We are entering the golden age of data mining and analytics. Today, digital tools are catching up to the technology advances; we are closing the gap between online and offline interactions with customers. However the challenge of managing all the data, and picking out the metrics that matter, are overwhelming for organizations.
Thankfully our Mine Your Own Data panelists have come to the rescue and shared their insights from years of experience in data analysis. The panel of experts included Rachel George of HARBISON and Joseph Sauro of Direct Agents and Terry Rice of Brooklyn Digital Marketing and instructor at General Assembly. If you missed the panel, you are in luck because here are the key takeaways from the discussion.
Make Sure Your Data is accurate
Before you start driving into your google analytics account, make sure your data hasn’t been tampered with. If your Google Analytics has been set up incorrectly, you could be making business decisions on biased data. Terry Rice recounted how one of his clients had 20% of their website traffic was coming from their office IP address, so their employee’s traffic was included in their customer traffic. Think about how much time your employees spend on your site compared to your customer. It all adds up. By adding a simple filter to your Google Analytics account, you can ensure that your customer traffic is the majority of your data analysis. To add to the point, Joseph Sauro recommends his clients look into where site traffic is coming from. Without a deeper look, your site traffic could be included in automated by a bot. In other words, your site is getting spam. Set up your data tools with filters and parameters to ensure that you are analyzing clean data from your website.
The Rise of Mobile Analytics Tools
Also her role at HARBISON, Rachel actively consults with startups and brands on their digital endeavors. Over her seven years in the digital marketing field, she has seen the increasing need for mobile data. According to KPCB mobile technology trends, mobile digital media time in the US is now significantly higher at 51% compared to the desktop (42%). With the rise of mobile-first companies, there is the increasing need to track users engagement within mobile apps and website traffic. As the industry has been catching up to the mobile data, she recommends Localytics, AppBoy and Flurry as mobile analytics tools. Clearly, it is a good time to ensure that invest in mobile analytic metrics and data point as mobile usage increase.
With the growing public concerns over privacy, all the panelists emphasized the need to build trust with your clients. As Terry highlighted, if your customers don’t trust your website they aren’t likely to buy your products. He recounted a time when client’s website data showed a trend of customers adding products to their shopping cart and then clicking on the company’s about page and then abandoning the cart. In this example, the client was checking that the company’s legitimacy before making a purchase.
Your customers are humans, not just data points
When making a decision based on data analysis, companies should still use their judgment and common sense. Although all the panelist is data advocates, they didn’t want companies to follow data analysis blindly. As Terry stated “Don’t think about metrics, think about people. Data is a proxy to your customers” to the audience. Your customer experience should be at the forefront when making a business decision and not data.
We would like to thank all our panelists for sharing their advice and experiences with our audience, and to Techspace for lending their conference room. Lastly, we would like to thank our audience for attending the event. As an attendee, you’re entitled to receive a complimentary e-commerce consultation on your current website from the Verbal+Visual team. If you missed our panel, don’t worry, We will be hosting another eCommerce panel next month. You will have to wait until October to find out what the hot topic will be. See you next time!
Anshey: Thanks everyone for coming today, much appreciated. Welcome to the ‘Mine Your Own Data’ panel. This is our third panel that we’re hosting. We feel like we have got in the hang of things at this point. We’ve had some really good panels in the past. One of our former panelists is here in the audience as well, which is great.
So I want to thank everybody, my name is Anshey. I’m the CEO of Verbal Plus Visual. We create digital experiences for challenger brands in the retail space. So a lot of ecommerce, as well as a variety of different digital experiences and digital platforms.
In the past, we’ve worked with clients like Citizen Watches, Wacoal Lingerie, Supima Cotton, and many more. And then for anybody here, or for anybody who is watching online, we have one hour of complimentary audit time for anybody who is here or on our website so we’d be happy to chat about that at the end of today.
So, the other thing I wanted to bring up was our hashtag for today is #MineYourOwnData so if you’re on Instagram or Twitter please hashtag us so we can get the word out about these events. We definitely want to grow these events into bigger and better events as we continue to grow. And then the last thing that we want to do before we start is thanks to TechSpace, which is the office space you’re in here. So if you’re ever looking for office space, give me a shout and we’ll get you connected with the TechSpace folks.
