Annie Cushing

About Annie Cushing

Annie Cushing lives for data. By day she gathers it, gives it a good home, and makes it runway ready. By night she dreams about her next vis fix. The only thing that thrills her as much as getting knee deep in data is teaching other marketers how to sexy up their data. But she sometimes she takes her sexy data analogies too far.

In her Learn Inbound talk, Annie will explore how as marketers, too often we keep the data we analyze locked away in bubbles. Resigned to a life of solitary confinement, the insights we could gain from taking a more holistic approach to data collection are lost in our departmental silos. Annie will provide practical tips and examples of how to free your data to romp and frolic with other complementary data sources with real-world examples found out in the wild (and her hard drive).

Key Takeaways

  • Use tools that allow you to marry data from different sources.

  • Rock APIs even if you’re not a developer.
  • Build dashboards that help you keep your finger on the pulse of your organization’s performance.

Video Transcription

Today, I am going to talk about bursting data bubbles. One of the things that I've learned in my years of working with data and I work with lots and lots of data is that even good data can go bad. It can go horribly awry with very expensive, catastrophic consequences.

So just one example, in '97, a U.S. Navy warship went out to sea and its propulsion system suddenly shutdown without warning. The reason for that was its database couldn't divide by zero and no one had accounted for that. So it had to be towed back to Virginia and it was a very, very expensive mistake. A couple of years later, in '99, NASA lost a $125 million Mars orbiter. The reason for that, the Lockheed Martin engineers, that's a big engineering firm in the U.S., were using English units of measurement where their agency was using metrics. And so, $125 million loss on that one.

In 2008, in the United States, the Lehman Brothers declared bankruptcy which was quickly followed by a meltdown of stock markets worldwide. A company in the U.S. called Barclays agreed to buy up some of their assets, but the Lehman Brothers wanted to hold back 179 contracts. So what they did was they took an Excel file. Now, it seems kind of unbelievable that a company of the size of Lehman's was working with Excel files, but they held them back by hiding those rows. But a first-year associate took that Excel file, converted it to PDF, not understanding that those 179 rows of assets were now revealed causing a huge frantic panic. It was all over the press, it was a really big deal. Again, an Excel error caused that data drama. 2010, there was a volcanic eruption in Iceland and they shut down thousands of flights because they were afraid that the ash was going to get inside the jet engines, and cause catastrophic damage so they cancelled thousand of flights. However, they found out later that the model that they used to make this decision was based on bad data, and in five days of flight shutdowns, it cost the airline industry $1 billion of unnecessary shutdowns all together.

So, as marketers were generally not dealing with those kinds of catastrophic losses from data drama, however, we still deal with very serious losses. I once had a client, they were a large publisher in New York and they had gone through three or four agencies, and they... no, I think they had gone three agencies and had brought in a consultant, and said, "You know what, no one knows SEO, they don't know what they're doing, we're not getting the results that we need." So she knew of my work, she knew that I offered analytics audits. She said, "Listen, you've gone through this many agencies in three years. I don't even wanna touch your site until you get an analytics audit. I just wanna make sure that the data is good." And I went in there, it was the most dramatic analytics audit I had ever done. I actually told them, "You guys need to declare data bankruptcy." They had borked their data so badly with campaign tagging, they had put campaign...if you know anything about campaign tagging, if you don't, if you just search campaign tagging, it's my one claim to fame. I outrank Google for the term because I don't this massive guide on campaign tagging and one of the big no-nos in campaign tagging is you never ever, ever, ever put campaign tags, those are those little, like, UTM parameters, on links on your site.

Well, one of their developers probably read a blog post somewhere and got so excited about campaign tagging, he added campaign tags to every link in the navigation of their site. Their top nav, their footer, and their side nav, and in a single month overwrote one 1.6 million sessions. So I was like, "You know what, you could take spaghetti, throw it against the wall, it's gonna be as predictive as your analytics."

