# Does your restaurant need a CRM? What a restaurant CRM actually does

> What a restaurant CRM holds, how it differs from your reservation platform, and what it changes for one room versus a whole group.

Author: Maggie Tang
Published: 2026-10-04
Canonical: https://www.loyalist.com/blog/does-your-restaurant-need-a-crm/

## What does a restaurant CRM actually do?

It holds the things a great maître d' would remember if they could remember everything and never left.

Visits and how far apart they are. What someone spends, and on what. The allergy. The anniversary. The banquette by the window, always. The fact that they drink Burgundy and will look at the list for ten minutes if you let them.

On a Tuesday that looks like this. The host opens the book at four and sees that table 12 at eight is a fourth visit, not a first. The server knows before they walk over that this is the guest who orders the same white every time, and that there is one bottle of it left. The kitchen has the shellfish allergy on the ticket, not on a sticky note. The manager has been told that the two-top at nine came in last month, had a bad night, and wrote about it afterwards. Nobody performed a miracle. The information was simply there.

The difference between that Tuesday and a normal one is not effort. Your team already wants to do this. The difference is that the knowledge was in the room instead of in one person's head.

A sales CRM like Salesforce tracks deals moving through stages toward a signature. A restaurant CRM tracks people moving through a relationship, and the thing being managed is recognition rather than a pipeline. That distinction matters when you go looking, because most software called a CRM was built for the first job.

## Is your reservation platform already a CRM?

Partly. It is the first thing worth checking, because you may already have more than you think.

OpenTable and Resy both keep a guest history. They will show you how many times someone has booked, what notes your team left, and whether they showed up. For a single restaurant that runs on reservations and nothing else, that is a real guest record and you should use it properly before you buy anything on top of it.

What it cannot see is everything that happens outside the booking. It does not know what the guest spent, because the check lives in your point of sale. It does not know they left you three stars last month, because the review lives on Google. It does not know they had dinner at your sister restaurant on Friday, because that is a different account. And it does not know they enquired about a buyout in March, because that sits in your events tool.

None of that is a failing. A reservation platform is built to fill tables, and it does that well. It simply sees one side of the guest. A CRM is what puts every side together into one [guest profile](/glossary/guest-profile/).

The test is easy to run. Pick a guest you know is a regular and try to answer three questions from your reservation platform alone. How much have they spent with you this year. What do they order. Have they ever been unhappy. If you cannot answer those, you have a guestbook rather than a guest database.

## What a CRM changes at each size

The answer is yes at every size. The reason changes.

**One location.** You already know your [regulars](/glossary/regulars/) by face. What you do not have is a way to notice the ones who stopped coming. A person who came every three weeks for a year and has not been in for two months is invisible to memory, because nobody remembers an absence. That is the single highest-value thing a CRM does for a small room, and it works the day you turn it on, using history you already have.

**A restaurant that takes reservations and runs marketing.** Here it stops being about memory and starts being about sending the right thing to the right people. A wine dinner email to the 200 guests who actually drink wine will outperform the same email to your whole list, and you cannot build that list without spend and preference data joined to names. The same is true in reverse: knowing who not to email is what keeps a list healthy.

**A hospitality group with multiple locations.** This is where it stops being optional in practice, and it is the case least well served by most software.

A VIP at one concept walks into the sister restaurant and is a stranger. That is the moment a guest notices your group is not really one group, and it happens in front of them, at the door, in the first ten seconds. No amount of brand work fixes it afterwards.

Underneath that there is a harder problem. Each of your rooms is running its own systems, its own tagging habits and its own idea of what a VIP is. One general manager marks people as VIP generously, another almost never does, and a third uses a tag nobody else recognises. So even if you could see across the group, the words would not mean the same thing in each room.

A group-level guest database does two things about that. It [resolves the same person](/glossary/identity-resolution/) across every property, so the count of guests you have is the real one rather than the sum of four overlapping lists. And it gives you one definition of a regular, one definition of a VIP, and one place where those are decided rather than four.

There is a reporting consequence too, and it is usually what convinces a CFO. Once guests resolve across the group, you can see which rooms share guests and which do not, which concepts feed each other, and which location is actually acquiring new people rather than trading the same ones back and forth.

## The guests you already have, and cannot see

You are not starting from zero, and this is the part most operators underestimate.

The guests are already in your systems. The problem is that the same person is in there several times, under several spellings, attached to several phone numbers, and no single system has ever put them together. Your best regular can read as three strangers who each came once.

