TL;DR — Blings co-founder and COO Yosef Peterseil has spent six years building a patented video format, the MP5, that renders a personalized video on the guest’s own device at the moment they click. One HTML template, unlimited variations, roughly 60 kilobytes, and the guest data never leaves the brand’s servers. Habit Burger used it to build what Yosef says is the world’s first video referral program and saw a 42% share rate and a 47% click-to-register rate against a typical referral rate of two to five percent. This episode is about what happens when the loyalty email stops being an email.
Who’s on this episode
- Guest: Yosef Peterseil, co-founder and COO, Blings
- Host: Jeremy Julian, President & Chief Revenue Officer, CBS NorthStar
- Episode: 359 · Runtime 38:08 · Released 19 October 2026
- Topics: personalized video, loyalty marketing, CRM, lifecycle journeys, referral programs, QSR marketing technology
What problem is personalized video actually solving?
Yosef’s framing is blunt, and most operators will recognize it immediately. Guests live in a video-first world — nobody reads the washing machine manual anymore, they search for a two-minute clip — and yet every first-party channel a restaurant owns is text. Email is text. SMS is text. In-app and push are text.
No one’s going into your newsletter and saying, my God, I can’t wait for the next newsletter. This was amazing — because it’s all text and some images.
Yosef Peterseil, Blings
The gap between how guests consume media everywhere else and how brands talk to them in the inbox is the whole thesis. It’s the same gap Brett Linkletter described from the attribution side on Ep. 354 — operators are spending on channels they can’t measure and can’t make feel personal.
How do you render a million personalized videos without making a million videos?
This is the part that sounds impossible until you hear the mechanism. The conventional approach is to pre-render: decide you need a million personalized videos, and generate a million files. That costs real money, the data is stale the moment it renders, and video files are heavy.
Blings inverted it. The MP5 is a single video template built in HTML. You drop it into your CRM the way you’d drop in any other template. Nothing is pre-rendered. When the guest clicks, the video is assembled on their device, pulling live data at that instant.
We can actually create billions of videos in a millisecond, with a file format that weighs 60 kilobytes — and that data is never exposed to us.
Yosef Peterseil, Blings
That last clause matters more than the throughput number. The guest data travels between the brand’s server and the guest’s device. Blings never holds it. For a marketer who has spent the last three years fighting their legal team over data-sharing agreements, that’s often the sentence that unblocks the conversation.
Where should a restaurant start if the CRM data is thin?
Yosef is refreshingly unromantic about this. He does not want you to start with extreme personalization. He wants the lowest-hanging fruit you already have.
Do you have a birthday field? Do you have a first name? That is enough to A/B test a birthday video against your current birthday email. Same for the welcome series, the win-back, the anniversary, the end-of-year summary — all journeys that already exist in every loyalty platform and almost all of which are underperforming.
What he describes next is the interesting second-order effect. Once a brand sees video outperform on the journeys it already runs, it starts asking what else it could be doing with the data it already has. The technology becomes the forcing function for the personalization strategy, not the other way round. That’s the same dynamic Guestologie’s team described on Ep. 344 about turning a first visit into a second.
What actually renders on the guest’s screen?
Jeremy asks the obvious operator question: is this Colonel Sanders on a mic, Shaquille O’Neal from Big Chicken, or a flyover of the menu? The answer in 2026 is “any of them.”
If you have rights to talent, two minutes of them talking can be personalized for everyone afterward. It can be an avatar. It can be two-dimensional animation — burgers flying, the guest’s name on screen. Blings built one template for Meta that ran in 33 languages, with accents and narration varying by geolocation, from a single asset.
Because the video is HTML, it is also clickable and it is live. It refreshes when the page refreshes. Every video carries a KPI — a click, a redemption, a purchase — and the interactivity is designed backwards from it. A welcome series might be “Hi, first name, we have a gift for you. Click to reveal.” The click is the point.
Can the offer change after you’ve already sent it?
Yes, and this is the part that most clearly separates it from an email. A static email is frozen at send. If your Nashville hot chicken sandwich is a ten-day LTO and the guest opens on day twelve, you have advertised something they can’t buy.
Blings is running a campaign with Papa John’s where the offer rotates every two weeks. The video carries a live countdown — open it three days later and the clock has genuinely moved. When the window closes, the same video changes: the offer is gone, here’s what’s next, click to keep shopping. Or it simply swaps in the new offer.
Layer A/B testing on top and the system optimizes per segment — one variant with a click-to-reveal gift, one without, running in parallel, with the algorithm pushing budget toward whichever converts for that population.
Does this work on SMS, or only on email?
Both, and Yosef is careful not to let anyone compare them directly — they’re different channels with different economics and different guest behavior. What he does address is the technical worry every operator raises: what happens when the guest is on a train with two bars?
Blings uses HLS streaming, the same adaptive approach YouTube uses. Load time stays at effectively zero; the quality flexes with the connection. The video always plays.
How do you keep this from becoming AI slop?
Jeremy pushes hard here on behalf of every skeptical marketer, and Yosef doesn’t deflect. He says the fear is legitimate and that a brand which isn’t afraid of it should be. His own example is a nature channel he unfollowed after an iguana briefly grew a second head — once trust is gone on the content, it’s gone on the brand.
If someone sees AI slop, they won’t trust the brand anymore. That’s done. For us, there’s always a human component. Nothing is ever 100% AI, and everything is approved by the client.