So I mean, that's probably the most dramatic an example I've seen of good data going bad, but typically the main cause that I've seen of even, like, little disasters which can turn into big disasters especially if it cost you your job or contract or something like that is typically manual workflows. So what we've all seen, so, you know, we come to conferences, we talk, big talk, you know, people are talking about APIs and stuff like that, but in reality, a lot of reports that I see that, you know, either agencies or consultants or even in-house people do look like a lot like this is like grab a screenshot from Google Analytics, throw it on a PowerPoint slide, boom, you're done. Or something like this, grab screenshots from wherever, throw them in PowerPoint, throw them in a Word doc, send them off to whomever.

The problem with this is that you're always repeating the same processes. It's like you are having to fish for the data and you're having to do the same thing over and over and over. So it's like you go into Google Analytics, you get your reports, you export to Excel, and you do whatever you're gonna do in Excel. Maybe someone else in your agency, you know, processes the data a little differently, they might use different formatting, but they're all very manual processes. You do research for one client, and then another client needs similar research and you start from scratch each time. Not only does that cause a lot in terms of wasted resources, it also adds a margin of error.

So I'm gonna talk a little bit today about some really cool and easy, I mean really easy ways where you can overcome this, and I'm going to speak specifically to people and teams who are not developers. If you have the advantage of working with a sophisticated developer team and they're actually data savvy because a lot of times developers are very good at what they do on websites, but they're not data savvy and they're not thinking in terms of data. But so, if you don't have the advantage of working with a data savvy, sophisticated developer team, this presentation is going to be for you because I know just enough development to be really, really dangerous, but I've learned to work with tools that give me as much power and, in many cases, more power than most people who are developers.

So, in order to do that, we're gonna have to talk about something called APIs. Now, I'm going to simplify this quite a bit to try to explain it, but basically with an API, what it does is you're taking a URL. So, your data, let's say Google Analytics or Facebook Insights or whatever marketing data you're working with, it's hosted on a website somewhere and you're given a URL to access that data. Usually like a subdomain of that website, and then, you have a number of parameters that you can work with. What's a parameter? If you ever see an equal sign inside a URL, that's a parameter. What comes right before it, that's the parameter. What comes right after it is the value. So, you know, sort equals high to low, color equals orange, you know, and there are all kinds of parameters.

And with APIs, you have all of these different parameters that you can use to filter your data, to sort those types of things, but there are tools that give you an interface where you don't have to build out. Like when I first started working with APIs, I was working with a dashboard tool called Klipfolio, and at that point, it was a huge pain because you had to learn all the little nuances of the APIs you were working with. But now, there are tools with interfaces that help you get the data that you need. And what happens is when you work with an API you have one central hub which is what you use to build out the visualizations, whether it's Excel, Google Sheets, dashboard tool, we'll talk about some of those, and you just plug in these APIs into that dashboard. And now, instead of you having to go and fish for the data every time, it's as if the fish are coming to you. They're just jumping right into your bucket.

So one other little caveat about that is one thing you'll learn if you start working with the APIs is APIs are very, very different. You have some APIs that are very generous, like the Google Analytics APIs there are issues with it, there are caveats, there are little gotchas, but it is very generous. It gives you lots and lots of data. Facebook, not so generous. So you have to kind of work around some of their limitation, how much data they'll let you get in a single call and, you know, things like that. Some APIs are totally free like the Google Analytics API. Other APIs are free to a point then you have to pay for the data, those types of things. So, you kind of after awhile, you just get used to searching tool plus API and learning what you can about it.

But one thing that I absolutely love that makes it so, so easy to work with these APIs are something called data aggregators. So I have a couple that are my absolute favorite and they are hella cheap like so, so cheap. You could be a total badass pulling in all of this API data for like Blockspring and I benefit in no way from this. There are no affiliate links in my presentation. These are just...I'm constantly tinkering and playing with new tools, and any time that I come across a tool that I'm like, "Okay, that's easy to use. That's not intimidating." I get really enthusiastic about it because I'm like, "Okay, this is a way that you can bring API data to the masses."

So Blockspring is like $15 a month, and it pulls in, it has over a hundred connectors. You'll see these are all of their different connectors. They're broken into different categories. You have everything from analytics connectors, you have email, you have, you know, Google Drive and Dropbox, and those types of things. If you've ever worked with, it's a very, very expensive tool. This API tool pulls in the free version. I mean, it's just really, really fantastic and it makes it so easy.

