Stories are the one Instagram format where you can see exactly where you lose people. Every frame reports how many watched it, and the gap between frame one and frame seven is a straight line down. Most accounts accept that decline as a fact of nature. A lot of it is actually sequencing.
The unit is not the frame. It is the sequence, and sequences can be built well or badly.
What the numbers are telling you
Instagram gives you per-frame data: how many people saw each frame, and what they did next. The actions matter more than the raw count.
- Forward taps mean someone chose to skip the rest of that frame. Normal. High rates on a frame mean it was slow or the point landed early.
- Back taps mean someone wanted to re-read something. Counterintuitively good — it usually means a frame was interesting and too fast, or the previous frame was clearer than this one.
- Exits mean they left your Stories entirely. This is the number to care about. An exit is not "the story ended", it is "I am done with you".
- Next-story swipes mean they jumped to the next account. Similar to an exit for your purposes.
Look at exits per frame across your last few sequences and find the frames where it spikes. That spike is a specific mistake on a specific frame, and it is almost always one of a small number of things: a frame with no new information, a wall of text, an advert with no setup, or a moment where the sequence stopped going anywhere.
The shape of a sequence that holds
A Story sequence is closer to a conversation than a broadcast. It works when each frame gives a reason for the next one to exist.
A structure that reliably holds attention has four parts:
- Open something. Frame one states a situation, a problem, or a claim. It should not be a title card. "Today: a thread about pricing" is a title card, and people leave on title cards.
- Complicate it. Frames two to four add detail, tension, or an obstacle. This is where most sequences collapse, because people jump straight from setup to conclusion and there is nothing to follow.
- Resolve it. Deliver the answer, the outcome, or the point. Do not stretch this. Once you have paid off, you have paid off.
- Ask for one thing. A poll, a question box, a reply, or a tap. Exactly one, at the end, after value has been delivered.
Four to eight frames is a sensible working range for a sequence with a point. Longer works if the thing is genuinely unfolding — a build, a journey, an event happening in real time — because then each frame carries new information by definition.
A worked seven-frame sequence
Take a baker explaining why a customer's loaves come out flat. Here is the sequence written out.
- A photo of a flat loaf. Text: "Someone sent me this and asked what went wrong." Situation established, no title card.
- "Nine times out of ten it's not the yeast." A claim that removes the obvious answer, which forces the question forward.
- A photo of dough at two stages side by side. "It's this." Visual, no explanation yet.
- "Over-proofed. The structure is already spent before it hits the oven." The mechanism, in one sentence.
- A short clip of the poke test. Showing, not telling.
- "If the dent springs back slowly and stays slightly dented, bake it now." The actionable version.
- A poll: "Have you had this?" / "Yes" / "Constantly". One ask, at the end.
Notice what is not there. No introduction. No "hey guys". No frame that exists only to say a frame is coming. Every frame contains something the previous one did not, and frame two deliberately withholds the answer so that frames three and four have a reason to be watched.
The rules that do most of the work
- One idea per frame. If a frame needs two sentences of explanation, it is two frames.
- Text large enough to read at a glance. A frame is on screen for a few seconds. A paragraph in small type is an exit.
- Keep text out of the top and bottom edges. The interface, your own username bar, and the reply field all sit there.
- Alternate the frame types. Seven consecutive text-on-colour frames feel like a slide deck. Mix photo, video, text, and interactive.
- Put the ask at the end. A poll on frame one is asking for a favour before you have given anything.
- Do not repeat a frame's content in the next frame. Repetition reads as filler, and filler is where people leave.
- Front-load anything time-sensitive. Most viewers never reach the last frame, so if something has to be seen, it cannot be frame nine.
Where interactive stickers actually help
Polls, quizzes, sliders, and question boxes get used as engagement decoration. Used deliberately, they do two useful things.
First, they interrupt a passive tap-through with a decision. Someone who votes has just spent attention, and attention spent tends to continue — people who interact with frame five often watch frame six.
Second, they generate content for you. A question box answered by thirty people is thirty frames of material for tomorrow, and answering them publicly is the single easiest Story sequence there is: question on screen, your answer underneath, next.
What they do not do is manufacture interest. A poll on a frame nobody cared about gets ignored, and a quiz about your own account is a request for effort with nothing offered in return. Ask about them.
How many sequences, and how to sell something in them
Two questions come up constantly, and both have practical answers.
On volume: several short sequences across a day generally beat one long one. Each new sequence re-enters the tray at the front, so it gets another chance with people who were not looking earlier. A single twenty-frame block gets one entry and asks for a large commitment in one go. Three separate five-frame sequences are three openings and three shorter asks.
On selling: the reason promotional frames spike exits is not that people hate being sold to. It is that the promotion usually arrives with no setup. A frame that says "link in bio" after four frames about something unrelated is a hard stop.
The version that works keeps the same structure as everything else. Show the problem, show it being solved, then mention that you solve it. If frames one to four demonstrate the thing genuinely working, frame five can say what it costs without anyone flinching. The order is the whole trick, and it is the same order you would use if you were not selling anything.
How to run this on your own account
- Take your last five Story sequences and write down the exit count for each frame.
- Circle the frame with the biggest jump in each sequence. You will usually find the same type of frame circled repeatedly.
- Name what that frame was doing: title card, wall of text, ungrounded promotion, repeated point, or filler.
- For the next two weeks, plan sequences on paper before you shoot: one line per frame, four to eight lines. If any line duplicates another, cut it.
- Re-check exits. The improvement usually shows up in the middle frames first, because that is where filler lives.
Planning a Story on paper feels excessive for a format that is supposed to be casual. It stops being necessary quickly — after a few weeks the structure becomes habit, and you go back to filming loosely with a better instinct for when a frame is not earning its place.
The underlying idea is simple. People tap through Stories one frame at a time, and every single frame is a decision to continue. Give each one a reason.
