English Español Deutsch Français Italiano Português (Brasil) Русский 中文 日本語

Storytelling with Data




Shelly Fagin: Hi, my name is Shelly Fagin. I've worked in digital marketing for about 17 years now. I do a little bit of work with SEMrush, I helped to work as the community manager for their group, and now I'm doing corporate speaking as an ambassador.

I get so excited about meeting some new people, and I got a chance to meet Jessie and Jonathan from Conductor. Jessie, you started your career in PR, right?

Jessie Cohen: I did.

Shelly Fagin: You found your niche obviously in data, in digital.  You work at Conductor, and if you guys aren't aware of what Conductor is, it's an enterprise search marketing platform. It's a technology, let's see, that is driven by a deep passion for helping companies help their customers, which I'm all about.

Jonathan is also with Conductor. You're the Manager of Customer Insights. You studied cognitive science in college, which is another background that I imagine would be so useful in marketing. You believe that search data is a window into human behavior for sure.

Since you joined Conductor in 2017, and you've been analyzing keywords to tell stories about how search data can inform overall SEO and digital strategy, which is very fascinating. I love that you have that background. How have you found that it's helped you in this industry and give you a leg up over others?

Jonathan Bloom: I mean, it's a tremendously helpful background to have. Cognitive science is a really exciting area that encompasses a lot of different disciplines. There are a lot of things that actually go into how people behave and how people act, and a lot of that is mirrored in the way people search.

Shelly Fagin: Most of you on here, you guys probably know Bill. Bill Hunt is such an awesome guy. I've been following him online for many, many years. He is the CEO, and make sure I pronounce this correctly, Black Azimuth.

Being able to go in and really analyze properly so we can take away insights and create a story, can be one of the most valuable skills there is. It's so exciting. Actually, I cannot wait to hear the presentation that Jessie and Jonathan have for us. There's so much great information. I'm going to let you guys take it away.

Jessie Cohen: Awesome, we're excited to get started. Our presentation today is Storytelling With Data, and we'll go into data collection analysis in communication with Conductor. Our agenda for today is why storytelling with data, and then going into the collection analysis and storytelling components.

Telling Stories with Search Data

Starting with why storytelling with data. I started at Conductor about five years ago, and I was the first data analyst responsible for keyword research on behalf of our customers, informing them of any insights I found, and what they should be tracking. I found that I would be pouring over these excel sheets of data, and I'd get so excited whenever I found something fascinating. But, whenever I would try and share it with a coworker or a customer, peoples eyes tend to glaze over when they see Excel sheets like these.

The only way I could get it to really resonate is by putting it into a data visualization that really got the point across. Why I say storytelling with data rather than data visualization is because I do think it's a full story. There's the exposition, so data that sets the scene, rising action, which is doing analysis on that data, a climax which is when you find an insight, the following action, which is visual, and then the resolution, which is actionable items.

The worst thing possible is when I would show someone data and an insight that I found, and they say “so what”. Every story should end with something that they can take away.


I might be a little biased but I often find that organic search is crucial but it's often overlooked and underfunded. If we can tell compelling stories with the data that we find, it's way easier to demonstrate that value. Then, we get the opportunity to get more bandwidth, buy in or budget, to make a bigger impact in our day to day jobs.

How to Collect the Right Search Data

Jonathan Bloom: I'm going to talk through the data collection methodology that we use here at Conductor. Data collection isn't the most exciting part of this journey to insights, but it's definitely super important because the data that we collect, the keyword sets, are really going to be the foundation of all of the insights and all of the storytelling that we are going to show.

So really, the first step starts before you select a single keyword. If you can spend some time taking a look and understanding what is the keyword universe around the topic that you're researching, this can save a lot of time down the road.

In this case, we did research on the travel industry, specifically into leisure and business travel. Just as an example of something that we found, we found that business travelers were really split into two distinct intent groups. Some people were searching for travel for business, so they had to travel on a business trip, whereas others were searching for basically traveling as their job, people who wanted to travel as their entire career.

Those are two different distinct intent paths that we wouldn't have known had we not looked into some of the keywords prior to even diving into research. This is a pretty important step.

Then once you have an idea of some of those keywords that you're going to use, you have to generate sets of thousands and thousands of relevant related keywords.

We use a couple of methods for this. One is to basically use several seed terms, so terms that we know are relevant to the topic we're interested in, and then we can return words and phrases that are either related because they contain one of the words in the seed, or they're semantically related. They don't necessarily contain the word we put in, but they're clearly considered relevant by topic.

