10 Questions with Atlanta Analytics

About a week ago, a really nice guy reached out to see if he could ask a few questions about analytics and how it could (or should) play a role in a business that’s just getting started out (although I felt like the questions were relevant regardless of business age). I thought his questions were really great and were thought-provoking, requiring this guy (me), who can typically be a little too general, to face the music a little more directly. Here’s what we talked about…

Update: It was pointed out to me that these questions are uncannily similar to one of Avinash’s most recent posts. Not sure how I didn’t remember the questions/answers (I blame sleep depravation), but I’m sure some of our answers will be very similar, since as you all know, I want to be like Mike.

Q: If you were to put together an effective web analytics plan to measure my client’s website, what tools would you use and why?

Google Analytics, 4Q survey, maybe crazyegg to appease some people. If we were going to be making some changes, feedback army and usertesting.com can often be great. It goes without saying this is without knowing the types of decisions people will be making, but this is a good standard suite of tools I could live pretty happily with.

Q: If you had one wish that would force all analysts to do one thing perfectly all the time when analyzing, what would it be and why?

Stop thinking about the web site and start thinking about the business. We tend to just keep our heads down and tie what we are seeing to the web site, rather than realizing that the implications are probably much, much bigger. When we think about the business in our daily/weekly/monthly/etc. analysis, we can be a lot more creative with our solutions and not treat things like the product, branding, offline efforts, etc. as “fixed.”

At their core, most companies are configured for failure online. The typical compensation plan where product development is compensated on upside and IT/development is compensated for cost-reduction reveals a planned point of failure. As web analytics people, not only can we find ways to improve outcomes, but we can also find ways to remove key roadblocks in the business, from an unbiased, external perspective. When we start doing this, we will own much more expensive shoes and cars.

Q: How do you sell the value of web analytics to a skeptical client in 30 seconds or less?

If I were trapped in an elevator and had to have a conversation with a person this insane, I would probably just tell them that I’m not going to waste 30 seconds of my life explaining why driving down the interstate with a blindfold on doesn’t make sense. There are too many smart businesses out there for me to spend my time convincing someone that the earth isn’t flat and that analytics matters. And believe me, if they are skeptical about web analytics, just wait until you find out how this type of person will get in the way of actually doing anything with that information once you have it. Hmm…guess I just spent 30 seconds explaining what I said I wouldn’t…

I believe, in the big picture, that we’re going to be a lot more successful in the future if we sideline people with brain-lock, rather than trying to convince them of something they subconsciously aren’t on board with. It’ll take years, but it’s time to dump some chlorine into the corporate pool.

Q: What is the best way to attribute an offline sale to an online assist? This is very important, as our business provides its services both online and offline.

There really is no best method. You have a number of ways to do this, and each one has huge holes somewhere in it. But I would use a mix of coupons, vouchers, bar or QR codes (something that people can bring with them physically), maybe with email registrations or some other type of pre-sale PII that you could use to link the activity. Ultimately, you’ll have to look at how offline sales happen in your organization and find a method that won’t be disruptive to the natural flow of those sales, but will help you get this data.

Also, don’t fall into the trap that so many businesses fall into: this data is like a drug. Know when to stop chasing rabbit holes and get back to work. You’ll just frustrate yourself and you’ll be no closer to discovering a colony of jackalopes, which is what you’re doing when you’re trying to make this data perfect. 80/20 rule all the way.

Q: What is the best way to measure and communicate “available demand” from available channels (social, search, display) for forecast modeling?

For search, you can use Google’s set of tools to get volume data fairly easily. For display, there are similar tools that will talk about reach that are a part of your ad network and delivery service. Social, it depends on the medium, but I would say that social is a marketing arena, not an advertising arena. The difference being that marketing is about learning, not just about sales. You have to know your market, whether there is cash or no cash in the interaction. What’s the ROI of your business plan, for example? Probably the whole future value of the company, arguably. So ongoing marketing has a similar value proposition. If you don’t do true marketing instead of just broadcasting your message in social, I’d be so bold as to call that downright idiotic, given how incredible what you can learn about your business through social media is.

