Expanding Your Reach: How Cannabis Lookalike Audiences Amplify Ad Campaign ROI

[ {Introduction}

hello welcome to cannabis marketing live I'm your host Jake litkey today we have H Matt TNA are a uh a frequent guest what is it when like uh on like the TV shows like news they have like someone that's maybe we need to you the robe from contributor like the well like the five time post yeah one of those jackets yeah exactly uh all right well thanks for joining everyone today we are going to be talking about lookalike audiences um and all the things that go into that although before we jump into that given the date here um well uh well I can talk a little bit about 420 um which is usually the largest day of the year um we have a bunch of States uh this year that will be their first year of wreck so people are getting a lot of new customers um Matt we were talking about this earlier you know the 420 is a time when I think new people come in or in frequent Shoppers or people that haven't been to dispensers especially in a new market right so somewhere like New Jersey where you know there was no place to go shopping in state in 420 last year um you've got dispensaries that are getting a lot of new customers um and those new customers even if they come and Shop once start to represent an interesting data set um and yeah go ahead and I think it's also it's also a time when uh stores and dispensaries are um doing a lot of promotions so hey if you sign up for this list you get a uh you know the kind of tongue and cheek joke is you get a lanyard um if you you know if if you buy something or even you go in the store and usually there's some kind of you need to give your email address or your phone number to to get that lanyard or the free lighter or whatever U you if if you're lucky enough to get a little rolling tray or something even more uh useful um hopefully and hopefully this dispensaries are collecting that data and doing some data management um when they get new customers and new people coming to their to their stores so yeah I was just at the uh benzinga conference and I was just listening to I think it was uh guy from flowhub was talking about this um and how they're they're increasing the touch points of where you can get a consumer to give you their email address right so um people are doing things like simple surveys they've got in some cases like a QR code

{How to Create Cannabis Lookalike Audiences}

like right on the checkout or at the kiosk that takes you to a survey um and really trying to incentivize people to to give you their their email address it could be phone number but I want to talk mostly about email address because email address is is a key that unlocks a lot of things in the advertising ecosystem so um now email can be considered personally personal information so one of the things that happens in the advertising industry is you take an email and you hash it right so you run it through a hash function which turns it into a long string of numbers and letters that are unintelligible to humans but act as sort of a Rosetta Stone to on board that data into the rest of the ecosystem and do things like append demographic information or create liik audiences which um we're going to talk about now so what's the uh what's the easy definition of a I mean it's it's somewhat yeah so look like audience is kind of it's defined in the name I mean it it's

{Defining Cannabis Lookalike Audiences}

finding um groups of potential customers that look like your customers and using your own customer data to kind of find that uh group so if you have um you know you have your loyalty program or your email you know list or your your your text list that is a few thousand people hopefully it's more than that um you uh you can do a lot with that with that data and so um usually what

{Legal Considerations and Ethical Practices For Cannabis Customer Data}

happens and Jake talked about you know personal information we call P I it's personally identifiable information um there are state laws um kind of that govern the use of that data and so when somebody gives you that email address um that's a relationship that you have with the customer as a as a dispensary and what so what you do with it is you know governed by state law um but there are ways to utilize that that are legal and that are kind of industry standard and one of them is to to take a look at that list in Aggregate and um use you know like uh thirdparty data processors to who can help you anonymize that data but then to look at it an aggregate against other data sets so at theera we have a uh 260 million person um identity graph that's a that's a database we have every American um every you know adult American on our database of of consumers but we keep um that information um in kind of a in a in a clean room or clean space so we can get customer data match it up against that file and in an anonymous way find uh like personas that then we can pass on to um advertising channels like MediaJel like direct mail and in in in in ways that um keeps the privacy of that initial customer list safe and also protects the privacy of your potential customers yeah and so when you when we