So one of the things I'm gonna do is I sifted through a ridiculous number of video tutorials because I was trying to find something that would kind of give you an idea of how cool this tool is in terms of being able to take data from different data sources and be able to do research with it, and the value in it is that it's all repeatable, like you can just set this data to refresh, you know, like, daily or hourly or weekly or monthly. So if you have a monthly report for your clients or your boss, it just automatically refreshes and there you go, there's the new data in Google Sheets or Excel or wherever you have it pulled into.

So what I did was I took this one video tutorial and I just, like I edited the video to shorten it, and I'm going to show this video because he uses...this is one of the guys from Blockspring, and you'll see an example, I think a really good example, of how he uses a number of different APIs to do content research. So to generate ideas for your blog. Hopefully this video works. Do we have a sound? Is it working? There we go.

Male: So first, let's find an interesting topic.

Annie: So the first thing he does is he goes to Twitter. So I'm gonna kind of talk over him a little bit.

Male: Twitter functions, and find the trending topic.

Annie: So he's looking for...he's using the Twitter API to pull in trending topics...

Male: Let's get this global...

Annie: ...and he's using Google Sheets.

Male: But I am gonna look at...

Annie: But there's a parameter in that API where you can look up a particular region, a particular town, state, country. So he pulls in Chicago.

Male: And then...

Annie: And those are the trending topics that are gonna come up here for that time in Chicago. So you can already kind of get an idea for where you could go with this.

Male: Let's pick a topic and we'll make it...

Annie: Now what he's doing is instead of hard-coding these parameters into the URL, he's dynamically referencing different cells. So he's going to point the API to say instead of the hard-coded topic, that's just there for a placeholder. Go ahead, and pull in whatever is in F4.

Male: If there's a lot of money on the line.

Annie: And here are the top 20 or so tweets about that topic. Now he's going to go into the "The New York Times" API because they offer an API, he's gonna point to that same topic, Mayweather, and pull in articles for that time period that "The New York Times" published.

Male: So search, get recent news about a topic.

Annie: Now he's using a different recent news API, different from "New York Times", and is pulling in just more topics. He pulled the top 10, but you could set in a parameter 20 or a hundred. He opens up one of the examples.

Male: Let's summarize these articles.

Annie: So now he's using a different API that summarizes articles like explain it to me like I'm a five year old, there's an API for that, Explain to me.

Male: So let's summarize down into a single sentence.

Annie: And he could set that he only wants one sentence, a one sentence summary for each of these articles just so you could, kind of, get an overview so you don't have read all of the articles.

Male: Okay, it looks like people wanted to see Mayweather lose. It looks like people talk about Periscope and Meerkat, about streaming footage of the fight, looks like the fight didn't live up to expectations. So generally...

Annie: Next, what he does he uses another API and pulls in the full article. So he pulls it into a cell, so you can't really tell, but he pulls... Oh, I'm sorry, he does sentiment analysis because that's using a sentiment analysis API to see if it was positive or negative.

Male: And we could do the same thing for, I'll set it to pull directly all the content from "The New York Times"...

Annie: Then he goes uses another API, it's a web scraping API and scrapes the whole content for each of these articles. In this you can see he's doing here by pointing to the URLs from a different API, so you can see how you can start to use these different APIs to build on each other.

Male:...all the text from these articles. And it would be fun to actually trying to figure out...

Annie: Then he decides to do text analysis. This is an API that tries to figure out the topics covered in the articles and what he does is he points to that whole column which is pulling in all of the text for all 10 of the articles and it nails down that it's primarily about boxing.

This is just using API layered upon API layered upon API, but you never have to touch a URL string other than to point it to the cell that you wanna pull. So, and you're talking a hundred plus different APIs and web services offered by these, and then, any number of combinations of functions within each one of those services. It's a really, really astounding, but then with something like this, all you'd have to do is change out the topic or set it to refresh every week or every month. And you could, if you're looking for content for your particular area, you could just set it to refresh, check in, see what the trending topics are, you will go back looking like a freaking badass even if you had to pay for it yourself, 15 bucks a month for your career, you know. And you'll just be, like, monitoring all of these things, and all of these... Now, he took, kind of, a more of a chaotic approach. He has things all over the place. I tend to be a little more OCD, so I would wanna put like a lot more structure to the layout to make it easier to replicate, but if you just updated the whatever is in cell F5, everything in there would update. Then you can just build visualizations from this. You could use it for research, etc. It's really fantastic.