Ultimately, regardless of the channel you’ll have to take the plunge with some level of budget to see if you can generate a true business ROI. Not a media ROI (revenue/cost), but true cash flow. The good news about the web is that you can learn quickly and adjust. So if you’re building a business plan or working with people who will be providing capital to the business, make sure that they understand it is impossible (and unintelligent) to go into launch with too rigid a plan and/or forecast of channel performance and budget allocation within the media mix. Also remember that if you are ROI positive, you should be able to reinvest those funds in your marketing, rather than having to continue exhausting capital or pay back investors or debt early in the game.

Q: How to gain insights to improve competitive keywords rankings of websites? And to how to visualize it in a report?

I’m assuming this is about search engine optimization. Step 1: hire a tenacious and reputable SEO. Step 2: see step 1.

The first thing you need to realize is you don’t care about rankings. You care about customer creation. Again, keep it about the business, not the web site. Now that you have your business hat on, figure out where you need to focus to gain customers via organic search. It’s probably going to take content development, particularly around the types of things that help people make decisions and learn, not just about some landing page that says “buy today.”

This is where businesses fail online where they often succeed in-person. I don’t walk into a Home Depot and see a bunch of people who just keep running on listing features with a button on their forehead that says “buy now!” I instead find people who are genuinely helpful in helping me pick a color to paint the room, talk to me about how that will make the room feel, and make me feel informed about what I can expect with the project, especially where to start painting because the first wall I do will definitely be done wrong, before I learn how to paint better than a 5 year old. Or I might talk to someone who walks me through their own experience with the Weber grill I’m considering buying.

When we bring it back to the web site, though, most sites in the commerce vertical have so little genuinely helpful content, or content that isn’t purely bottom-of-the-funnel focused. Inject some humanity into your web site and not only will it help your customers, but it’ll show Google that you really know a lot about a topic, which leads to rankings.

The key to these rankings is in how comprehensively you cover topics, not little tricks that fool Google. Just like how the experts you know personally understand their topics deeply, Google wants the same from the sites they promote to the top of rankings for certain keywords. You’ll also notice that competitive, high-volume keywords are mid to high-funnel, more information-driven keywords, not imminent conversion keywords. Winning there requires real content.

Your SEO partner will handle the report. This will be the easy part, and it’ll need to be somewhat tailored to your business, so that’s a little hard to answer, generally. But if they focus on your business, they can tailor reports to what you can act on.

Q: What is the most superficially tempting metric that is almost never worth worrying about?

I would say it’s not a metric; it’s pathing reports. Pathing is a consequence of your design, and doesn’t even remotely reflect consumers’ true intent, because your web site is designed in a way that influences behavior. While a web site is much more fluid than a real world store, for example, obsessing about pathing is still a little like asking people what the first thing they see when they come into Best Buy is: duh, it’s the stuff in the front of the store. Customers’ paths in the store reflect the layout of the store. On the web site, it is definitely more fluid and you have more options, but the layout of your interfaces still influences behavior in a major way. Plus, the paths get so fragmented so quickly (in both cases) that they’re barely usable under any circumstances.

We have to ask ourselves: What are we trying to learn with pathing? Is it what our customers want? We know now we can’t do that. Are we trying to learn if our design actually works? Well, testing is a heck of a lot better way to gain insight into what works, which is a lot more valuable than knowing whether something works. It’s a lot easier to react to analog data like, “What types of things work better?” than binary data like, “Does it work?”