Yosef Peterseil, Blings
The working model: AI gets a video to 70–80% quickly — turnaround inside 72 hours — and humans on the Blings team carry it to 100% with client sign-off. Only after every scene is approved does the algorithm start optimizing between the approved variants. The machine picks the winner; it does not pick what ships.
Do you need a studio to make these?
No, and there are three doors in depending on how much of a creative team you have.
- Full control: a Blings extension for Adobe After Effects, for brands with a video editor in house.
- Templates: existing proven templates in the Blings platform — swap the logo, the colors, the copy.
- Paste the email: drop your existing email into their AI and get to 70–80% of a video, then refine it yourself or have their creators finish it.
There’s also a Figma extension, so a design team’s existing work can be pulled straight through and animated into an MP5.
What did Habit Burger’s video referral program actually return?
This is the case study worth writing down. Habit Burger — the West Coast brand now owned by Yum — came to Blings with the idea themselves: turn friend referral into a video.
The flow: the guest gets a video saying they’ve won a free burger. Click to redeem — that’s an API call, and the burger lands immediately. Then: want to share the love? Yes. Because the video is HTML, they type the recipient’s name into the video itself, hit share, and the device’s native share sheet opens — social, text, whatever they use. The recipient receives a video that opens with their own name and says Jeremy just sent you a free burger.
The numbers against a two-to-five-percent industry referral rate:
- 42% share rate
- 47% click-through to register
Yosef’s read on why it worked is worth sitting with: the recipient didn’t get spam, they got a video that someone they know generated for them. Two-sided incentive, personal medium, zero friction.
Signups are up and sales are flat. Does video fix that?
Jeremy raises the complaint that runs through every loyalty conversation on this show — registrations climbing while the top line doesn’t move. Yosef’s answer starts with KPI discipline: a referral video is an acquisition play and should be measured as one. Second and third visits need a different video.
The McDonald’s example: a guest who has made at least one purchase gets a video showing their actual point balance, the products they previously bought, what each costs in points, and what they can redeem right now. Click through and it opens to “here you go” — buy this now, or take this one free.
Just showing a member that you know them. We’re not showing you random things — this is what you like. You’re a McFlurry guy, and this is what you need.
Yosef Peterseil, Blings
He makes an aside that food and beverage operators should take as a compliment and a warning: QSR and fast casual send far less spam than retail, where two emails a day is normal. That restraint is an asset. Every touch point should count. This is the same discipline we covered on smart menus and personalization in Ep. 292.
Does this replace my loyalty platform?
No — and Yosef is emphatic, because it’s the objection that kills deals. Blings sits alongside what you have.
It’s email on steroids. You have email — we’re not going to replace that. You have a system set up — we’re not replacing that. We’re a copy-paste snippet.
Yosef Peterseil, Blings
They’re already an official Braze partner and are working on formal partnerships with Paytronix and Punchh. Nobody wants another line item on the tech stack; the play is to become native inside the platforms operators already run. For the wider picture of how these pieces fit, our conversation with Pairzon on data and AI for restaurant marketing is a useful companion.
Frequently asked questions
What is an MP5 video?
MP5 is Blings’ patented video format. Instead of a pre-rendered file, it is a single HTML video template that renders on the end user’s device at the moment they click, pulling live data at that instant. One template produces unlimited personalized variations, and the file weighs about 60 kilobytes.
Does Blings see my guest data?
No. Because the video renders on the guest’s device, the data moves between the brand’s own server and that device. Yosef Peterseil states on the episode that the data is never exposed to Blings at any point, and that the company is SOC 2 compliant.
What kind of lift should a restaurant expect?
The Habit Burger video referral program returned a 42% share rate and a 47% click-through-to-register rate, against an industry friend-referral rate of roughly two to five percent. Results vary by journey and by channel, which is why Blings A/B tests against the brand’s existing email rather than promising a fixed number.
Do I need to replace my loyalty or CRM platform?
No. Blings layers on top of existing journeys as an HTML snippet inside your current CRM or loyalty platform. It is an official Braze partner and is formalizing partnerships with Paytronix and Punchh.
How much CRM data do I need to start?
First name is enough to begin. A birthday field lets you A/B test a birthday video immediately. Deeper personalization — point balances, past purchases, tier status — improves results but is not a prerequisite.
Takeaways for operators
- Your guests are video-first and your CRM is text-first. That gap is the opportunity, and it costs nothing to test.
- Start with the journey you already run. Birthday or welcome, first name only, A/B against your current email. Don’t wait for a data project.
- Match the video to a KPI. A referral video is acquisition. A point-balance video is retention. Measuring them the same way is how loyalty programs end up with rising signups and flat sales.
- Interactivity is where the lift comes from — the click-to-reveal, the type-a-name-and-share, the in-video redemption. A video that only plays is still a broadcast.
- Keep a human in the approval path. AI to 80%, humans and client sign-off to 100%, algorithm optimizes only between approved variants. The brand-trust downside of getting this wrong is permanent.
- Layer, don’t rip and replace. If it requires a platform migration, it won’t happen.
Read the full transcript
Prefer to read? The complete, searchable transcript of this conversation is here: Yosef Peterseil on Personalized Video for Restaurants — Full Transcript.
Connect with Yosef
- Website: blings.io
- LinkedIn: Yosef Peterseil
Thanks to our sponsors
This episode is brought to you by Restaurant Technologies (RTI), who take the mess, the risk and the guesswork out of cooking oil management, and by Yelp, helping restaurants get discovered by the guests already looking for them.