We use a number of tools for this. We use explorer, Conductor's explorer, we use SEMrush, but there are a number of free tools as well, including Google Search Console, as well as customers’ paid campaigns. Another way that we also generate this list of keywords is by looking at domains and URLs that we know are relevant to the topic, and seeing what they already rank for.

Cleaning Your Data to Make it Useful

The next step is about taking those sets and really making sure they're super clean. This step is probably the most time-intensive and can be a little tedious. But again, this is so important because a few high-volume terms that are irrelevant to the data set can really skew the data and it means you have to go back and re-analyze things, and re-clean the step, which takes even longer ultimately.

Also, if this is data that you're going to end up sharing with the customer or a client, if they see a few bad keywords here and there, it really undermines their confidence in the entire set. You want to make sure the datasets are super clean.

In this case, we cleaned it in a number of different ways. Obviously, we cleaned out misspellings, we cleaned out clearly navigational searches, so things that said like www or .com in it. We also chose to exclude all branded searches.

For example, Expedia flights to Florida, we would exclude that search because that person is clearly looking to navigate onto Expedia, and we want a more natural market share of which sites are coming up when people don't have a brand in mind already.

Bucketing Keyword Data into Relevant Categories

The next step, once we have this final set of keywords...we bucket that data into thematically relevant categories. This helps us because we can obviously look at the set as a whole, but this also helps us to dive deep into specific topics or modifiers that people are using, and tell even better stories and understand how people are talking.

In this case, as I said, we bucketed all the keywords, high-level, into either business or leisure travel. But within each of those, we did a number of different ways. We bucketed them according to intent, we did it by persona, we also analyzed the different modifiers and types as we'll talk about, that people are going to take.


Sometimes when you find that you bucket the keywords into categories, there aren't actually enough in certain categories, or there's one category that has way more than others. In that case, the process continues. You have to go back and generate more keyword sets, clean them, and then put them into the categories.

It's an iterative process and it definitely can be time-consuming but I think the main overall point here is that, having these clean datasets are going to be essential to really understanding the data and the insights that you show down the line.

Getting Insights from Your Data

Jessie Cohen: Then, it's time for data analysis. We really want to take a deep dive into what we analyze to find an insight. There are numerous factors that we can analyze. We find that the most common and easiest to find a quick insight for that quick win is with these eight major themes.

I start off with historical data because historical data provides context. If I have that benchmark of knowing how your keywords have ranked, or what their monthly search volume was in the past, then I can spot anomalies for the future. Once I have an anomaly, this is something that I can easily make an insight out of.


I also like to look at monthly search volume, and this is a great indicator of how popular a specific search query is, how often it's searched on a monthly basis. Also, if you're using customer voice, at Conductor, we use the word customer voice a lot, and that's to determine if a customer's jargon is aligned with the same way people are searching. Oftentimes we find misalignment, and we can use that monthly search volume to say this is how people are actually searching, and this is how you should speak to it on your website.

Competitive data is also extremely important because we can understand if our competitors are using the same customer voice. With this, we can find where there are opportunities for us to gain in that organic market share, or where areas are overly saturated, and we might want to scale back and put our efforts into different content ideas.

We also want to look at branded versus non-branded. Typically, we find that people don't want to track their branded keywords because the assumption is I'm going to rank in first place for my brand. But, and this might be from my PR background, I think it's great to have an understanding of that entire search. Who else is talking about your brand? Are people saying good things about your brand or bad things? This helps you monitor and control that conversation.

Search intent is also extremely important. When I say search intent, I'm really talking about the buyer’s journey. I look at this as early, middle and late stage. Early tends to be informational questions, who, what, when, where, why. The middle stage is, especially when we're talking about products and services when someone might be comparing one product or service to another. The late stage is transactional in nature so, when we think people are ready to purchase that online.



PPC terms are also extremely important because if a customer is willing to bid and spend money to show up for that query, it should be important to them from an organic standpoint as well. If we can align PPC campaigns with our organic search strategy, we can see where we're completely dominating a query in terms of share from a paid and organic standpoint. We might be able to scale back paid spend if we're consistently ranking in a very high position from an organic.


Localization is growing in importance because we know that, more and more people are searching on mobile, and performance, popularity or demand may depend on a region. Just because you're doing really well in New York, you may not be performing as well at your brick and mortar or organic standpoint in Dallas. We want to make sure that we have a really accurate depiction of this data from a local standpoint so we can make the best-informed decisions from a business standpoint and from a website standpoint.


Lastly, seasonality. We know that rankings may fluctuate, but also search volume fluctuates. If we have an understanding of how seasonal particular search queries are, we can make more informed decisions in regards to our editorial calendar. For this specific webinar, we wanted to go over that travel industry analysis that Jonathan did, and we focused on competitors search volume, PPC, localization, and seasonality.