So why don’t we hop right into our panel. Thanks again everybody for being here and a special thanks to our panelists as well. I’d like to start with a question for each of our panelists. They’ll introduce themselves and answer the question.
So the first question is- “What is the most important data point to be aware of in targeting consumers or customers on your digital platforms?” I’m going to start with you Terry.
Terry Rice: Alright, so my name is Terry Rice and I am a digital marketing trainer and consultant at Brooklyn Digital Marketing. That’s my company and it’s really located in Brooklyn.
In addition to that, I also teach digital marketing classes at general assembly here in New York.
So to answer your question, for me, the most important data point is CPA, meaning cost per acquisition. Therefore, from that they know well, alright, this audience is coming in at a price point that’s aligned with my goals. Now could I find more people this? Or is this one didn’t work out so well I should probably stop doing that- not a good idea. It’s so long as your tracking that, you can course correct and continually optimize your campaigns. So for me, that’s the most important one.
Joseph Sauro: I’m Joseph Sauro and I am the analytics team lead at Direct Agents, a direct response digital marketing agency. I would have to say that the most important metric to look at for this would kind of correlate to Terry’s over here with conversion rate. You really want to make sure that you’re targeting the user group that is most likely to convert which ties into CPA then. So making sure that it’s a highly runned group will definitely go a long way in driving the digital marketing campaigns.
Rachel George: Hi my name is Rachel George. I do start up consulting in the- I work with a bunch of different startups in the marketing spaces focused on mobile Google analytics as well as brand and product management. I would say from the mobile perspective, it’s kind of an interesting one because there is definitely a focus on cost per acquisition and conversion rates but it’s really sort of a two fold mixture. One with getting them to download your application and engage with you in the first place and then measuring how they engage with your application through to the point of a full conversion within the app.
Anshey: Great, thank you everybody, welcome. So I’d like to start with Google Analytics because that’s a very common topic, of course. So Terry, what are some of the more advanced methods and aspects of Google Analytics that most people working in GA miss?
Terry: Yea good question. So yea, with that I think a lot of people use Google Analytics out the box but essentially there’s three things you can track with that. One is conversion, meaning someone hit a URL. Another is time on site and another one is pages viewed. I think if you go a step further in actually starting tagging events, such as a button being clicked or a video being played, you can find how that correlates with someone making a purchase, for example, if someone watches this youtube video on my landing page, are they more or less likely to convert versus people who have not. And you usually make a segment of these people, should you choose to do so, look at the different conversion rate, cost per acquisition, things of that nature and see if this interaction actually benefits your site, or maybe it takes away from it and then from there you know how to further optimize your website experience.
Anshey: Great, Joe do you have anything to add to that? I know you work with Jay all day long so.
Joseph: Yea exactly. So, I find the common one really lies in utilizing the admin sections. And that is done in a variety of ways that can really help you bring together your data points so they are easily analyzed. Many of those align to make sure that your traffic is being grouped into the correct channel so when you’re making an analysis you really do know how each of your marketing channels are performing and they don’t have miscategorized information.
Another one is really cleaning up your page URLs so that you don’t have extraneous parameters that don’t really add any impact full of insight to you as an analyst but maybe they’re more there as a back end tracker that doesn’t play into that. So really cleaning that up so it doesn’t split your data up into one page into ten different rows of data that you have to combine it yourself. You could place specific roles on there and filters that will bring that data together into one seamless row so it will be easier, more easily analyzed.
Anshey: One thing I’ve seen a lot in Google Analytics- you know we worked on a lot of different sites and I don’t know if any of you have seen it, is spam. And that can really infiltrate the data and mess up what you’re looking at from an analytics perspective. Have any of you experienced that as well in the past?
Terry: Yeah, I have. Has anybody not?
