What 668 retention audits reveal about dispensary customer data

Steve Riparip has opened the books on 668 dispensary accounts. He runs Tact Firm, a retention agency working across 26 states, and every engagement starts the same way. Before anyone writes a campaign, he goes through the store's customer data and finds out what is actually in there.

He calls it a retention audit. Most operators expect it to turn up a list of campaigns they should be running. What it usually turns up instead is that the numbers they have been reporting on for years are off, sometimes by a lot.

We sat down with Steve to go through what he finds. Five things come up often enough that he now expects them before he logs in.

1. The customer count is inflated

The first thing Steve looks at is how many real people a store actually has. The number in the dashboard is almost never it.

"After doing almost 700 retention audits for stores out there, we see a lot of messy customer data, and oftentimes a lot of customers are showing up multiple times. A lot of these systems don't have really good ways at deduplicating those. It just inflates the amount of customers that a store has actually had."

Someone signs up with a phone number in store, then orders online with an email, then shops a second location. Three records, one person. Anonymous transactions make it worse, because most systems have no clean way to separate them from identified customers in the data.

The operator is not doing anything wrong. They are reading a number their system gave them. The number is just counting the same person more than once, and everything built on top of that number inherits the error.

2. Your loyal customer looks like a new one at every location

This one is specific to multi-location operators, and it is the finding Steve says surprises people most once he points at it.

"If a store has multiple locations, a lot of the other systems will count a customer as a new customer if they've shopped at a store for the first time, not if they shopped at any store. So if one customer shops at three of the different locations at different times, it's going to count them as a new customer every time."

A regular who visits all three of your stores shows up in your reporting as three new customer acquisitions. Your new customer growth looks healthier than it is, and the person getting a first-time welcome offer at your third location has been shopping with you for two years.

Steve's point is that the distinction changes what you should say to them. "It's important to know, are they a new customer at that store, or a new customer to our entire store's brand? Because the messaging matters."

3. The same purchase is being credited to four different campaigns

This is the finding Steve says most operators have never heard of, and it is the one that changes how they read every report they have.

Most marketing systems use broad attribution. If a campaign went out and the customer later bought something, that purchase gets credited to the campaign. Whether they opened it does not enter into it.

"If you sent someone a message and they never knew it existed, it's going to count that purchase towards that campaign, and it's going to give a false attribution that it did something. All these other platforms, they typically do it like that. What I found is that a single purchase from a customer might show up in two to four different campaigns."

Last touch attribution works differently. It asks which marketing touch actually happened before the purchase, using opens and clicks rather than sends, and credits one campaign. Steve is careful about what that does and does not prove. "It doesn't give us 100% assurance on anything, but it gives us really good direction."

It is worth doing the arithmetic on what broad attribution means for a quarterly report. If a purchase is claimed by four campaigns, the revenue attributed to marketing can exceed the revenue the store actually took in. Nobody set out to inflate anything. The system is counting the way it was built to count.

There is a test any operator can run this afternoon. Pick a single customer purchase and see how many campaigns claim credit for it. If the answer is more than one, every campaign ROI number you have reported is built on that.

"In these other platforms, you're guessing, mostly, with the marketing. We don't have to guess. We know very, very exactly what's going on."

4. Duplicate profiles end up on your balance sheet

Most of what an audit turns up is a marketing problem. This one reaches the owner.

Outstanding loyalty points are a liability. Every point a customer holds is value the store owes them, and it sits on the books accordingly. When a customer exists as four profiles, their point balance is fragmented across all four, and the total outstanding liability the business is carrying gets harder to know.

"Especially with point liability, that's on the balance sheet. If any store is getting ready to enter a transaction, whether it's to get more debt or sell their company, and you have these extra points that are accruing there, that's going to lower their valuation because it shows up as a liability."

The same fragmentation shows up at the counter, where it is more visible and less expensive. "They might come in and say, hey, I thought I had more points than you're telling me, not realizing that their other profile had 500 points and the one they're coming in with only has 50."

Steve's word for this is second and third order effects. A duplicate record looks like a tidiness problem right up until someone is running diligence on the business.

It also means deduplication is not a marketing project. It is a financial hygiene project that happens to live in the marketing system.

5. People buy on their schedule, not yours

The last finding did not come out of a report. It came out of a metric Steve says he has never seen anyone else measure: how long it takes between a customer opening a campaign and actually making a purchase.

Across email and text, roughly 2 to 5 percent of people who open go on to purchase within the hour. Another 10 to 20 percent purchase within 24 hours. Everyone else takes one to three weeks.

That changed how Tact Firm advises clients, and it is why Steve now talks retailers out of flash sales.

"Think about the actual end consumer. They have a schedule. They're not gonna leave work early to come in for your flash sale. They're not going to not pick up their kids from school to come to your flash sale. People have these set schedules, and retailers need to lean into those more rather than trying to break them."

The alternative he describes is less dramatic and considerably more effective. Most customers are on a rhythm already, buying every week or every two or three weeks. The work is being relevant and present when that rhythm comes back around.

"Let's be top of mind enough with the right relevant content at the right times, so when they're ready, our store is the first one they think of."

Or as he puts it more plainly: they buy when they need to buy, not when you want them to buy.

What the findings have in common

Five separate problems, and the same thing underneath all of them. The operator is working from numbers that look authoritative and are not quite true. Customer counts inflated by duplicates. New customer growth padded by multi-location double counting. Campaign ROI claimed four times over. Point liability nobody can size. Flash sale strategy built on an assumption about timing that the data does not support.

Steve keeps coming back to the same word for what he is after. "We're always trying to figure out as much truth as possible."

That is the standard he holds a platform to. Tact Firm runs its entire practice on AIQ, and when asked what he would lose first if he had to move tomorrow, his answer was data availability. Open API access, endpoints for everything, a verifiable chain from send to open to purchase.

"That's how we can prove that we're making a difference, and we can also prove that in the bottom line of the retailer."

None of these five findings require a new platform to go look for. They require a system that will show them to you. If you want to know whether any of this is true in your own account, start with the attribution test. Take one purchase and count how many campaigns claim it.