Plain English · No code required

How Osbert knows who to call

Every night, Osbert scans every customer at every venue and asks a simple question: who should we be talking to right now? Here's exactly how it works — and why it matters for revenue.

Most venues lose 30–60% of first-time guests permanently. Osbert finds them before they're gone — and tells the team exactly what to say.

The Osbert opportunity engine

The data

What Osbert knows about each customer

Right now, Osbert pulls data from the venue's Square POS system — the same system they already use to take payments. No new hardware, no manual data entry. Every time a customer visits and pays, Square records it. Osbert reads that history every night.

From Square, Osbert builds a profile for every customer: how many times they've visited, how much they've spent in total, and when they last came in. That's enough to spot every meaningful pattern below.

What we read from Square
What it tells us
Number of visits
Are they a first-timer, an occasional guest, or a loyal regular?
Total lifetime spend
How much revenue is genuinely at risk if they stop coming?
Date of last visit
How long have they been silent? Is that unusual for them?
Feedback / reviews
Did they leave unhappy? Do we know about it?

The five signals

What Osbert is watching for

Each night the engine checks every customer against five patterns. When a pattern matches, Osbert creates an opportunity — a card in the dashboard that tells the team who to reach out to and gives them an AI-drafted message to send.

😟
Unhappy customer

A customer left a negative review or a staff member marked their feedback as needing care. This is always the highest priority in the queue — a bad experience that goes unacknowledged becomes a lost customer and a bad review.

Example: A table complained about slow service on Friday night. Their feedback was marked "needs care." On Saturday morning it's the first card in the queue — ready for the manager to send a personal apology before the guest posts publicly.
High-value regular going quiet

A customer who spends significantly more than the average guest hasn't been in for over a month. These are the customers who drive disproportionate revenue — losing one quietly is a big deal.

Example: A corporate client who books the private dining room regularly and spends 4× the average hasn't been in for 6 weeks. Their card appears in the queue with estimated recovery value and a suggested outreach message — an exclusive preview of the new menu.
🔄
Lapsed regular

A customer who has visited three or more times — so they clearly liked it — but hasn't been back for more than two months. These are the highest-conversion outreach targets: they already know the venue, they just need a prompt.

Example: A couple who came in 6 times over spring haven't visited in 10 weeks. Osbert flags them as lapsed, calculates their average spend per visit, and drafts a "we miss you" message with a seasonal menu update.
🤫
Going quiet

A customer who has visited at least twice hasn't been in for over six weeks. Not yet a lapsed regular, but showing a pattern worth a gentle nudge before the habit breaks entirely.

Example: A guest who came in twice in March and hasn't been back since — 7 weeks now. A simple check-in message with the weekend specials might be all it takes to get them back.
👋
First visit — no return

A first-time guest hasn't come back after 1 to 8 weeks. The first-to-second-visit conversion is one of the highest-value moments in any venue's customer lifecycle — most of the guests who become regulars return within 60 days of their first visit, or not at all.

Example: A guest visited for the first time 12 days ago. A warm personal message — "We loved having you, here's what's on this week" — converts first-timers to regulars at dramatically higher rates than any other outreach.

The workflow

What happens after Osbert spots an opportunity

Spotting the right customer is only half the job. Osbert then guides the team through turning that signal into a real conversation — without them having to write a single word from scratch.

1
Opportunity appears in the queue

Every morning the manager opens Osbert and sees a prioritised list — unhappy customers first, high-value regulars next, then everyone else.

2
AI drafts the message

One click generates a personalised draft based on the customer's history, the opportunity type, and the venue's voice. The manager sees the draft instantly — no prompting required.

3
Manager reviews and approves

The manager edits if needed and approves. The whole process takes under a minute per customer. The system records who approved what and when.

4
Message sent

The outreach goes out via the venue's preferred channel. The opportunity is marked sent — it won't appear again unless new signals emerge.

5
Customer returns — revenue attributed

When the customer comes back, Osbert records it. The dashboard shows which outreach led to which return visits and exactly how much revenue was recovered. This is how we prove ROI.

The numbers

How Osbert estimates the revenue at stake

Every opportunity in the queue shows an estimated value — the revenue Osbert thinks is recoverable if that customer comes back. It's calculated from the customer's own history: their average spend per visit, multiplied by how likely they are to return multiple times if re-engaged.

Customer type
How we calculate the opportunity value
First visit, no return
2× their first visit spend — if they become a regular, we expect at least one return within the conversion window
Going quiet
1× average visit spend — a single return is the goal
Lapsed regular
3× average visit spend — proven regulars return multiple times once re-engaged
High-value going quiet
4× average visit spend — the relationship has more potential; the loss is bigger
Unhappy customer
Relationship value — here the goal is recovery, not just a single sale. A resolved grievance prevents a negative review and preserves lifetime value
These are conservative estimates. A regular who's been coming in for a year represents far more than 3× their average visit — but we show the recoverable value to keep the numbers honest. The actual attribution is recorded when they return.

Why the queue is ranked

Not all opportunities are equal

A venue team can only send so many messages in a morning. Osbert ranks every opportunity so the most important ones are always at the top — the manager never has to decide where to start.

The ranking takes three things into account:

Type
Unhappy customers always top the list, followed by high-value regulars going quiet
Silence
The longer someone has been absent, the higher they float — automatically, every morning
Spend
Top spenders get an extra boost — their recovery is worth more to the business

The priority score is recalculated every single night. A customer who was fifth on Monday is automatically first on Friday if they've been silent for another week. No manual sorting needed — the queue always reflects reality.

Quality control

The queue stays clean automatically

One of the things that makes venue teams lose faith in CRM tools is stale data — a customer who came back last week still showing as a "lapsed regular." Osbert solves this without any manual work.

Every night, alongside creating new opportunities, the engine also closes any opportunity whose conditions no longer apply. If a customer who was flagged as "first visit — no return" walks back in on Wednesday, by Thursday morning that card is gone from the queue. Automatically. The team is never chasing someone who already came back.

Common questions

What sales teams ask

Do venues need to do anything to set it up?

They connect their Square account — that's it. Osbert does the rest. The first night after connection, the engine runs and the queue is populated with real opportunities from real customer data.

What if a venue doesn't use Square?

Square is the current integration. SevenRooms is next. The architecture is built for multiple connectors — each one adds a new stream of customer data that the engine can use. This is a roadmap conversation, not a blocker.

How does the manager actually send the message?

Through Osbert's dashboard. The approved message is ready to send via the venue's preferred channel. The outreach is personal — it looks like it came from the manager directly, not from a marketing tool.

Can we show a venue how much revenue Osbert has recovered?

Yes. Every return visit after an outreach is attributed. The dashboard shows total recovered revenue, by week and by month, with a breakdown by opportunity type. This is the core ROI story.

What does "fully idempotent" mean when engineers talk about it?

It means the engine can run twice accidentally and nothing breaks — no duplicate cards, no double messages. It's a reliability guarantee. For sales purposes: Osbert is production-grade, not a prototype.

Want the technical deep-dive?

The full developer doc covers every threshold, formula, and architecture decision — with diagrams.

Read the technical doc →