Terry: Spam free! So yeah, I absolutely have and if you can identify the common source of that spam, you can actually set up a filter to block that out. Within your google analytics, when you set it up you can hit a button that says ‘block out bots’ like known spam. So that can help with that because it does throw off your data and I think with those filters one thing people don’t think about is to actually filter out their own internal data, meaning the data you’re generating into your site all day while you’re on it. You know I was working with one of my clients and I said to them, you know doing the math, it looks like about 20% of the data that you’re looking at an analyzing is actually internally generated. She was like ‘wow this person is spending a lot of time on the site’ and I’m like ‘that is you!’ So what you want to do is go throughout your own internal data by looking at your IP address or the IP address for your company and blocking it out because you’re looking at a lot of false positives which can totally skew your data.
Joseph: Going off of what Terry said with that is good practice too, just to have multiple views. So you have your master view which is all your data points coming in, you have your external view which is everything but your internal tracking and then you can have internal tracking too as well to ensure that your data is split up that way and you are not analyzing those false positives that are there. As well as the spam traffic, it’s always good to see and isolate where exactly is that spam traffic coming from? Is it coming from a specific browser, a specific operating system, or a combination of the two? And that can really help you dive in and isolate that so you can go ahead and exclude that IP.
Anshey: Joe, have you seen data used to predict customer behaviors and increased conversions?
Joseph: Yea, definitely. It really comes down to, the example I’ve used, comes down to user experience and what users are interacting most with on the site. I’ve had clients where they had a really small search bar on their site. And when looking at the google analytics data, you can see how people are interacting with search, people who utilize search and people who do not. You see that those who did have much higher AOV and a much higher conversion rate than those who did not. So with that, recommendations going into place on how to redesign the header of this ecommerce site, suggesting that you have much larger search bar that goes pretty much 75% of the way across the top of the screen and then when the user scrolls down you have a consolidated version of that header so the search bar is always visible to the user, whether they’re at the top of the page or the bottom of the page. And with that, we saw much more engagement with search bar as well as revenue being driven through that, which increases the bottom line.
Anshey: Before we get to Rachel and Terry, can you explain what AOV is?
Joseph: Sorry, AOV is Average Order Value.
Anshey: Great, sounds good. Rachel would you like to add anything?
Rachel: Yeah, I was going to say the exact same concept that applies to mobile space, so any points within your mobile application that you want to test you can set up AB testing to figure out which ones have the most positive impact on the raw flow, the average order value, the conversion rate, all those types of things. But essentially, constantly testing your assumptions and developing different kinds of theories and putting those to the test can really help you maximize the overall user experience.
Anshey: Terry, do you have anything to add to that.
Terry: Yea, with that, I think it’s great to know data, but I think you don’t want to remove the human element from the equation. And by that I mean for example that I’m an eCommerce and I have a recommendation engine on my site that says ‘people that bought jeans we’re more likely to buy this product.’ Well it’s probably going to be another pair of jeans, right? So that in my decision making process with consumers were ‘well I had this pair of jeans in my cart now should probably check out but now they’re showing these other ones so now I’m kind of thinking about this’ and now I’m going back and forth and now the doorbell rang and I abandon my cart. Does it make more sense to say the people who bought these jeans might need something else? A Shirt, a belt, something that’s a compliment as opposed to a substitute? Because essentially you’re competing with yourself now when you’re on your own site and you want to take a look at that and say is this a good idea or a bad idea and don’t let the machines take over essentially.
Anshey: Yea we don’t want to terminate our situation.
Anshey: Sounds good. One thing that we hear a lot about is the term big data and Terry I know this is one of your areas of expertise- can you explain the difference between big data and little data and what that all entails?
Terry: Yeah, so I mean there’s not necessarily a litmus test for big data versus little date. Sorry all data. But essentially very large streams of data that you might have have on the server to actually manipulate as opposed to popping it into an excel. So we’re using sequel or R to actually manipulate the data. That would be my original definition of it. But I think some people feel like because they don’t have large amounts of data they can’t analyze something. That doesn’t mean they should be quick to give up. Start from somewhere, right? The data you have coming in you can still see what’s the optimum experience? You’re going to look at exposure events, conversion events and you’re going to want to reach statistical significance, like a percent confidence in saying this version works better than this version. So yeah, I mean if you’re not using some cool programming which you still could have math, which could be a lot longer time if you’re dealing with big data, so use that to your advantage to inform your decision making process.