{Propensity Scoring to Identify Similar Audiences}

talk about finding like people um what we're what we're looking at is okay so we these are people that have purchased from you right and and you have their email addresses they can be scored in aggregate to say that there there are some Trends which and that could be across age gender voting habits ethnicity propensity to have dogs there's all kinds of different factors in there there's you know many many dimensions um one of the things that that we have found working with a lot of advertisers is that and this is just human nature because people tend to bias towards themselves just inherently is that a lot of people have some um they don't fully understand their customer they they make some assumptions about who they think their customer is and and okay we live our store is in an affluent area and so our customers all must be high income people they must be high income they must be you know conservative or republican or democrat or whatever it is um and when you look at the data um you get a more informed decision right because I mean I think so that's it's when you so like you know talking about this big list again what what what we actually have is a um small little small little bits of data on um individuals so it's the de demographics that Jake mentioned but also other kind of like other you know consumer behavioral Trends and so when you look at a 10,000 person or 20,000 person list you may think you know as Jake mentioned that um income is a driver or gender is a driver or even location to the you know the dispensary is is is a driver for engagement with that customer um and those things might be true but when you look at the you know we have all this information on on individuals and I I put this string out if you think of like an Excel spreadsheet that's what I'm kind of doing um that information uh is interesting on an individual but it's not really useful what's useful is the aggregate so you look at the all the rows together and what when we're building a lookalike model we're looking for like what pops when say what pops is which of all these attributes all these different data points we have are similar across all of the you know records or or or most of the records and sometimes that is income and sometimes that is location sometimes that is gender but there are things that are uh you know much better to uh to actually like compare across the entire data set than just like the simplest demographics and that's really what look like like models do and then when we score we're we're really we're creating a propensity score so it's the propensity of what is this what is the non-customers propensity to shop at the dispensary in relation to the customer so we know the customer is the yes class they are 100 and so what are the rest of the you know the rest of the surrounding custo uh people around that dispensary we want to reach who should we send that direct mail piece to who should we Target on online who should we you know go after with um advertising and we score everybody from on on a one to 100 Scale based on the similarities uh to their to the known customers yeah so one to 100 um is the scale but generally when we're doing a lookalike model we're looking at numbers above 90 right right um and and the number get smaller the tighter you get so you know the difference and and actually it it's it changes quite a bit so the difference between like a 95% score and a 96% score could reduce the size of your audience quite a bit um and you know there's there's taking the data in okay here's our here's our audience we're going to

{Unexpected Insights From Cannabis Customer Demographics}

score them for their de demographics and propensity and then we're going to go find people in your target market that are not currently your customers that look very similar to them um but there's a middle step that you get out of that um that is pretty valuable which is that when we go through the scoring exercise it actually kicks out a report that that shows you of your customers what the what where they index high in in specific characteristics um and people are frequently surprised by that by those results um actually we did this recently with uh a uh dispensary chain in the Central Valley in California which is largely an agricultural area um and the owners of that were they thought that they were going to go with the we mentioned this before like like a very conservative approach because they themselves are conservative people but if you actually look at the demographics and the voting habits around them it's like 80% of Their audience is actually Democrats right um which doesn't necessarily have a direct correlation to cannabis but it tells you something interesting about your audience um and in this case the other thing that happened was like the majority of their uh customers it's a are Hispanic also right so we actually worked with them like well let's we were talking about 420 420 is interesting um but also Synchro too is interesting in this case for your audience right and without knowing those types of things you don't have as informed of a marketing strategy of who you're trying to reach Market um and so just you know you can look at the demographics that's a lot of people start with this is what I think I know about my audience and then I'm going to go build a Target Persona and that's usually where we start with the marketing campaign right who do you want to Target um frequent well sometimes we get the answer is why I want to Target everyone which is not a good answer um you need to have messages that are tailored um so then the next step is well I think I want to pick you know High income or whatever it is I'm going to pick these things and then I'm going to build that as my targeting criteria um but it really works a lot better if you let the data do the work for you right yeah and it and it's why I kind of when I come on this uh show the show as it is um I kind of like talk about data management and data collection I was in

{Cannabis Data Management and Collection}

a uh on on 420 I went to a it wasn't at 420 it was the week before I didn't want to go in the rush but I it was in the spirit of 420 uh I went to a dispensary here in my state of Maryland and it was an MSO and I got to the front door and they said well you can either give us your information or you can sign in as a guest and I was like well I'll sign in as a guest and they just handed me a blank sheet that and I was like checking off the mistakes and where I was going to kind of take notes and it's like the you know at every step of the way from you know when you enter a store to you know the purchase and the point of sale and kind of syncing all that data and I and collecting the data particularly when you have a promotion particularly when there's a actual gate to be kept that you can't do something unless you give an email address or a phone number to get the discount or or get the lanard um you shouldn't let people off the hook like that uh because then you miss out on the chances to actually do the type of analysis and um the type of effic like bring the efficiency to your advertising uh that the industry really needs um because it's not only about those kind of two things obviously you want to be um getting people to your stores and talking to the right people but you also you don't want to be talking to the wrong people because um it's just it's literally a way of money so so so depending on um simple demographics and depending on kind of guess work uh because of where you your the location of your your dispensary is um is really wasting dollars that are um frankly uh sacred in the you know 280 environment that the industry presides right now so yeah marketing is um unfortunately more expensive than should be due to those tax laws um and another thing to understand here is when when you build a lookalike audience now you you can't just like you obviously can't start sending sending text messages because you don't have an optin but but when you create a look likee audience I mean and this is really what Facebook was great at and and and still is um but that's why Facebook marketing worked so well right which is that they know a lot about every single like and click yeah the person so they have this they have this interest graph and and demographics about people that they know so they're largely doing that work for you which is so I want to Target people that you know match this type of brand they're what they're doing in the background is building you a lookalike audience now Canabis industry is restricted from Facebook so we essentially built the same type of solution now I'm not going to say we have the same level of information as Facebook does um because they've got very very