That is not a theoretical problem. One group we worked with had 3,434 guests who had been regulars and had quietly stopped coming. Not lost to a competitor, not unhappy, just faded. Every one of them was sitting in systems the group already owned and paid for. Nobody could see them, because seeing them required joining four tools together.

When those [lapsed guests](/glossary/lapsed-guest/) are visible, the work becomes obvious. [Win-back](/glossary/win-back/) campaigns aimed at lapsed regulars get between 10 and 30 percent of the people who receive them to book again, with the top of that range coming from the best-performing ones. That is not a marketing trick. It is what happens when you ask someone who liked your restaurant to come back, at a moment you would otherwise have missed.

The numbers around regulars are worth knowing before you make this decision. Resy and Toast's 2026 Regulars Report found that 7 percent of guests can drive up to half of a restaurant's order volume, and that regulars book ahead 83 percent of the time against 48 percent for everyone else. The same report found recognition beats points roughly two to one as a reason people come back, and that only about 30 percent of guests consistently feel recognised.

That last number is the opportunity. Most restaurants are not failing to recognise their best guests because they do not care. They are failing because the information is in four places.

## What actually happens when you turn one on

Worth setting expectations, because the first month is not the month it starts working.

The honest version is that the data is messier than anyone expects. When we started building guest journeys, we assumed the hard part would be the journeys. It was not. It was the state of the guestbook underneath them. The first onboarding we ever ran turned up one guest who appeared six times under three different email addresses, and that turned out to be normal rather than unusual. We wrote about that in [what we learned cleaning hundreds of guestbooks](/blog/introducing-guestbook-studio-what/).

So the first thing a CRM does is not send anything. It reads every connected source and tells you what condition your guest data is in. Duplicates. Five years of free-text allergy notes with no structure. Tags somebody created once and nobody used again. Guests with a phone number and no email, or an email and no name.

Then it cleans, and this is the part worth insisting on when you evaluate anything: a human should approve the cleanup before it is written. Matching two records is a judgment call, and a system that merges guests silently will eventually merge two people who happen to share a name.

Only after that does personalisation work, and this is the order that matters. A CRM built on an untrustworthy guestbook will confidently send the wrong email to the wrong person, which is worse than sending nothing. Expect the first few weeks to be about the foundation. The pay-off starts once it is solid.

## Who owns your guest data?

Worth asking early, because the answer shapes what you build.

Your reservation platform, your point of sale and your delivery apps each hold a slice of the guest, and each has its own terms about what you can export and what you can do with it. A CRM that sits on top of your existing systems gives you a copy of the picture that belongs to you, and keeps belonging to you if you ever change one of those systems.

The short version is that if the only complete record of your guests lives inside a platform you do not control, you do not really have a guest list. You have access to one.

## Sources

- Resy and Toast, "The Regulars Report" (June 2026): 7 percent of guests drive up to 50 percent of order volume; regulars book ahead 83 percent versus 48 percent; recognition beats points roughly 2:1; only about 30 percent of guests consistently feel recognised.
- Loyalist platform data: win-back journeys return 10 to 30 percent of recipients to a booking, with 30 percent representing top-performing journeys.

## Frequently asked questions

### Are Resy or OpenTable CRMs?

Partly. They keep guest history, notes and visit counts, and for a single reservation-led restaurant that is a real guest record. What they do not have is your point of sale spend, your reviews, your events enquiries, or the guest's history at your other locations.

### Do I really need one if I already know my regulars?

You know the ones who are in front of you. The ones a CRM finds are the regulars who stopped coming, who are invisible to memory because nobody notices an absence.

### How is a restaurant CRM different from a sales CRM?

A sales CRM moves deals through stages toward a close. A restaurant CRM tracks a relationship that has no end state, where the thing being managed is recognition. Tools built for the first job tend to ask you to model your guests as leads, which is the wrong shape.

### What do restaurants actually use?

Most full-service restaurants run some combination of a reservation platform, a point of sale, a review tool and an email tool, and call that a stack rather than a CRM. The question is not which products you have. It is whether anything joins them to the same person.

### How long before it does anything useful?

The first weeks go on the guest data itself. Expect the first useful output, usually a list of lapsed regulars worth contacting, within the first month, because that one runs on history you already have.

### When is the right time to get one?

Earlier than most operators think, because the value comes from history and history takes time to build. A CRM connected today can read the last two years of guest activity you already have. A CRM connected in two years cannot recover the two years you spent without one.