Anshey: Great. Rachel, can we chat a bit about what methods you use on mobile to segment data in targeting users on a more granular level?
Rachel: Yea, sure. So I think, well this is an offset, what’s interesting about the mobile space is that you can actually pull all of the data points and pin it down to one individual consumer. So if you use a tool that gives you insights into the general app analytics and visits and higher level insights, but you also use a tool that gives you more granular, sort of a CRM system almost where you can have a demographic and a second graphic information about each individual user, you can then start to do really interesting segmentations. Giving you, instead of a very robust segmentation engine with lots of filter options and rules around it will allow you to narrow it down to let’s say women who are between 25 and 35 in New York City that go in a certain geographical area and also have visited your app within the past 3 days and made a purchase in the last 2 weeks. So you are able to narrow it down to really interesting segments like that and do a lot of exciting stuff around push notifications and rewards, offerings and other things to promote, whatever your end business goal is for that application.
Anshey: And backing off of that, what are the tools you use? And I’d love to hear from Joseph and Terry as well- what are the tools that you use in your day to day use within the data analytics phase?
Rachel: That’s a great question. There’s actually a ton of tools in the space now so it’s kind of crowded, but I’d say if you’re looking at picking a a mobile analytics tool the most important thing is to figure out who’s going to be using it in your organization and what you want to get out of it. So certainly, a lot of the first to market in the space tools are really focused more for the data science crowd. Now there’s a lot of tools that are available to google analytics that give you a high of robust data options but also more user friendly and give you those marketing actions that you can take immediately. It really depends what you’re trying to get off of it.
Anshey: Can you name some of the platforms, if you don’t mind?
Rachel: Yea, sure. I think I would say Googlytics and AppBoy are great if you want to be able to see a lot of data and also take action from a marketer’s point of view. If you’re specifically focusing in pushing out lots of messaging, Urban Airship is great for that. If you want to dive deep on data, Mix Panel is obviously a great one. If you want to pull third party data, Flurry is a good one to use a lot of page sources. Also Google Analytics and Omniture if you’re already working within those platforms.
Rachel: Lots of options.
Anshey: What about you, Joseph?
Joseph: Yeah, a little bit of overlap here, but Google Analytics is definitely one that lots of you are probably already using. Also Adobe, Omniture, Psych Catalyst, whichever one you’d like to call it at this point- they tend to change it quite often. On pairing with that, Adobe Audience Manager definitely helps, especially helps you if you’re going into that programmatic space and really isolating those high heavy engaged profitables as well as understanding which segments aren’t as engaged for you and finding out a way to reactivate them and to turn them into converters and be those high profile audiences for you.
Terry: Yeah, I agree with your recommendations. And speaking from the Omniture that we stand for- I used to work there so I have to say yeah it’s a great platform. It’s also somewhat cost prohibited as well though, so the reason why I like Google Analytics a lot though is because everyone has access to it, which is great because then if everyone has access to it, there’s a lot of user generated content around how to do it, right? So if it’s a larger community of people who are going to be filling out help documents, Reddit and Tumblr, forms like this, helping you if you have a challenge as opposed to waiting for this company to get back to you. So there’s definitely a benefit to having a large audience to pull information from.
Anshey: Great. Joe- how have you used data in advanced ways to optimize digital ad spends? Obviously there’s building the website and optimizing the website, but then there’s marketing for the website. I’d like to move onto that topic a little bit more.
Joseph: Yeah, a great way to look at that is to understand what is the correlation between your marketing channels and your efforts? Is there a positive or negative correlation and is it significant too? Because we want to make sure that the decisions we are making, especially when it comes to ad spend, that they are actually statistically significant decisions. So, a way to look at then with an example there is seeing is there any correlation there for when an email goes out to your users, are you seeing an influx in paid search traffic? If so, you need to really make sure you have cohesive messaging across your initiatives there so that your email message that moment users are targeting that message and that it’s just driving home for them to increase their conversion rate in the long run.
Anshey: Great. I think we’ve heard statistical significance a few times here, but I’d love to go over that a little bit. For each one of you, I’m sure there’s some different meaning for that. I’d love to hear what your feedback is on achieving statistical significance and when data becomes valid, so to speak.