{Leveraging Cannabis Customer Data for Advertising Strategies}

detailed information and a lot of it but we do actually have quite a bit of data um and it's proven to be very effective when building look- likee models especially when you want to start to get out of just sort of your retargeting retention using your loyalty program to you know send messages to your existing customers um it's a really good way to efficiently spend marketing dollars to find new customers yeah exactly and I mean I think it's like you know kind of success Builds on success and so if you have um processes in your advertising that are dependent on data then you know that front door interaction you know you can kind of breed the um the kind of environment where your employees like actually are wanting to actually don't like like not letting somebody up the hook taking somebody's name and address or email address to do that um and kind of just like trickles on down um because you know obviously look like modeling is is interesting and important that's why we're here to talk about but the most important thing is to actually know your customers and to be able to talk to them and if you have those chances to collect that data and are missing it um you're missing the chance for look like modeling but you're also missing a chance to do another touch to that customer so yeah what are some uh I'm going to put you on the spot here um what what are some examples that you've seen really in in

{Modeling Effective Lookalike Audiences in Any Industry}

any industry where you've seen some particular successes using lookalike modeling let's see I think yeah I mean I I I think when there are um kind of like you know I I I've done a lot of work in politics and a lot of work in advocacy um making making something to be a like a like a big tent issue um kind of issues that um perhaps maybe not now they're benign but 10 years ago 20 years ago soly they were benign um something like the American Heart Association or the American Lung Association and um there are there's success in like people who are year-over-year donors who every Christmas holiday season these people are donating to um their charity of choice um you know having a like a multi-year um record of a donation or multi-year record of an action in like in politics we would call them ones and twos because we kind of rate rate everybody from one to five on the un certain propensities and so looking at their one looking at your ones and twos like your people and knowing um their perally taking action that like those kind of like those kind of lookalike models are the most important I mean are the most effective so like in the parallel in cannabis would be so you know somebody's a customer that's like the great first step but then if you know their wallet size if you know what kind of products they like if you you know that they come in uh I don't know if they're a weekly Shopper or a monthly Shopper um if they come in on holidays if they come in on 420 that second piece of information is really where you can really get successful with a lookalike model so it's not only are they a customer that's great and and customer look like models are great but then the second layer is really what what brings the most success that you know so in politics like it's you know someone's taking an action um it's that that second piece and because it's based on the theory of the best indicator of a future uh behavior is past Behavior so that's true for your a customer and it's it's it's more true and it's a better way to find uh lookalike customers yeah some of the ways that's a good point is is actually segmenting your customer base into categories like

{Cannabis Customer Segmentation Strategies}

people that purchase gummies people that purchase fate people that purchase flowers basket size um is another one as well as frequency um and and then you can tailor your marketing messages around that I mean and there's some simple obvious things right like I mentioned if you're going to break people into you know Edibles vape and flowers three broad categories you you should put them you should create three segments you should create lookalike models for each one right and then when you run your advertising you want to use a creative that's tied to that right so you know you know the product of picture in the creative itself should be an edible for people that are indexing similar to your frequent edible purchasers Y and then your and and your offer for to get those people to come into the store is based on those products rather than just a general you know bottom line percentage off of a whole whole card um which I think is uh kind of you're you're giving away the store there um and then on wallet size or basket size um one of the things that becomes interesting is when you do an offer are you marketing it as 25% off or are you

{Cannabis Discounting Strategies}

marketing it as $25 off a purchase of $100 or more those are psychologically different things right for someone who spends less dollars particularly when the prices are bottoming out I mean mature market right yeah so you can you can say like okay well 20% off doesn't mean that much to someone that's spending $40 um so in that case you you will frequently use like a dollar amount or it's a buy one get one for a penny or some other thing that on you know from a financial perspective when you run the numbers it's actually the same offer right you're giving them the same amount of value but for someone that spends more money giving them a percentage offer and then giving the people that spend less money a similar discount but done in dollars or product uh turns out to be much more effective yeah particularly then if you also take their email address or phone number or some other way of Engagement when they you know if they are to uh cat in that offer it's the the biggest mistake ac across the board that that we're making and I hope nobody is making not making this uh is um having a customer to be accepting an offer uh without them giving you something other than the purchase they should be taking stock in your in in your store in in a in a loyalty program or marketing program yeah another way that you can segment your audience and then Target is people who make online

