Subscribe
Share
Share
Embed
Accelerate your eCommerce growth and become an analytics and tracking expert. Join Elevar's Founder & CEO, Brad Redding, as he discusses how to address data collection issues that D2C brands face today. Plus you'll learn real world examples of transforming data into insights to help your business grow faster.
We're back, baby. Season 3, I don't know if we have seasons, but we'll go with it, of the Conversion Tracking Playbook podcast. I'm your host, Brad Redding, here with John, the co host. Today, we are talking about 3 things to know about event and match quality scores from meta. I know we both get this topic and question literally every day, all day.
Speaker 1:It's everyone's I I swear everyone has a pull up on their second monitor, and they're just looking at it all day every day, which is totally cool. So let's just get right into it. Event match quality scores, three things to know. So what are EMQ scores? Again, a k event match quality scores inside business manager.
Speaker 1:When you go to your data source, you will see your pixel. And on the screen, you will typically see your list of events. You'll have standard events, like page view, view content purchase. And if you have set up custom events, so it could be a new purchase, a returning customer purchase, maybe you have custom events for category views or gender based activity. You'll see those show up here.
Speaker 1:But in general, you'll see your standard events, where they're coming from. So is it brow client side, a k browser side, or server side, and then the event match quality score associated with this. Now, if you've been in this game for long enough, it could be a week week to week basis where one day it's a score, the next week it's good or great or poor, the next week it's something completely different. We don't know. I swear there's jokes inside of meta where they just want to pull our string a little bit and, I don't know, cause some fuss.
Speaker 1:When you click into those scores, it'll give you a breakdown, which is the image on the left on the screen. And this is where you'll have a little bit more information around this the actual score for that particular event. So in this case, we're looking at a purchase event. This is from an Elovar customer where we have the score of 9.3 out of 10. You'll also see the additional attributed conversions from this event due to the CAPI integration.
Speaker 1:And then this is where a lot of the nuance and questions and back and forth happens, especially in our technical support with our customers is the parameter by parameter breakdown. So email, first name, IP address, phone number, click Id, a k fbc everything in between. So we're gonna peel back the knowledge or the insight and where these come from and why they matter here in a minute. But that's event match quality scores in a nutshell. So, John, let's start with that.
Speaker 1:How are EMQ scores calculated?
Speaker 2:Okay. Well, we I recently learned a little bit when we did some research for this podcast. So let's jump into the learnings. First of all, they're only calculated on CAPI data. So your you know, how everyone's got the redundant client side and server side tracking?
Speaker 2:Well, the client side events, whatever data you're sending on those client side events, those don't get calculated into EMQ. So it's server side only, which is interesting. It's it's CAPI only. It's only over the last 48 hours, and there's a little bit of a nuance here. I don't know if you wanna quickly talk about
Speaker 1:Yeah. Do it.
Speaker 2:So the nuance is if you're turning on a system like Elevar that does a whole bunch of stuff in the background to get more information into these events, the 48 hour time period is probably not the time period you should be looking at to understand if it's doing better or not. You need more time because our systems and competitor systems need time to warm up. So 48 hours is the timeline they do their grading on, but it could take longer to all the information that we have into the system. So just a little side note, don't look after 48 hours. Give it a little more time.
Speaker 2:Amedda does not score data quality. I think that's really that's gonna be pervasive throughout these next couple minutes. They're scoring the number of events excuse me, the number of parameters you're sending with each event. So Brad just showed you a list of a whole bunch of parameters like FPC and phone number and email and all that good stuff. They're scoring based on the number of those parameters you're sending in.
Speaker 2:They aren't scoring on how accurate those parameters are. That's that's a really important nuance. And then lastly, maybe we can go back to that last slide because I think in the last slide, it shows the FBC isn't full. It's a little bit hard to see, but it's really important to note that the number of events with an FB click ID is very closely related to the the amount of traffic that's coming to your site with an FB click ID. Right?
Speaker 2:So if you have a site where there's no ads running, you're gonna have a very low FBC percent, and it's going to affect your EMQ score. So that's I mean, it's within your control, but kinda, sorta, not really. Because if you're not running a lot of Facebook ads, you're going to be limited by what you're scored for EMQ.
