Your agency has designated you as the analytics gal/dude to join the core team embarking on a brand overhaul initiative for a client with national reach. It’s also a new client, so the stakes are high – if your agency hits the mark early in brand development, the potential for agency-of-record-dom looms large.
The initial client immersion session is a success. Your team interviews stakeholders from every department, from sales teams to hipster product developers to high-level c-execs. You riff on native advertising with their social media managers; talk automation with their email marketers; you even get to geek out with the SEO & analytics lead, who generously gives you the keys to their Adobe Analytics suite.
Days later, your team regroups on the initial discovery meeting. The creative director has the whiteboards out. Account executives have culled notes and drawn timelines to deliver on a killer brand positioning – with ad-ready design spec and maybe even video. So where do the analytics come in?
If you’re an (or the) analytics gal or dude at your agency, something like this has probably happened to you. If it hasn’t, it will, because the age of data-driven branding is well upon us. Agencies and brands alike are rightly showing up to brand discovery – a stage previously dominated by creatives – with a business and/or web analytics specialist among their corps.
But let’s be clear about one thing. Branding is an odd stage for the analytics role. The valuable insights you glean from combing through large data sets from Google AdWords/Analytics, Webmaster Tools and Adobe SiteCatalyst can inform the decisions UX leads make in design; they can facilitate identifying conversion points in email sequences; and are arguably THE basis of your content and channel strategy. But how can this data be applied to re-branding? And when all you have is web and marketing analytics data, is that good enough?
The true answer to that last question is, not always. The value of web analytics is subject to the particularities of sample size (are your data sets statistically significant?), and platform setup: are page loads, custom conversions and events firing reliably? Was data skewed by seasonalities or campaigns, and were those campaigns consistently tagged?
This also isn’t your time to be profound. Even when the planets align favorably in all the above regards, the best you can really prove is often just correlation, not causation.
Worse yet, you’re the only one who knows this.
The remainder of this post is about how to proceed from here. I’ll draw on the approach our analytics team took in our most recent client case, which looked exactly like this. And while the approach we took isn’t perfectly relatable to every data-driven brand research scenario, certain aspects are. Without further ado, here are:
10 Ingredients of Data-Driven Brand Development
Ingredient 1: Expectation Setting
Sometimes the greatest asset you have are the expectations that surround you. People seldom realize how much control they actually have over those expectations. The digital analyst has enormous leverage in this area, so use it responsibly. Brief your team in plain language on the limitations of regression analysis using someone else’s analytics paradigm (which itself had limitations). In this case we found it particularly useful to remind our branding team that our contribution would provide correlative findings at best. There will be plenty of opportunity to shine in the post-branding, strategy phases of this client engagement, especially if you can help set up new test-&-measure scenarios. For now, under-promise.
Ingredient 2: Asking the Right Questions
This is harder than you might think. When it comes to devising a marketing strategy, it’s easy to look at your findings in analytics and say: “The data on user behaviors tell us X, therefore you should do Y.” But in branding, you’re delivering a statement that looks more like: “The data on user behaviors tell us X, therefore you should be Y.”
Sound weird? That’s because it is. And it’s why focusing on the right question is critical to maintaining your bearings as you go plunging through the data. In the case of this scenario, we realized the question we needed to be asking was: What needs does the current brand not fulfill, and what can website user behaviors tell us about this? This helped steer us to the next logical ingredient, which was:
Ingredient 3: Well-Defined Audience Segments
This is arguably where the real work begins. Brand development is essentially an exercise in understanding audience, so defining audience segments in your analytics is key. For this particular client, national presence was a given, so we had the windfall of reliable sample sizes spanning the country’s top 10 urban hubs, and a lot in between.
Early into our analysis, we decided to focus on the brand’s top five geo-segments, three of which were major urban centers, and two only mid-sized metros. And knowing that the brand was hugely popular in its own hometown, a mid-sized metro hub, gave us a “control geo-segment” in brand awareness. That made six geo-segments.