{Identifying Cannabis Audiences Who Are Most Likely to Purchase Online}

purchases versus just going to the store right so um we work with a bunch of dispensaries that are trying to incentivize people to make an online purchase um because it cuts down on the amount of time that people spend in the store and they can service more customers or in some cases I know we've got company working with where they've got an unfortunate uh freeway construction project happening so their off-ramp is like messed up and so they're trying to get people to understand that they can order online to save time because you know traffic has become an issue for them um and then you can incentivize people will run campaigns that are specifically for that which is like get X offer for online orders only um and and then you can you can then take that you can take the offer group of people that are purchasing online and you can again use look like modeling for people that are more likely to purchase online um and drive people through that funnel uh for the people that are here on Zoom they're um because we're on a couple platforms but I can only see the questions on Zoom if you are um in the zoom thing uh feel free to ask uh any questions if you like and we'll we'll try to answer them I want to throw that out there because we're we're probably going to I think this was scheduled for an hour but I don't think we need an hour to talk about look- likee modeling I think we're we're probably only a few a few more minutes it's it's not very complicated um that it has a lot of interesting applications but the underlying concept isn't isn't too complicated yeah and I think what we should do um I you know as we've been talking I've been thinking about my hand gestures and I think rather than is the international sign for spreadsheet rather well it's like string string string string of data data string um we should put some data viz onto uh next time in the webinar but also as a followup um onto onto the different channels that where people are hopefully tuning in um we can you can put some database on like what a look like model is and the kind of the process of of creating that rather than just lab about it I think there's one thing we didn't cover yet which is there there is while a look Lo look like model is an anonymized audience um and can be used for doing digital marketing there are still some things that you can't do right and some of them may not be obvious um one of them actually this isn't even a cannabis thing because we work in other regulated Industries but real estate um actually has some interesting regulations which is that you cannot Target someone based on their income so that's a data point that's in there but because of the fair housing act that's considered discrimination on income for housing so that means that you if you're building your own lookalike model um or you know we we help our customers with this to avoid those traps but you actually can get yourself in trouble if you are using the data points that you shouldn't be using to create an audience depending on what you're doing with it yeah and I and I would put um voting behaviors in that as well um and like the actual like the actual uh source file voting behavior that we know this person's registered to vote and um but like certain you can't use the voter registration use the vo registration that's not allowed um age obviously comes into play here uh you know one of the benefits of using an identity graph like the one that you have is that we do know people's ages um you want to make sure that when you create a look like audience you're not targeting people that are under 21 or yeah and we have ways when building the look like model to even exclude um the that 18 to 21 segment from the from the analysis at all so not only will they not be in the targeted Universe they will not be included in the analysis um which gets to kind of you know the the compliance with the regulations that we've all uh been been dealt um in this industry yeah and that also will skew that that can skew your lookalike audience as well there's there's another distinction actually that we should mention there are there are demographic and other behavioral information that are tied to a household and there are some that are tied to individuals right so um you really want to be doing individual level targeting for this especially for the Cannabis industry because the it's a it's a pretty personal choice and the other people in that household may not match that um you do there are some weird things that happen with canvas because we know that like there's a weird anomaly of the fact that you have a lot of more males that are shopping for their household so frequently shopping for themselves and then sometimes shopping for their partner or their parents sometimes right yeah there's that still happens like my dad won't go into a dispensary for some reason I I don't know why I guess it's just because seems weird um so I end up like picking stuff up for him right um because he doesn't want to go the dispenser and meanwhile it's like my my dad's like carrying cases of wine out for Christmas out of the liquor store like yeah know um all right so let's see here I think we covered most of our topics we talked a little bit about um some of the other ways that actually I guess there's a couple other things we can talk about you know when you build a look like audience obviously that becomes your audience but you then have all the other tools at your disposal right like geographic history where did they come from um and then we when we're doing digital advertising on the display side the other thing that we do is we we have thousands of Publishers that we can run ads on right so there in any given Market there tends to be a set of Publishers that are like the top 100 Publishers or applications that people are using that seem to be more relevant for cannabis consumers in a given market and so we also optimize on you know where we're serving the ads for a specific brand in a specific Market um so the lookalike model that you make is really just the foundation for you to use and then you you know you want to use all the other targeting um targeting tools at your disposal on top of that right yeah yeah and so knowing the you know the individuals and loc you know their contact information um it it just makes it so much easier to do that it it gives you a little bit more you know you you never able to kind of laser focus but you're turning it from a um you know scatter shot to a more focused scatter shot great I think we've uh I think we've exhausted the look alike all right we go for some look out for some data viz uh on on LinkedIn because I think it's post yeah great uh so uh Matt TNA from sta if you want to reach him uh his phone number is no I don't now email matt.com um and we are coming out with a a relaunch of one of our products which kind of works with data uh data management that I'll be talking about on LinkedIn and other spots soon yeah great uh well thanks for tuning in everyone um Jake liy your host uh see MediaJel we do digital marketing for regulated Brands and do a lot of work for the Cannabis industry thanks for your your time today and reach out uh I am Jake media.com if you want to talk about look- likee audiences or any marketing topics in general thank yeah have a good one by ]