Joseph: For statistical significance, it’s important that when you’re measuring that 1.) you have a large enough sample size and that is a key part there. There’s many times that you see your data and think it’s significant after one week, but after two, three, four weeks go by, you see it flip on you. Also it’s important that when you’re doing these tests and running these optimizations, that you really have control over your test environment. You’re measuring one variable at a time, or different interactions amongst variables so that you can really isolate what was the impacting cause of this change that I am seeing and then how can I leverage that to increase profitability.
Terry: Yeah, I’d agree and with that, it’s important to test elements at the same time whenever possible to get rid of any kind of noise in the data. So that this week goal trying to hit a next week goal, you had a super bowl commercial next week and that’s why numbers are looking so different. And when it comes down to the actual math of it, we should have that 95% confidence that we talked about so much. I typically use an online calculator before I put in exposure events and conversion events for two different experiments and then see which one won. So I sound mathy because I go ‘Oh yea I got 85% confidence’ but I’m really just entering some numbers. But why it’s important to know this is because people are going to come to you, you know, a client or a stakeholder and say, ‘when are we going to know when are we going to know?’ and next week, you know, next Friday, I don’t know. I need data, not time and to predict how long it’s going to take to predict that data. But I can’t tell you by next Friday for sure because I didn’t have enough exposure events and conversion events to make a call. So you honestly need to get comfortable with pushing back math instead of kind of fumbling with that pressure and saying next Wednesday I’ll let you know.
Anshey: Fair enough. Rachel, how about you?
Rachel: You know, I go with what these guys thought about it. I think this is sort of a side bar, but one interesting thing I found with working with social media analytics that’s always interesting to keep in mind is I found that in that space a couple of years ago, generally most tools were flawed in some way. It comes a point of when I, for me when I’m looking at data where if there’s some sort of underlying issues that might be concerning for me, I don’t feel that I always have to necessarily take something else’s statistical significance. I can take it at directional and that might have an impact on high maintenance decisions so that might not personally go into it with a full amount of confidence but I do say, well directionally the data is leaning towards this way so it’s just an additional insight when you don’t have the option for statistical significance.
Anshey: Right, so it really is a mixture of time and actually having the amount of people you see using the service, right? So it is both. Would you say it’s one over the other? It’s definitely the number of people, I’m assuming?
Anshey: And we’re back to users to achieve that?
Anshey: Cool, sounds good. So for everybody, you know, where do you see the future of data and all of this going?
Rachel: Gosh, I think it’s really exciting to think about the possibilities of where we’re going. Across the board, I think it, data in general, enables so many opportunities in the future that are going to impact every element of our lives. I as a marketing consumer, I love that data is becoming more and more centric to the way that companies make decisions. For me personally, I leverage mobile far more than any other channels, so I love being able to be targeted for personalized notifications and relevant in app alerts and constantly being able to passively consume all the information that I want from every company that I’m in contact with so I hope and I look forward to the ability for data to make my life more efficient and overall I think improve it in certain ways.
Terry: I’d agree, especially on the mobile standpoint because, you know, for anyone who does retail and ecommerce like the holy grail is like online and offline, right? Like how do you actually improve and join? And with Google and now Facebook they tried searching map ads. Okay say someone saw an ad on mobile or on desktop at home, or who are now tracking people who are doing store visits and then saying ‘okay they were near the register for a long time, they probably bought something’ right? So I’m making that join between online and offline is this holy grail we’re all searching for and it’s somewhat being solved more efficiently now.
Joseph: And to piggyback off of that, with the future of data and analytics is I’m excited because we’re continuing to improve. We’re filling those dark spots where previously we weren’t able to track it such as the online offline as you mentioned. And we’re going into new spaces as technology progresses, such as going into connected TVs, which weren’t even a thing years ago. But now, that’s a new era that we’re able to track and see how users are engaging with apps across the board desktop, mobile and it’s an exciting time.
Anshey: Great. And with all of these- go ahead.
Rachel: Oh I was saying yeah.
Anshey: Just yeah. So with all of these advancements, is there any major concern- and we see this with consumers all the time about data security- both on the consumer side as well with the folks that are in charge of that data.