Jonathan Bloom: I'm excited to walk through some of the insights that we found based on this travel keyword set.

First up, we have market share. Market share is an incredibly powerful tool to understand where the competition lies beyond just the traditional competition that you think about as your business' traditional competitors. Market share means the percentage of available search volume on page one that you're actually owning.

In this case, on the right-hand side, we looked at the competitor set for business travel searchers. There are a number of interesting things to call out here, just on the basis of this graph alone. First of all, Expedia, as you would expect, is doing pretty well for this search, but you also have Indeed and Glassdoor showing up with 4% market share. These, I think of them as career search sites. To me, this indicates that they are creating some form of content around business travel to get that relevant demographic onto their site in an earlier stage than just looking for a job.

You have those competitors, you also have YouTube showing up, which again is an indicator of the increasing importance of video searches, and really showing up on Google's other enormous search engine.

Then moving down on the right-hand side, you have very specific industry influencers, so people like Expert Vagabond, the Broke Backpacker, and Location Indie. These are people that again, you don't think of them in the same league as Expedia, but they are taking up valuable real estate on these SERPs.

It's important if you are a business travel provider asking are you familiar with these influencers, do we know what kind of content they're writing about, and have you considered partnering with them as a way to get more eyeballs on our brand?

Then another way you can look at this is really by breaking it up by hyper-specific competitors. If we look at just keywords that contain the word sabbatical, you have this site, Fairy GodBoss owning the sabbatical market, and it's really owning it with this one page. There's an article on what is a sabbatical, and why should you take one.

This page alone ranks for about 60 keywords, and based on its ranking positions and click through the curve, it generates an estimated 7400 organic visits per month. Again, this is with one piece of content. If writing about sabbaticals is an interesting topic for you, this might be an interesting page to look at and say, how can we emulate this, how can we provide more value than this page, or if you can't beat it, how do we ultimately get our content or our brand onto this page whether it's with a partnership, or with a display ad on this already high performing page?

As Jessie said, a story is only as good as the actions you can take from it. The action is really about finding these non-direct, non-traditional competitors, and you can use that to inform a whole lot of strategy, whether it's partnering with the site, whether it's finding new communities to engage with, or whether it's using other forms of marketing like display ads on high performing pages. There's a whole lot you can do with competition.


Next up we have search volume or word frequencies. In this case, I looked at leisure travel searchers, and we wanted to see the different types of trips that people might be taking. We found that by far, the most commonly searched for type of trip...is a spring break trip. It's about three times higher than the second place type of trip, which is a weekend trip.

This is a really interesting way to...figure out what are the categories that people are most interested in. There are a number of different ways you can visualize this. On the left-hand side, I have a bubble tree, a bubble map and on the right-hand side, you have a word cloud. Basically, it's just about showing the relative demand for different topics.

In terms of action items, again, there is a whole lot you can do with these. I mean, the first question I would ask is, do you have landing pages that are actually targeting these types of high-valued trips? You might have thought to put a spring break landing page, but do you have the destination wedding landing page? If not, that would be a very relatively quick way to try to capture some of this high intent traffic.

Then the longer term, you might, especially for some of the higher volume topics, you might want to create and optimize really valuable long-form content. This we find that this type of long-form content is what Google really values for pulling info onto the answer box, which is now above position one. We think that this creating long-form content could be a really great way to capture them.

Lastly, you can really analyze the competitors that are doing it right. It's combining this with the market share that we see in the previous slide. What is the market share for each of these areas? Are there hyper-specific spring break competitors for example? And how are they performing? How are they doing it right? How can we learn from them?

Next up we have paid. This can be a really, really valuable way for us to basically quantify traffic because in organic, we sometimes have trouble doing this, but paid is a great corollary, that gives you an average bidding cost for a keyword. In this case, we look at the amount of the cost to buy a certain percentage of the traffic for our entire travel set based on the average CPC that we pulled from SEMrush.

The real takeaway here is that even just getting a tiny fraction of the clicks from the set about 1%, you're looking at an annual spend of over $700,000. Of course, the second you actually turn off that spend and decide I don't want to invest in this channel anymore, all of those clicks are actually going to go away because you've been paying for them all along, versus investing it in organic experience, or an organic piece of content, that's something that can really continue to deliver year after year.