Speaker 1:Yep. Cool. That's a little background on how EMQ scores are calculated. Now, let's move on to number 2. What's the purpose of EMQ scores?
Speaker 1:So I have a few slides here to go through this. I'll touch on this first one. This image is from I think Maurice created this from Disruptive Digital in their blog. You can look at this on the Facebook docs, Facebook site docs as well. But, basically, there's a there's a bid formula where your total value, which is how you are going to rank against other advertisers that you're trying to essentially capture visibility and meta serving your ad.
Speaker 1:It's a 3 part calculation. It's your bid times your estimated action rates plus your, perceived user value. So for tracking an emq scores, we're really focused on estimated action rates. If we keep this very narrow to EMQ, if Facebook if we're just sending a 100% of events to to Facebook, but we have very limited user data, so translated low low EMQ scores, MET is not gonna be able to connect as as many of your ad clicks and impressions to user activity on the site. So in this case, this would actually cause your estimated action rate.
Speaker 1:So instead of being, I don't know, an 8, it might be a 2. So you're in this formula, your total value number will actually go down. So that would mean, if you wanted to actually have your ad shown to a particular user, you'd have to potentially bid higher to move that needle. So that's that's a very very I'd say the quickest overview I can give on the improved auction performance, but really the estimated action rates, that's what what we're focused on here, which is a combination of events plus user data or matching quality data. Anything to add to this one, John?
Speaker 2:Just think about this as if you owned Meta and you wanted to understand the behavior of everybody. You would wanna know so you're always receiving events from a whole bunch of sites. You wanna know who those people are on the sites that are generating the events so you can profile them, understand what they're interested in. That's what Meta is trying to do. They're trying to match random behavior that's coming in to somebody they know and aggregate and profile and, you know, remarket and all that.
Speaker 2:So that's what's going on here under the hood.
Speaker 1:Yep. The next one, what's under this umbrella of number 2, what's the purpose of EMQ scores? Really simply, it's to improve ads manager reporting, a k attribution. So this a lot of this does have overlap with sending the maximum amount of events or conversions to Meta. But again, if you just, let's use Google Analytics.
Speaker 1:If you just send a 100% of conversions, but no attribution data to Google, what would happen? You'd have a 100% of conversions attributed to direct none because Google doesn't know where they came from. So that's where EMQ scores has that direct or indirect correlation to helping improve your reporting inside of ads manager. Wanna stay away from any of these third party attribution tools, so this is really focused on ads ads manager reporting. The next one, lower funnel events plus, I'll put this in quotes, mediocre EMQ scores.
Speaker 1:Let's say, scores between 46. John, why don't you take this one?
Speaker 2:The thing here is, does it matter? Like, if I'm not optimizing for page views and most people aren't or view items, why do I care if the EMQ score goes from, like, a 4 to an 8 when you use a tool like Elevar? What's the point? And I think that's a that's a totally fair question, but it just it goes back to what we were just talking about. If you are Meta and you're trying to profile and understand what people are doing, then if we send you an email address on a page view and there's no click ID, so this person's kind of unknown to you, but we're able to provide you an email address, then the profiling becomes easier because you can see somebody's on this on a particular site, and maybe they're interested in a particular product.
Speaker 2:And then, obviously, you using all the crazy machine learning that Meta does, obviously, you're gonna be able to market to that person better. So it really does matter. It matters all the way to the upper funnel.
Speaker 1:Yeah. Maybe touch on the the reality that most advertisers, we won't have high m q scores, because there's usually no reference to PII, and maybe they're they're not as useful. So if you had a page view that had a score went from 4 to 8, is that really gonna be like, how beneficial do you think that'll be? I know this is conjecture a little bit. This is initially the the source of truth, but what are your thoughts on there based on a lot of conversation, the combos that you have with the Meta team directly?
Speaker 2:Yeah. Like, speaking with some of their engineers, it would be nice if we could quantify the answer to this question. I don't think you can. Yeah. But they they go through the exact same logic that I just did, and I'm not sure I can add anything more beyond It's just easier for them to profile if they know who you are.