Another thing that they offer is they have templates. So let's say you're starting off and that learning curve is really high, there are so many APIs and stuff, this is a more recent offering. You can download these templates, load them into Google Sheets or Excel or whatever it is that you're using and it will tell you the third-party services... Do I have a laser here? Yeah. They will tell you the third-party services it uses. So if the third party service is a paid API or you need an API key, it will actually walk you through the steps of how to get the API key. So you're never left feeling like, "I don't know what to do next." So, this is another example of a template that uses the SharedCount API and the free API. API is just a scraping API tool which is absolutely amazing, but it's very, very expensive, but it only uses the free version. And this is just a template that helps you do a content audit. But once you start getting more comfortable, then you could, you know, anything that you can reference in a cell, you can use in another API. These are the different platforms that you can publish data to.

So, here's one example of how I used this for a client. They needed a dashboard and they were having issues from... they had some bad publicity which was causing issues with traffic to their sites and conversions, but then it was also impacting their company internally, like their morale was low. So I build out a dashboard to be able to track the external stuff, but I thought, "How can I track the internal stuff? How can I use data preferably from an API to be able to measure this?" So I used the Glassdoor API which was available via Blockspring. It's a free API and they had on a number, I think it was like a total of 12 different ratings and I compared them against their competitors. So they were...the client was this far in each of the areas, but it gave them the opportunity to see how they compared against their competitors and help them learn where they needed to kind of shore things up.

Another aggregator tool is called Supermetrics. This is another really, really good API tool. So my biggest complaint with Blockspring is, I mean, it's super, super amazing, but its Google Analytics and search console if you do a lot with those tools which I do, they are not super amazing. Supermetrics is much, much better for Google Analytics and search console. If you go in knowing exactly what you want and you only have to build this for one site, Blockspring is probably going to be okay for you. But, if you don't, one of the issues I ran into is if I when I first building out like a base template for this one client, we were kind of working through, "Well what do you want included in the dashboard and stuff?" And any time I would like go in and try to change anything like change a filter or something like that, or even just change like a date range, anything in the query, I would actually have to rebuild all of the filters. So for this particular client, they had a lot of like really detailed filters. And so, it was a pain in the ass, like it just wasn't built really well to scale. So Supermetrics was much, much better for that because one, you know, you can very easily go in and change anything in the query. You can also, when you build out a query, so this is what like what my dashboards typically look like the raw data. I'll have, like, you know, different types of data in different sheets, and, you know, I color them differently. These are tire kicking campaigns and issues with their site, those type of things whereas these are like more of the data that they wanted to monitor every month. So that's what I meant by like my only criticism with that video is, you know, he kind of puts it all over the place.

But, if you build out an API query and you have very specific filters, you can then, in Supermetrics, just take that cell, copy-and-paste it into another cell, and just change out whatever you need to change out. So it's a very, very easy. You can't do that in Blockspring. It's a little more expensive and it doesn't have, let me back up, it's a little more expensive and it doesn't have nearly as many connectors, but like I said, in my opinion, if you're doing a lot with Google Analytics, search console, and some of those tools where you maybe have to build a dashboard for multiple sites or something, I personally think it's worthwhile.

So then what I did with this particular client, I then took all of that data that you saw in that Google Sheet and pulled it into Tableau, that's my personal favorite as far as dashboard tools. And you can see over here, I was able to use different metrics and dimensions as filters for the dashboard. So the entire, in this case, this particular chart would update if I chose sessions or page use or a particular goal conversion, but then I also had filters up here. These are my two favorite filters to include in any dashboard channel. So being able to filter the entire dashboard by organic, social, paid search, whatever it is that you're doing, the entire dashboard updates and also device category. So, you know, is it desktop, mobile, tablet, that sort of things. So this just kind of gives you an idea. It's one thing to look the raw data, but this is kinda more what you can then do with it because Tableau you can just connect Google Sheets to it and have it update automatically.

