New: AI-powered Google Review automation is liveLearn more →
VoqadoWiFi
Use Case

Measure How Many Customers Actually Come Back

The job to be done

Replace a gut feeling about regulars with a number you can compare month to month.

Ask any operator what share of their customers are regulars and you get a confident guess. The guess is usually built from the faces they personally recognise, which skews toward the people who talk to them and toward the shifts they work. WiFi logins give you a measurable version of the same question. It only covers guests who connect, which is a real limitation, but it is consistent, it is yours, and it moves when the business moves.

The four numbers worth tracking

Repeat measurement gets complicated quickly, so start with a small set and hold it steady.

  • Returning share: the percentage of sessions in a period that came from an identity seen before. The single headline number.
  • Median gap between visits: how long a returning guest typically waits. This defines what "lapsed" means at your venue instead of guessing.
  • Visit distribution: how many identities have one visit, two to four, five or more. Most venues discover their base is narrower than they assumed.
  • Cohort retention: of the guests first seen in a given month, what share came back within sixty and ninety days. This is the number that tells you whether a change worked.

How identity is established, and where it fails

A repeat visit is recognised when a new session matches an existing guest record, which in practice means the guest used the same login identity. Hardware addresses cannot carry this job, because phones randomise them per network specifically to prevent silent tracking. That design choice is good for guests and inconvenient for measurement, and pretending otherwise produces numbers that quietly drift. The consequences are worth stating: a returning guest who logs in a different way counts as new, a guest who swaps phones counts as new, and a couple who alternate whose phone connects will look like two occasional guests rather than one frequent pair. Every one of those errors pushes the returning share down, which means your real repeat rate is at least as high as the number you see and probably higher.

Reading the number without lying to yourself

A single month of returning share means very little. The signal lives in the direction of travel and in like-for-like comparison. Compare the same month across years to strip out seasonality. Compare weekday to weekday rather than to weekends. When you change something real, a menu, an opening hour, a loyalty sequence, mark the date and compare the cohorts either side of it rather than the raw monthly figure, because a busy tourist month can bury a genuine improvement in retention under a flood of first-timers.

What to do once you can see it

The measurement is only interesting because of what it enables. A defined median visit gap tells you exactly when a guest has gone quiet, which is the trigger a win-back campaign needs. A visit distribution tells you whether to invest in acquisition or retention this quarter, because a base of mostly one-visit identities is an acquisition-heavy business regardless of how it feels. And a cohort curve turns every operational change into a testable hypothesis rather than an opinion argued about at the end of the month.

Keeping the measurement clean

Three things corrupt this data more than anything else, and all three are fixable in an afternoon. Staff devices on the guest network create phantom daily regulars and drag the returning share upward. A dwell threshold set too low counts pavement connections and doorway pass-throughs as visits. And a portal that makes returning guests log in a different way from last time manufactures new identities out of loyal customers. Fix those before you trust a single chart.

How to set it up

In order. Skipping a step here is usually why a portal ends up showing a spinner instead of a login screen.

  1. 1

    Move staff and devices off the guest SSID

    This is the first and largest correction. Anything that connects every day and is not a customer must live on a separate network before measurement begins.

  2. 2

    Set a dwell threshold that means a real visit

    Choose a minimum session length appropriate to your venue and write down what you chose. Comparisons are only valid against the same definition.

  3. 3

    Make returning logins identical to first logins

    The returning guest should land on the same portal with the same options. Every variation in login path is a chance to create a duplicate identity and understate your repeat rate.

  4. 4

    Collect a baseline before changing anything

    Run at least one full month, ideally a quarter, untouched. Without a baseline every later number is unanchored and every improvement is arguable.

  5. 5

    Pick a review rhythm and stick to it

    Look monthly, compare year over year, and record the date of any operational change alongside the figures. The log of what changed is as valuable as the chart.

Who this suits

The venue types where this job comes up most often, and where the data the portal produces is dense enough to act on.

Coffee shops whose economics depend on weekday frequency
Restaurants deciding between acquisition and retention spend
Bars measuring whether an event night builds a base or just a night
Gyms and studios watching attendance decay after joining
Retail stores with a repeat browsing pattern
Multi-site operators comparing retention across locations

Run this on your own network first

The Starter plan is free forever: one location, 25 logins a month, a branded splash page and consent logging. Growth is $49 a month and covers up to three locations. Works with TP-Link Omada and Ubiquiti UniFi.

Common questions

Is this the same as my true repeat customer rate?

No, and it is important to hold that distinction. It is the repeat rate among guests who connect to your WiFi and log in consistently. Because every identity failure pushes the number down, treat it as a conservative floor and watch its direction rather than quoting the level as a fact about your whole customer base.

How far back can I compare?

As far as your retention setting allows in the product, and further if you export a monthly snapshot. Year over year comparison is the most useful view in seasonal businesses, so start keeping snapshots before you need them.

Why did my returning share suddenly jump?

Before celebrating, check for a staff device that joined the guest network, a dwell threshold that changed, or a quiet month with fewer first-timers inflating the ratio. Returning share is a proportion, so it rises when new visits fall even if returns stayed flat.

Do I need a paid plan for this?

Starter is free forever but capped at 25 logins a month and one location, which is too small a sample to measure retention meaningfully. Growth at $49 a month covers three locations and enough volume for the comparison to mean something.

Read next

These jobs share the same plumbing, so operators normally set them up in sequence rather than one at a time.

Use Case

Get Real Analytics From Your Guest WiFi

Find out how many people actually come in, how long they stay and how many come back, using data you already generate.

Use Case

Win Back Customers Who Stopped Coming In

Notice when a regular quietly stops coming in, and reach them before the habit belongs to somewhere else.

Use Case

Run a Loyalty Programme Through Your WiFi

Recognise and reward guests who come back, without asking anyone to install an app or carry a stamp card.

Terms used on this page

Related pages

← Back to all use cases

Stop reading.
Set it up in an afternoon.

Free forever on Starter: one location, 25 logins a month, branded splash page and consent logging. Enough to see this whole playbook running on your own network.

TP-Link Omada and Ubiquiti UniFi  ·  No card required