This isn't to say that paid is bad, or that we shouldn't be investing in paid. But it's just to show that it's an expensive acquisition channel, and it's a way to quantify the traffic, the dollar value of our keyword sets. You can actually take it even further when you pull year over year data. You can do this in SEMrush as well. We actually found that, across our entire travel set of keywords, over half of them increased in paid costs since this time last year. They also increased by an average of $2 per click, which is significant when you consider the volume associated with this set.

This tells us that by continuing to rely on paid as a primary source of the traffic to our site, that cost is going to continue to go up. As far as action items, it's kind of what Jessie already mentioned, which is informing your budgets for both SEO and SEM.

We're not saying stop spending money on paid, but rather, maybe consider pulling back on spend for keywords that you're already in position one for, or the opposite, increasing spend on keywords that really you have no chance of ranking, of displacing the top performers for.

I want to talk through some local insights as well. I think this stuff is all really interesting. I apologize it's a little small, but what this chart shows is, it looks at the business travel searchers for our keyword set, and it breaks out that search volume by the top 15 or so cities in the US.

What you find here is that search to search demand is by far the highest in New York and LA, Chicago, Houston, et cetera. If you show this to someone, they'll probably say, "Great, this isn't surprising to me. As a business travel provider, I know that New York has the highest search volume for business demand. These are just the biggest markets." That actually is validated when we add this yellow line which shows the population of each of these locations. It really does scale with population, the demand.

Something that we've been doing to really bring local demand to life, and surface cities that you might not otherwise be thinking about, is we take a ratio. We look at the demand in that city divided by the population of that city. What that shows us is, the cities where they're relatively high demand for business searchers. There's high demand as a proportion of their population. That's what this chart shows. This chart shows the relative demand for business travel in different cities across the US.

Southeastern cities have a relatively high demand for business travel, so, Atlanta, New Orleans, and Miami compared to some of those bigger places. Again, this isn't to say, "Oh, I need to stop investing in New York because that's not where the relatively high demand is”.

But this does inform us about potential new markets that you might not be thinking about a business travel provider. So again, as takeaways here, you're going to ask, "Do I have landing pages that are going to be capturing this traffic from these high intent or these relatively high demand places?" This is the foundational approach to local SEO.

Finally, we wanted to talk through the seasonality of data. Obviously, when you look at a keyword, you're looking at it at a point in time, but really, what that search find is showing you is that it's a rolling average over the last 12 months.

When you actually break it up by month, you can tell some interesting stories. Probably not surprisingly, you find that ski demand really spikes in the December, January timeframe, during the holidays, people might be on break. Beach demand spikes around what you would consider as spring break time, so March, April, May.

You can even pair this with local data for example, and maybe find that ski demand fluctuates quite a bit on the interior of the United States, whereas beach demand stays relatively flat across the year in coastal cities. You can really add layers of other data on here that can make this super valuable and interesting.

As far as so what, I'd say the main thing that this can help inform is your content calendar. Obviously, we know from search data what we should be writing about. But what we don't necessarily know is when we should be writing it, and when we should be promoting it. This is super important.

We would really recommend reading the book on the left, it's called Storytelling With Data. A lot of what we do here at Conductor is informed by this. It's a great resource for understanding at a very practical level, how do I make my charts and figures as compelling as possible.

Will AI Change the Way People Search?

Shelly Fagin: Thank you, Jessie and Jonathan. You guys, that was awesome. We've already got a lot of questions actually.  "Do you think that AI would change the way we search, or do you see the algorithms changing drastically with this technology?"

Jessie Cohen: That is a good one. I do think that it will change the way we search. I think the more that we use voice search. We're going to get lazier and lazier in how we pose our search queries. I think picking up on how the way that we phrase our queries change over time, is going to be great in matching the way that our content changes on our website over time so that we're still serving up the best results that answer the way that people newly phrase their questions.

What Clean Data Means

Shelly Fagin: That's so true. We actually got another question asked, what do you consider super clean data?

Jonathan Bloom: I'll take that one. When I say super clean data, I mean only words and phrases that are highly relevant to the topic at hand. When we see data, when I think of not clean data, I'm thinking of duplicates, keywords, misspellings, keywords that might have similar words but not have the same intent.

I'll give you an example. I was doing keyword research for a company that sells boots. They're based in the UK. There is a ton of searches around the keywords boots near me, which I assumed to mean someone who's looking for a pair of boots right now. But, that's not actually the case. Boots is actually the largest pharmacy in the UK.

When someone searches that, it's actually entirely owned by Boots.com, which is in the UK a pharmacy. That I would consider to be an irrelevant search that really skewed the data. Sometimes you really need to think about each and every search and understand what is the intent behind it, and intent can be a really exclusionary factor as well.