Ingredient 4: Third-party Socioeconomic and Demographic Data
In most cases where geo-segmentation comes into play, it’s worth venturing outside of the web analytics realm. This is where subscriptions to premium vendors like Forrester and Iconoculture becomes golden, but don’t underestimate what you can get for free from City Data and Zillow. As it turned out, economic indicators and broad industry trends had direct relevance to our client's audiences. So it was critical for us to have a clear sense of the socio-economic environments in which our six audience segments existed.
Ingredient 5: Branded & Non-branded Keywords
It should go without saying that meticulous vetting of branded and non-branded keywords from organic and paid search visits are a central ingredient to a data-driven brand audit. It’s like the broth in the soup. There are challenges to this data collection, not least of which was the cloak of “not provided” that gradually eclipsed everyone’s data over the past year. But there are workarounds: measuring what’s available in organic keyword analytics as a percentage; combining what’s left of matched search query reports; cross referencing Webmaster Tools reports. It helped that this client gave us access to their AdWords account, which had state-tiered campaigns.
Ingredient 6: Corroborating Brand Signals
With all the workarounds you need to get consistent and accurate keyword data these days, it’s good to toss in additional indicators of brand awareness, be they social referrals (it’s tough to get adequate sample sizes from these), direct visits, or even search instances as a whole. These metrics help you corroborate what seems obvious in keyword analysis, and also keep an eye out for outlier influences (e.g., a major PR event).
What you begin to see over the course of obtaining these data (Ingredients 5 & 6) are seasonal trends that make sense in the context of your client. For e-commerce clients, that might be holiday sales; for real estate, it’s the spring home search. In the case of our client, we divided web traffic data into quarterly sets, which allowed us to draw logical assumptions about audience behaviors – and needs – at different times of the year. By comparing these with geographic and economic factors, we began to see which behaviors (as signified by length of visit, or length of time before return visit) were indicative of audience needs.
Ingredient 7: Identifiable Conversion Points
Until you have conversions, the narratives you’ve begun to see in your data are nothing more than assumptions. Conversions are what ground your analysis to consequences. You can reliably assume your audience likes sci-fi, because who else would buy full storm-trooper body armor online? So pick a conversion that not only tells you something conclusive about your audience, but that enough of them are doing. And it may be as boring as filling out a contact form.
Ingredient 8: Corroborating Conversion Actions
With your baseline conversion point in place, it’s great to draw on the bits of audience evidence that you can gain from all those other conversion actions: checking a preference box, printing out a blog post. Not everyone in your data set is doing it, but if enough of them are, then what does that tell you about college students in Boulder in early Q2? (Answer: I have no idea.)
Ingredient 9: Role of Mobile
This is a bit of a red herring, as mobile analytics vary from one brand to the next, be they sourced from responsive websites, mobile subdomains, or mobile apps. And rarely is mobile data tracked perfectly across all of a brand’s applications. In the case of our client, we were fortunate that their analytics dude had created a custom event in Sitecatalyst on exit links that corresponded to the instance in which a user downloaded the mobile app from the mobile website. This itself indicated a critical second-tier conversion: If you’re going to the trouble of downloading a new app to reside on your phone (taking up your data bandwidth), you’re exhibiting a specific level of commitment to that brand. Using this we drew a few key correlations between peaks in brand awareness and frequency of app downloads. With the correlation between brand awareness and acts of confidence as a backdrop, we then explored data points that could tell us something about audience need.
Ingredient 10: The Moment Where You Go “Wha?”
It’s not the “Eureka” but the “that’s funny” moments that tell you you’re onto something (to butcher an Isaac Asimov quote). These are also the most rewarding, and it’s easier to see them when the correlations you’ve established paint a picture of “the norm.” That’s when one irregularity can tell you not only that an audience need went unfilled, but when an audience simply had no idea. It’s in those spaces the brand should be better positioned.
At the end of our brand analytics audit, we came away with several helpful takeaways. Granted, some were useful in so far as they confirmed notions that people already assumed. But others were more groundbreaking, such as the idea that the client’s brand awareness struggled within certain niche industries, signaling perhaps missed opportunities. If nothing else, the act of conducting an analytics brand audit gives your team an embedded grasp of audience that can extend throughout the life of the project. But only provided the analytics gal or dude keeps everyone on task.