On the metric end, though, exit rate gets a bad rap. I do think this is a ridiculous metric if your aim is to keep people on your site for the rest of their lives, but if you look at your top exit pages, this is a good way to see if there’s the possibility of better serving that audience. Hopefully, one of the top exit pages is your order confirmation page, for example. If this is the case, do you think there’s anything of value you could put on this page? Tell them about your social media presence, your weekly specials email list, or just give them a link to something you think is cool about your company or some other wondrous thing on the Internet. Don’t just wave goodbye after you get their money. Genuinely thank them in some constructive way; humanize the site. Don’t mistake me as saying you should try to contort them to do something they don’t want. If your social media campaign is just a broadcast, do you think they want that noise? No. But if it benefits them, tell them.

Q: How can competitive intelligence be used to estimate competitors’ conversions?

Some tools like comscore will let you purchase this type of data for a fee. It’s fairly straightforward. But keep your eye on the ball because I’ve seen this type of data derail conversations and whole businesses for months on end. Every sliver of data your company gets, particularly when they pay for it, should be about one thing and one thing, only: action. If it doesn’t cause you to act quickly and frequently, drop it like a bad habit. If this type of data paralyzes your company in a whirlwind of meetings and everyone’s opinions, you need to seriously evaluate whether this data is actually value-positive.

Q: What’s the most accurate way to discern a competitor’s web traffic (at least visits)?

Hire one of their analysts!

Corporate espionage aside, go to a competitive intelligence tool, preferably a free one like compete (for this type of data), get a comparison of your site vs. theirs, see how wrong the data is about your site, realizing that at best, this is a scalar comparison, and get on with your day.

You already know how to earn or buy more traffic. You already know how to generate sales on your web site. Focus on those two things and growing your business. Even for a media company, focus on how engaged your visitors are, rather than basic traffic volume. While the display market today values uniques, visits, and pageviews, this really is old thinking and doesn’t really have any bearing on how successful their display campaign would be. I’d rather have 100 impressions on a small-circulation article about Oprah’s endorsement of my product than 10,000 impressions on CNN’s home page, if I’m smart.

Q: Since our company is new, how would I optimize with sparse data?

Starting a new company is like launching the space shuttle. The tiny changes in direction that happen just after lift-off will mean a difference of thousands of miles later in flight or can even throw you completely off balance. Later in flight, it’s easier to keep moving forward while you try new things.

If it were my business, I would say this, “I only know the slightest shred of what my business will look like a year from now.” I want to be highly reactive and a true marketer, not highly active and a loud advertiser. It can be hard to walk the line between being reactive and being fickle, but it’s something good businesspeople manage, and if you’re not a good businessperson, I can’t help you in a 10 question interview 🙂 But it’s important to use everything you have data-wise to make those early decisions count, because early decisions matter so much.

To make this more concrete, start out by looking at your visitors data, your traffic source data, and your content consumption data. Where are people coming from, what kinds of people are they, and what are they looking at? What attributes within these sets make those visits more or less valuable? Why might that be? Do some user testing or set up A/B and multivariate tests to confirm your thoughts on how people interact with your brand by testing the response of each of the “value drivers” you found. And for search traffic, use their keywords to help you understand what they were looking for in the first place. How well are you satisfying those array of needs? Are you the Home Depot store, helping people out with their broad base of questions and needs, or are you just a robot repeating, “buy now, buy now?”

I think what separates great business from regular ones online is that the great ones realize that they create the data they need to make decisions, and I don’t mean they do it in the same way as Enron. Everything you do creates information. You used a lot of information or research to figure out what to create in the first place, but the greatest value is in watching the data that flows in once your creation exists and using the information you actually give life to. For example, when Boeing makes a new model of aircraft, an outrageous amount of research goes into its design. But ultimately, the design will be tweaked thousands of times based on human feedback when pilots and passengers use it in the real world.

So I would say that the focus shouldn’t be on just reacting to sparse data but aggressively figuring out what data you need to create through your own actions and testing to learn more about your audience. Your sparse data is a lot more valuable when you create it with a purpose, like testing an offering, than when you call your site a finished product and just take what the world gives you.