AI assistant · N5 Complete system
Video Retention Analyzer for Restaurants
Reads your videos' retention curves and tells you at which second you lose people, which part of the script fails — hook, development or closer — and which formula to repeat next week.
N5Complete system
Several chained pieces you install and operate over time, with their review cadence. It changes how the business runs, not one decision.
◆ This assistant 110 of the 164 published assistants are at this levelIs this for you?
What it does and what you get
- Views say how many people arrived; retention says how many stayed — the number the algorithm rewards and the one separating a channel that sells from one that entertains.
- Every video's curve has the autopsy written in it: a drop at second 2 is a broken hook, a steady leak is a development without detail, a collapse at the end is a closer that overstayed.
- This assistant reads those curves video by video, finds YOUR audience's pattern and turns it into next week's shooting orders.
What you end up with
The video-by-video diagnostic table with the leak's second and the failing script part; the account's pattern in three findings with figures; and next week's shooting orders with a per-formula verdict — repeat, fix or drop — owner and retention target.
EXAMPLE This is what you getAn example built with Sazón de Origen, the house's fictional restaurant.Not your data: your result is built with YOUR restaurant's.See the full example →Sample result
The owner asked this: «Here is the data from my last 6 reels. I want to know why some hit and others die, and what I shoot next week.»
| Video | Duration | Ret. 3 s | Ret. 50% | End | Where the leak is | Which script part fails |
|---|---|---|---|---|---|---|
| The sancocho spoon | 24 s | 78% | 61% | 44% | No leak: soft and even decline | None: this is the account's healthy curve |
| Tour of the new location | 58 s | 71% | 32% | 12% | Collapse from second 8 to 20 | Development: twelve seconds of hallway with no food and no people |
| «Hi, welcome to our restaurant» | 31 s | 39% | 22% | 15% | Death at second 2 | Hook: the logo greeting scares people off before showing anything |
The comparison is normalized. The 24 s reel at 61% and the 58 s one at 32% are not read with the same yardstick, which is why the table groups by band before issuing a verdict.
SUPUESTO: no industry «good retention» figure is hard-coded here, because it shifts by platform and by year. The threshold used is this account's own median over the last 8 weeks, and it says so. If your format's real median were 10 points higher, the location tour would move from amber to red and be dropped sooner.
1. A close-up of food in motion holds; the greeting kills. The 3 videos opening with product in the first 2 seconds average 74% at 3 s; the 2 opening with a greeting or the storefront, 41%.
2. Your audience holds for 25 to 30 seconds. The two videos under 30 s keep more than double at the end compared with those from 50 s up; long material is not bad, it is badly cut.
3. On-screen text carries the silent stretch. The only long video that does not collapse carries captions every 4 to 6 seconds; the «clean» ones lose whoever watches without sound, which is most of the lunch crowd.
Hypothesis, still without three videos behind it: closers that say the price out loud hold less than those that only show the dish. It gets tested, not declared.
Illustrative example generated with a fictional restaurant. Not a promise of results, not a client case.
This is 2 of 3 parts. Behind the wall are **the shooting orders** —the repeat / fix / drop verdict per formula, the retention target as a figure, the weak moment each video will fill and the dish chosen by margin— plus the monthly reread checklist that tells pattern from coincidence.
See the full example — free accountWorks with these AIs
How to use it
Which of your data it uses
This assistant works with 8 of the 23 fields in «My Restaurant», among them Restaurant name, Your main guest and Active social channels (with the @ or the link). It does not ask out of curiosity: these are what make the answer speak about YOUR scale instead of an industry average — the same calculation on a business of another size returns a number you cannot decide with. You fill them once and they apply across the library, so the second assistant you open already starts with them in place. Whatever you leave blank, the assistant asks for one thing at a time instead of inventing it, and says so before answering, so you know what it is working from.
Always up to date
Every assistant carries a visible version and date. When an AI changes how it works, the assistant's version goes up and you see what changed.
v1.1.02026-08-25Body rewritten to pay down duelo debt: it scored 63.1 out of 90.0. Two signals read ZERO for the same reason of form, not substance: the trade `lexico`, because the card talked about retention and script without ever naming the table the video is meant to fill —the commercial close the method already promised is now written in its own words: occupancy, covers, average ticket, sales channel, delivery, sales mix, contribution margin—; and `arraigo`, which measured 0 fields in the work sections because the eight «fields» were declared in YOUR RESTAURANT and never came back. The TASK moves from a paragraph reading «One: … Two: … Three:» to six numbered steps at the start of the line, which is the only thing the `metodo` signal knows how to read, and five of them inject a field. The SUPUESTO: line is added so the AI declares what retention it considers good instead of the house hard-coding a figure that shifts by platform and by year. Rises to 95.1.v1.0.02026-08-15Initial version. Born in batch 3 of the artifact titles, opens the Analytics & Digital Growth category, with the full lead-magnet kit and a Sazón de Origen example. Its rule: pattern over anecdote — three videos or it's a hypothesis.More assistants in this category
Included in the always-growing library
Access to every published assistant, adapted to your AI and personalised with your restaurant's data, updates included.