Speaker 2:And remember, if if somebody comes to a site and signs up for a Klaviyo in a Klaviyo banner or some email banner, most of the time, your meta connector is gonna have no. It's not gonna use that email address. They're 2 totally disconnected systems. So once you start connecting things like Elavar does, we do get to use an email on the page view. I for in prepping for this podcast, I asked a friend at Meta this exact question.
Speaker 2:He said, well, you know, you never really have an email address for a page view anyways. It would be helpful if you did. But, actually, you know, Elovar does enable that.
Speaker 1:So yeah. Yep. And for those that are listening, not what not actually viewing the slides directly, I wanted to point out the note that if you are running campaigns against retargeting audiences, your upper funnel events with, again, mediocre EMQ scores, Those are valuable. And then with advantage plus shopping, where you now can define engaged audiences, that is another layer where this can help as well. And a lot of this information where, again, we work very closely with Meta.
Speaker 1:So we're trying to trying to give you the the latest and greatest insight here. Now, let's jump into the targeting aspect of what's the purpose of EMQ scores. And this one, I'll I'll take the first part and let you touch on the second part. But very simply, we just meta has a very advanced machine, and we need to train meta or tell meta how to differentiate between high and low value value users. So therefore, they are not putting your ads in front of low value users to you.
Speaker 1:And this is again through a combination of 100% of tracked events, plus high and accurate to go back to John's point. We don't wanna just stuff with dummy data and a phone number like a 111 dash 11 because that does no one any good, but high and accurate EMQ scores. So that's my take on the targeting aspect to EMQ. John, what about what about you?
Speaker 2:Yeah. Same thing. It's like, if you look in every events manager for any ecom store, it's the exact same events. So they know that what the funnel approximately looks like. And if they see people coming to your site and abandoning right away, they obviously know that that person's probably not the best to market to.
Speaker 2:But if they see somebody who checks out a product and adds it to their cart and then removes something from their cart and then begins to check it, clearly, that person is more interested. So they take that into account, despite what, like, lots of people think. Those guys at Meta are hyper focused on performance. So if you own the if you own the channel, what would you be doing if you were hyper focused on performance? You'd be doing a little bit of sort of CRO type stuff, and that's what they're doing too.
Speaker 1:Yeah. They want you to do well so you don't pull budget from meta and move it somewhere else. Alright. Number 3. Does higher event match quality scores equal better ad performance?
Speaker 1:John, drum roll.
Speaker 2:Yes and no. Yes and no. If the data that you're sending is guaranteed to be a 100% accurate, then yes. The answer is yes, and the people at Meta will agree with this. But there's a caveat here because if you're sending data that maybe it could be called probabilistic, information that you don't know is for sure true, that can juke the EMQ scores.
Speaker 2:It can make the EMQ scores look awesome, like, maybe get to a 10. But if the data's incorrect, the performance actually can drop, so they could be inversely correlated. Because if you're confusing the system and sending Brad's email on every page view for every single person, it's not going to help with selling to your your, you know, your broad market. That's not gonna do much. But, potentially, it could improve the EMQ scores.
Speaker 2:So it's too bad that EMQ wasn't a measure of match, like a true measure of match. May it it would be nice if it said 80% of the events that you sent were matched to a user, but that isn't what it is. That is not what it is. It's a rating on the number of parameters that you're sending. So it's a proxy for match, but it's not a it's not a match score.
Speaker 2:So you have to be really careful here because I think I think there are companies that are sending, you know, like you said, Brad, 111-1111 for phone number, and then it makes the phone number in the the ratings go to a 100%, but it's not accurate. It you're totally missing the point. So if the data's accurate, it's correlated with ad performance. If the day if your data is not accurate and you're guessing emails and stuff like that, all bets are off. It could be bad for you.
Speaker 1:Yeah. So the the TLDR here is accurate EMQ data leads to matched events, which will help convert help attribute conversions to ads and deliver ads to people who are more likely to convert, which can result in better ad performance and lower cost per action to go back to the bid auction. So there you have it. That is three things about event match quality scores. If you like this, let us know.
Speaker 1:Otherwise, we'll see you on the next episode.
Speaker 2:See you.