There's another example of a dashboard that I built where the data was coming from Stat, it was their ranking data and I was able to take their keywords, break them up into different keyword groups, then you could filter the entire chart by the keyword group or individual keywords. You could, you know, put a search term in there and only select keywords that included search term, you could filter by search volume, by device, by rank, you know. So there were all these different ways to kind of zero in on these and then I couldn't show an example of being able to hover this because it would show the client's URLs, but if you hovered over any one of these points, you could decide whatever you wanna put in there to show like the URL, the actual rank, etc., etc. So there's a ton of information that you could, kind of, shoehorn into there. Supermetrics, here are the connectors that it automatically pulls in. Like I said, a little more expensive, but much's a better tool for power users in my opinion.

Next up, we also have different dashboard tools. So I'm gonna just cover a few them. There are a lot of dashboard tools out there. I'm gonna cover my favorites in terms of, in order of simplicity to use. Right off the bat, if anyone has ever worked with Cyfe, it is an absolutely amazing dashboard tool, very, very inexpensive. I think it's like $19 a month, $19, $20 a month. That's chump change. And it also has a large number of connectors. I couldn't find on their site where they actually listed how many connectors they had, but it's a lot. And if you are intimidated by dashboards, if dashboards just scare the bejesus out of you, this is your dashboard tool to use because it is like stupid easy. It's like okay just login to Facebook. All right, what reports would you like to pull in? And you have like all these sexy reports and you can choose, you know, dark backgrounds which is what I like because it's a little more lusty or a white background. You could put your logo on it like if you are a consultant or an agency and you're building out dashboards, you can put your...white label it and put your dashboard...I mean, I'm sorry, your logo on there, and tons and tons of options.

You could divide this app into different dashboard. So typically, if I am working with a client like a smaller client and they really just want an easy dashboard tool that I can teach them how to use and add data to, I'm like, "Okay, Cyfe's your tool and we'll pull out, you know, one dashboard just for organic, one dashboard for social." I mean, basically, wherever they're putting their resource, but then you could pull in like MailChimp, or, you know, QuickBooks, etc., etc. And it just gives you one place where you're monitoring the data, you can keep your thumb on the pulse and see what is going on so you don't get those big surprises because that's what causes those catastrophes when you're just going along. And a lot of times, in the marketing world, you don't even realize that a catastrophe took place, because I'll get this panicked emails or calls from clients saying, "Our organic traffic plummeted two months ago. We think we've been penalized, what do we do?" Well, two months has now passed and now you have to dig out of that hole. So that's why I'm a big, big fan. I mean, here are the different tools that it automatically pulls in.

My one criticism with Cyfe, this is not for power Google Analytics users because I interacted with the CEO. He's a great guy. I just let him know, "You really need to surface filters." They surface segments, but not filters. And segments, if you know anything about sampling, segments will trigger sampling if you have over 500,000 sessions, so that can be problematic. So for power users, you will bump up against some limitations, but they do that to make it simpler to use.

All right, next up, Data Studio. I was already a Tableau user before Data Studio came out and I kind of was a little bit upbraided by them being like, "It's Google products only." I'm like, "It's 2015. There are other products besides Google out there guys." But they up their ante in February of this year when they made it free. So it's really hard to beat free. So that's the Google way. So I anticipate I'll be doing a lot more with Data Studio because that changes everything.

Tableau, like I said, that's my personal favorite. The learning curve for Tableau is pretty high and there are some things that it just doesn't do out of the box that I think it should like the, you know, having top level widgets to be able to just, you know, monitor like sessions or, you know, new Twitter followers, those types of things. So there are some drawbacks, but overall it's very, very sexy and the visualizations that you can build are pretty amazing. Like we built this for a client just to be able to show them how their top ranking keywords were comparing against their competitors. Who had the top ranking and who had the lowest ranking for each of those keywords? And here is just another example of just like the visualizations that you can build out even maps being able, if you have any kind of longitude, latitude data, you can bring that into a map and it's pretty mind blowing. And that's all I have. So, thank you.

you might also enjoy

Share your email address and we’ll keep you updated on all upcoming marketing related events and news so you never miss a beat...