's owner: you cannot move the block you care about out of the board until you have rearranged something else. A level is a 5-column by 4-row board. One target block must leave through the gap in the right wall at row 3. Blocks slide one cell at a time, up, down, left or right, into an empty cell. No diagonals, no jumping, no pushing a block out of your way. There is no timer, no move limit, and no lives.
The board at the start:
col1 col2 col3 col4 col5
row1 . . . . .
row2 . . . # .
row3 T . A B D
row4 . . . C .
T is the target. # is a fixed wall that never moves. Everything else is an ordinary block.
Drawn as a treatment, the sequence runs like this. The player drags the target one cell right. It moves to (2,3); nothing else changes. They drag it right again — and nothing happens. Block A sits at (3,3). The target nudges, a soft thud, A flashes. No move is spent, because there is no counter to spend. The player drags A up to (3,2), and now the target can advance to (3,3). But B sits at (4,3), and B is boxed in: up is the fixed wall, left is the target, right is D, down is C. To move B, the player has to clear a neighbor first — move D up to (5,2) and then B right, or move C left to (3,4) and then B down. Either way, D has to leave (5,3) eventually, because that cell is the door. The thing standing on the exit is usually the last thing you move.
Nothing in this board can be permanently stuck, and there is no fail state, so restart is a convenience rather than a rescue. That is worth a line in the treatment, because it tells the build team the reset control is not load-bearing.
Now the same art, a different activity. Same five blocks, same colors, same corner radius. The player taps any block, it shrinks and vanishes, a counter ticks up. All five blocks are tappable in any order, because there is no rule about cells and therefore no such thing as a blocked move. When the board empties, the end card appears. Restart restores the five blocks and zeroes the counter.
Mapping both against the five slots:
Sample of the real mechanic
- Starting state: the board drawn above. Five movable blocks — the target at (1,3), A at (3,3), B at (4,3), C at (4,4), D at (5,3) — the fixed wall at (4,2), and the gap in the right wall at row 3.
- Action: drag the target block one cell right.
- Response: it moves to (2,3). Nothing else changes. No counter increments, because there isn't one.
- Next: the target can be dragged again, but A occupies (3,3). Moving A up to (3,2) clears that cell, and the target can advance into it.
- Unavailable action: drag the target into an occupied cell. It nudges and returns; the occupying block flashes; nothing is spent.
Separate branded game
- Starting state: the same five blocks and a counter at zero. No wall, no cells, no exit.
- Action: tap a block.
- Response: it vanishes and the counter increments.
- Next: nothing. Every remaining block was already tappable and still is; no tap opens anything, because there is nothing to open.
- Unavailable action: tap empty space. Nothing happens. That is the entire failure vocabulary.
The exit side separates them too, and it belongs to the ending checklist rather than the five slots. Restart in the sample returns every block to its starting cell; there is nothing else to restore. At the destination, the same rule set continues on a different board — this board, and any sense of carried progress, do not. Restart in the tap game restores the five blocks and zeroes the counter; what continues is the artwork and the name, and nothing the player did in the ad exists anywhere in the product.
Matching artwork does not make these the same experience. One is a small puzzle with the product's rule intact. The other is a themed tap game. Both are defensible; only one of them is a sample.
The simplification that would delete the attraction
Sooner or later someone proposes the clean version: cut the other blocks, keep the drag, let the target slide straight out. Six seconds, no confusion, and it tests better in a meeting.
Run the check. Board: the target at (1,3), the fixed wall at (4,2), the exit at row 3. The player drags the target four cells right and it leaves. What survives? The drag gesture, and the artwork. What is gone? Every decision. There is no obstruction, so there is nothing to rearrange, so the attraction named by the product's owner is not present at all. A player who only saw this learns that dragging moves blocks. They do not learn what the game is.
Note that this is a different failure from the tap game. The tap game was never claiming to be the puzzle. This version is claiming it, using the real mechanic's control, and quietly deleting the mechanic's content. The visual continuity makes it the more dangerous of the two, because it looks correct in a deck.
There is a useful midpoint that gets mistaken for the gutted version. Keep one obstructing block and a two-move solution: the target cannot leave until the player slides something else out of the corridor. That is a real puzzle, small. It has a decision, a blocked action, and a consequence that matches the product. Two moves is thin, but it is thin the way a sample is thin.
Zero obstructing blocks is not thin. It is a different activity. So: restore the constraint, or stop calling it Keystone Line and pitch it honestly as a separate branded game — art, name, and all.
Write the limit into the proposition
A playable treatment should end with one bounded paragraph that someone could build from and someone else could object to.
Keystone Line, sampled. Drag the target block out of the board. The other blocks are in the way and will not move for you. No timer, no move limit. The full game has forty boards and larger sets; the ad shows one board using the same rules. A blocked drag costs nothing. Nothing you do here carries into the app.
That paragraph names the activity, the rule that makes it the product's activity, the scope reduction, and the omissions. It does not claim anyone will enjoy it. It does not claim a format, an eligibility status, a conversion rate, or an install. It does not claim the thing runs.
Those are separate questions with separate evidence, and nobody has any of it yet. What the treatment can establish is narrower and more useful: this activity belongs to this product, for this reason, and here is what it leaves out. A reader should be able to say both halves of that sentence back to you — why this is the right interaction, and what it is not. A screen you can click is not the same as an experience you can recognize.
Frequently asked questions
What is the core difference between a video ad and a playable ad?
A video ad runs on a timeline; a playable ad runs on a state machine. A state machine waits for input, changes something, and then waits again. That creates a question a video never has to answer: the thing the viewer just did — is that the thing the product does? Before anyone draws a frame, settle what the interaction is a sample of.
What are the five slots a state sequence needs, and why does the unavailable action matter?
For at least one path, fill in the starting state, the available action, what responds, what the player can do next, and what happens when they try something the ad does not allow. The unavailable action is where honest treatments separate from sketchy ones, because the player will try the unsupported thing. A linear movie with a destination button does not fill these slots: touching a button that moves to the next scene is navigation, while touching something that changes what the next touch can do is a state machine.
How can a writer tell a small sample from a simplification that changes the promise?
Cutting a board from forty blocks to six is a small example if the blocking rule survives; the player still plans around obstruction and learns something true about the product. Removing obstruction entirely is a removed condition: the drag may look identical, but there are no decisions left, so the ad is advertising a control rather than the puzzle. List what is absent and ask what each absence costs the player's understanding — which decision, constraint, resource, or wait is gone, and whether someone who only saw the ad would recognize the real product.
What ending and handoff details does a playable treatment need?
It needs completion — what counts as done and what triggers the end card — plus reset, any waiting and where it falls relative to the player's action, failure, and handoff. Handoff is what continues at the destination and what does not. If the ad's version had no timer and the app's early levels do, the viewer meets that difference within a minute of installing; if nothing carries over, no state, progress, or board, the treatment should say so.
What does the Keystone Line comparison demonstrate?
It shows that the same artwork can be two different activities. The sample uses the real sliding-block rules: the target cannot leave until the player rearranges an obstruction, a blocked drag costs nothing, and restart is a convenience rather than a rescue. The separate branded tap game has all five blocks tappable in any order, a counter that increments, no wall, no cells, and no exit; no tap opens anything because there is nothing to open. Cutting to a clean version where the target slides straight out deletes every decision and leaves only the drag gesture and artwork, so the repair is to restore the constraint or pitch it honestly as a separate game.