Terry: I’d say yeah. We’re concerned.
Anshey: What are things that are most concerning that you’re seeing for consumers out there and some of the things that brands really have to become synonymous especially in terms of protecting their data?
Rachel: I think on the high level of this stuff is that I think brands are just starting to understand that they need to establish that security relationship with consumers. So, basic stuff that’s putting in things like privacy notices, making, with mobile applications when they download their app, making the policies that you agree to a little bit more clear to see what you’re standard consumer is agreeing to this and that brands treat it with the reverence that your personal data should be treated.
I think that- I’ve seen this in a couple of different spaces, especially within the healthcare space, there’s so much aversion to digitalization of a person’s components in a sense, and it’s really scary for a lot of consumers to think about companies having multi data points on them and how this is going to be used, but also is that company, are you actually trusting the company enough to care and taking really strong measurements to make sure things are secured and locked down. I think that it’s- on the company side, it costs across the country we have a lot of work to be done to figure out what it looks like and how to make it secure and how to communicate that to consumers and we’re going to continue building those brand relationships.
Terry: I’d agree, and I think part of having empathy for the whole customer online journey and with that I’ll give an example- one of my clients, an ecommerce retailer, they were having challenges with their conversion rates and we actually mapped that path to conversion by Google Analytics, we actually noticed people would, you know, put something in their cart and then they would go to check out. And then at check out, they would go to the about us section and a lot of them ended up leaving. And I was like ‘that’s kind of weird, why is someone going from check out to about us?’ Trust- They didn’t trust the company so they were like ‘let me do a little background on you guys first and maybe they could have optimized their about us section a little bit better so that those people didn’t leave. Obviously, what we did was we improved the whole check out flow by saying yes, we are better than this rating, and we’ve been in this magazine, and this is a secure transaction- all of those things. So I have an empathy for what the person is going through and this is explicit data, this is not implicit data, or I’m sorry this is implicit data, not explicit data. They’re saying no, this is why I’m leaving, but you kind of have to keep that human element involved again you know, just think about what someone is going through.
Joseph: Yeah, I agree. Especially when you have one bringing forward to the consumers when they’re making a purchase. But also backing able to back that up on your end to ensure that none of their personal items or identifiable information will be leaked. So definitely those security precautions need to be made ultimately on both ends, the back end and on the frontend for the customer.
Anshey: Google is putting into play secure certificates being pretty much mandatory at this point which also helps, so making sure that security is really being upped by all these platforms along the ways is quite beneficial, I’m sure. Are there any other final points or final takeaways from today’s chat that you all would like people in the audience to go home with?
Joseph: We talked a lot about different types of data and how you use it, but I think the most important part of this is before you even go in and start analyzing your data, you need to make sure that it’s correct. Because you’re talking about all of these different ways to slice and dice it and what you can do with it, but if you’re collecting- if your data collection is flawed and there’s an error in it, the decisions and insights that you’re drawing from it could prove the adverse effect in what you’re looking for. So it’s always important to really vet it out and make sure everything is tracking as it should be.
Rachel: I was going to say that in mobile space, there’s so many things that you can track. If you have a mobile app, and that process can be- it’s a great process to go through but you can go through and you can track every single possible data point in an application, it can be really overwhelming initially so I think that for starting out and in general what’s really important to keep in mind if you really want to leverage your data is to understand what are your business goals? What are your key- what are your KPIs and what is it that you need to improve in this application to really show your success? And from there, build that down into what do those data points look like? Like how do I want to bring those together and then implement your data plan? That can make a huge difference in your efficiency going forward.
Terry: Yeah, I’d agree with that and I think overall we need to remember that data is a proxy for people. So when you look at your bounce rate, that’s a human going to your site and then leaving. Something is wrong- fix it, right? If you know it’s in your conversion funnel, people go to the shipping page and they leave, well maybe you charge too much for shipping. Or maybe you re-target them with an ad that says free shipping. So use that data to better understand the person you’re going after and just don’t think about metrics right then, think about people. I think, again, keeping the elements of the whole process as well.
Anshey: That’s a great point. Well thank you everybody. Next, we have a Q&A for anybody that has any questions.