Jessie Cohen: When we talk about cleaning datasets, we do at this point in time, still need a human to understand this had ambiguous intent. It could mean this, which would skew our market share, so let's omit it from the set. I'm a huge stickler for cleaning data.

Shelly Fagin: That was a really, really great example honestly because everyone assumes that eventually, tools will be able to do it all. Some people in this industry really lean on tools a little too much.

Tools for Sharing Data with Stakeholders

Someone has asked, what about presentation tools for sharing data with stakeholders? PowerPoint is a go-to tool, but is this the best way to present it? Any other tools that you can share?

Jessie Cohen: Yeah. Conductor Searchlight has a number of workspaces. I don't mean to plug our product over and over again. You can export it into a PDF, and then you can do what you like with the PDF. We find that more and more of our customers are using Google Slides. So Google Slides is equally as valuable to PowerPoint.

But I also love visualizations in Tableau, which also tend to export into PowerPoint and slide functionality. Those seem to be the two major ways to share the visualizations, but I'd love to hear from anyone else if they have other tools that they like to use.

Shelly Fagin: Yeah. I think someone mentioned in the chart about Power BI too.

Jessie Cohen: Power BI's awesome.

Jonathan Bloom: Yeah.

Bill Hunt: Same with Google Data Studio, anything I think beyond just giving them static data, I find that if you show them very easily how they can interact with the pull-down and watch things change on the screen, so similar to what you were talking about Jonathan.

Shelly Fagin: Another thing that you guys touched on that I thought was so true is understanding who is ranking for top phrases. I love that you create a top players analysis. It can add a lot of value for showing who's ranking for that set of terms and their market share.

Many businesses have a misconception about who their real competitors are. Especially like brick and mortars, they think it's the guy across the street or these people that they've been battling up against for a really long time. They don't realize that those aren't their search competitors online. They really need to be researching this.

We got a question that, other than the PPC cost associated with specific keywords, do you have any recommendations as to other financial metrics that would be helpful to persuade a management team that it's worth increasing investment in SEO relative to other channels?

Bill Hunt: Well, I mean, if you take like the model that was shown by Jessie around that, we call that the cost of not ranking, where the only way you can get traffic is to pay.

One of the other things that we can look at is, if a lot of times people use lead gen sites like a TechTarget, if you're in a B2B business, you'll go to a lead gen site. A lot of times they take your PDFs and white papers and that helps them rank. We've created some models to show that if we can beat them, then we don't have to pay. In some cases, it's as much as $15 for every lead they send us.

If we can capture those leads before they do, or with them, it saves us money. I think that's a big one there. But enough people I don't think do the basic one of, the only way to get this traffic is to pay for it.

I think as Jonathan said, think out of the box. Are there multiple ways that this data can connect to people? What's their cost of acquisition?

Using Data to Tell Stories to Different Audiences

Shelly Fagin: Awesome insights. It seems like there are multiple audiences for storytelling. There are executives that demonstrate value, creatives to find and locate those content ideas. What ideas, other audiences are there for the data stories that we talked about?

Jonathan Bloom: There are a number of different people that care about this information. Obviously, as the question alluded to, getting executive buy-in is a really important purpose. Even within the exec team, there are different things that might resonate. A CMO would really care about seeing logos, seeing who am I beating, what percentage of their market am I owning? Whereas, a CFO might care more about projections. If we were to own this percentage of the market based on our average order value, and conversion rate, how much revenue would we generate?

Then I think obviously, another area would be just SEOs and search marketers generally speaking, who might be more of the data nerd type to care about the relative search volume for different categories, and how it breaks out locally

Jessie Cohen: Yeah. We try and make sure that we're setting up proper stakeholder reporting for everyone who can leverage such data. So just to reiterate on what Jonathan was saying is, we need the highest level, most impactful information for the executives, to just understand the healthiness of their content and their organic performance on a regular cadence.

Then we love to drill down to provide what's most insightful to that copywriter, to that content manager, the email marketer, the merchandiser, the PR, as many people we can get involved, to leverage search data as possible.

I can't believe we haven't touched upon the product manager part. When it comes to storytelling with data, we actually have the most amount of impactful stories when we talk about product naming conventions. Because oftentimes, those naming the products don't consider SEO and the thought process and ideation for what that product should be named.

Shelly Fagin: You guys are incredible. I've had so much great conversation. I want to thank you guys for being on here with me today and joining us at SEMrush.

Jonathan Bloom: Thanks for having us.

Jessie Cohen: Thank you.

Shelly Fagin: Thank you.

Bill Hunt: Thank you.

All levels