Year-in-Review Brand Videos: Plan the Data Now
Year-in-review brand videos are made in December but built from data collected all year. What to track starting now, formats that work, and the AI pipeline.
Every December, marketing teams have the same meeting. Someone says "we should do a Wrapped-style recap," everyone agrees, and then the room discovers that the numbers it needs — the fun ones, the human ones, the "you ordered 47 oat lattes" ones — were never collected. What ships instead is a beige montage over generic stats: "a big year, thank you all." It gets polite engagement and teaches nothing.
Spotify Wrapped is not a December project. It is a data architecture that runs all year and merely renders in December. That is the actual lesson brands keep missing, and it is why this post is publishing in July: the year-in-review video you want to ship in five months depends on decisions you make this quarter. Here is what to start logging now, the formats worth building toward, and the AI production pipeline that turns the December crunch into a render job.
Why recap videos outperform (when fed real data)
The year-in-review format carries unusual built-in advantages:
- Specificity is the charm. "We shipped 12,438 orders and the single most-returned color was, inexplicably, teal" beats any brand claim you could write, because oddly-specific numbers are inherently credible and inherently shareable.
- It manufactures belonging. A good recap makes customers feel like participants in a story ("you were part of this"), which is retention content wearing a celebration costume.
- December feeds want it. Reflective end-of-year content matches the audience's own mood; the format swims with the current.
- Personalized versions get shared as identity. The reason Wrapped conquers social every year: people share their stats as self-expression. Even a lightweight segmented version ("you were in our top 10% of early birds") inherits some of that engine.
The failure mode is equally consistent: vague numbers, company-centric framing ("our revenue grew!"), and a montage aesthetic. Nobody shares your revenue.
The data to start collecting now
Work backward from the video. The December script wants three kinds of numbers, and each has a July prerequisite:
| Data type | December line it enables | What to set up now |
|---|---|---|
| Odd totals | "You collectively drank 31 tons of coffee" | Tag orders/events so units aggregate into human terms, not SKUs |
| Superlatives | "The busiest hour of our year: Nov 28, 9 p.m." | Timestamped event logging; monthly snapshot exports |
| Streaks and firsts | "417 of you ordered every single month" | Customer-level counters; define "streak" while it can still accrue |
| Community moments | "Your favorite post, by a landslide" | A running folder of top posts, reviews, comments, UGC — screenshot monthly, with permission notes |
| Behind-the-scenes | "We rejected 14 names for this product" | A team ritual: one weird internal number logged per month |
The mechanic that makes this painless: a single shared "recap ledger" document and a 15-minute monthly ritual where someone drops in that month's numbers and moments. Twelve entries later, December's script writes itself from the ledger. Skip the ritual and you will be spelunking through databases during the busiest retail month of the year, which is precisely how beige montages happen. If your metrics live across platforms, per-post engagement history is the other half of the ledger — Versely's built-in analytics keep post-level numbers you can mine in December without a spreadsheet archaeology dig.
Formats, from safest to most ambitious
The stat-card countdown (30–45s). Five to seven big numbers, one per card, typographic treatment, escalating from amusing to impressive, ending on a thank-you. The AI slideshow maker is genuinely the right tool here — generated backgrounds, one overlay stat per frame, music, export as video. Safest format; impossible to do badly if the numbers are good.
The narrated mini-documentary (60–90s). A voiceover walks the year's arc — the launch that flopped, the pivot, the month everything worked — over generated and real footage mixed. This is a multi-scene build: script it as 6–8 scenes and chain them in the AI movie maker so lighting and world stay continuous, with the voiceover generated in your brand voice. Honesty is the differentiator; include one failure and the whole piece gains credibility.
The "your year with us" segmented recap. Not full per-customer personalization — segment-level: three to five customer archetypes ("the early bird," "the bulk buyer," "the launch-day regular"), one short video each, delivered by email to matching segments. This captures a real fraction of the Wrapped share-as-identity effect at a fraction of the build cost.
The community awards show (60s). Hand out invented awards — best customer photo, most creative product use, comment of the year — with a presenter. An AI spokesperson fronts this format well, and a consistent character introduced in December can recur next year. Event-recap grammar applies directly; the mechanics overlap with AI event recap videos.
Whichever format: render a 16:9 hero plus a 9:16 feed version, and cut a 15-second teaser of the two best stats as the acquisition-facing edit.
The production calendar (working backward from Dec 15)
- July–October: the monthly ledger ritual. Nothing else. This is the whole trick.
- Early November: pick the format; lock the 6–8 numbers that made the cut. Write the script while November data is still arriving — leave slots for final totals.
- Late November: generate the visual system — backgrounds, scenes, presenter takes — while iteration time is cheap. Style-lock everything to your brand look so the recap reads as yours, not as a template. (If your visual identity isn't codified yet, do that first — the style-block approach in brand archetypes mapped to video content styles is the fast version.)
- December 1–10: drop in final numbers, regenerate the affected cards or scenes — this is where AI production pays hardest, because a changed number is a re-render, not a reshoot — then sound pass and grade.
- December 12–15: ship, ideally before the feed floods with every other brand's recap in the week before Christmas. Schedule the multi-platform rollout in one pass with Versely's social publishing rather than posting by hand across nine apps.
- January 2: post the outtake — "the stat that didn't make the recap." Cheap second wave from the same ledger.
Mistakes that flatten recap videos
- Company-centric numbers. Revenue, headcount, and funding are your story; the recap must be the customer's story. Translate every stat into "you/we together" framing.
- Round-number vagueness. "Thousands of orders" is beige; "12,438" is content. Precision is the aesthetic.
- Ten stats where six belong. Cut to the numbers that are funny, surprising, or emotional. Impressive-but-expected numbers are filler.
- All triumph, no texture. One honest stumble ("we sold out in March and it took us six weeks to fix — sorry, and thank you") buys credibility for every boast around it.
- Shipping December 23rd. The reflective window peaks mid-December; late arrivals compete with holiday chaos and every other brand's recap simultaneously.
FAQ
When should we start planning a year-in-review brand video?
Now — mid-year — because the video's quality is determined by data you can only collect prospectively. Set up a shared recap ledger and a 15-minute monthly ritual logging odd totals, superlatives, and community moments. The December production itself takes two weeks with an AI pipeline; the data architecture is the part that cannot be rushed retroactively.
What stats should a brand recap video include?
Six to eight numbers that are specific, customer-framed, and at least one of: funny, surprising, or emotional. Odd totals in human units ("31 tons of coffee"), superlatives ("busiest hour of the year"), streaks ("417 of you ordered every month"), and community moments beat any company-centric metric. Precision matters — exact numbers read as credible and shareable where round ones read as marketing.
Can small brands do a Wrapped-style campaign without Spotify's data team?
Yes, at the segment level rather than per-user: define three to five customer archetypes, produce one short recap video per segment, and deliver each by email to the matching group. That captures much of the share-as-identity effect for a build measured in days, not quarters. A single all-audience stat-countdown video is the still-simpler fallback and performs well when the numbers are genuinely specific.
How do we produce a recap video quickly in December?
Do the slow parts in November: lock format, script with slots for final totals, and generate your visual system early. Then December is drop-in-numbers and re-render — with AI generation a changed stat is a regenerated card or scene, not a reshoot. Stat countdowns assemble fastest in a slideshow-to-video flow; narrative recaps chain scenes in movie mode.
Open a recap ledger today, log July's numbers, and when December comes, build the cut with the AI slideshow maker or the AI movie maker. Free credits daily.