Figuring out how customers find your website is perhaps the biggest challenge of running any online business. As the president of technology at MTG Card Market, my responsibility is managing our technology stack, marketing and infrastructure. That means I'm the guy in charge of web analytics and SEO, alongside UI/UX.

I've learned a lot building Quiet Speculation and I've been applying the same skill set to MTG Card Market. A basic primer on how to figure out your traffic sources and user behavior patterns seems to be in order for those starting their own sites.

I use Google Analytics and, while others use things like MixPanel, I'm used to GA and have been using it for 3 years. It's easy to use, easy to set up and tells me what I need to know. I can also automate reports to send my team on a weekly basis. I set these reports up once, deciding which variables and dimensions to map, chart and visualize, and they get automatically generated and shipped out on whatever time table I so desire.

Having an analytics package in place is an absolute non-negotiable for an online business. You simply need to understand where your traffic comes from, what your users do, how long they spend on the site and from which pages they exit. While there are dozens of other dimensions you can track, these are crucial. Going forward, this piece will be assuming the use of Google Analytics.

Communicating With Your Team

To supplement these automated reports and infographics, I do a weekly traffic report write-up that I send to the team. In the early stages of development and launch, you'll want to explain to your team exactly what the metrics mean, how they compare to previous timeframes and the business implications behind them.

Remember that the majority, if not all, of your team is non-technical. I'm the Geek of Geeks: the rest of the team are MTG players/traders and/or investors. They have no idea what a good "bounce rate" is, nor do they know how to read a page navigation summary visualization. They want to know the take-aways, actionables and forecasts.

One of our investors relayed some great advice he received from a mentor: He explained that I don't need to give him the whole story, just the 5 basic points that the executives need to understand.

Initially, I will explain the whole story to ensure a baseline level of competency on the team, but going forward I avoid rehashing the fundamentals. They can ask for clarification if they need it.

Different Traffic Sources

Direct Traffic: This is traffic that comes from a visitor typing in your URL (http://www.mtgcardmarket.com) or clicking a bookmark. This is not an especially useful number, since there is no one source you can verify. We hand out business cards and stickers at events, which can easily turn into a direct traffic referral. If you get a biz card and wish to visit our site, you'll just type in our URL in all likelihood.

Organic Search: Visitors who are searching on keywords relevant to your site's topic or name. We probably get a lot of people who search "MTG Card Market" in Google instead of typing in the full URL. Again, this tells us very little. Unfortunately, recent changes to Google Analytics make it impossible to know which keywords are being searched, so we have to play stupid guessing games.

Paid Search: Only applicable when you run an AdWords campaign, these visitors are directly correlated to the ads you pay to display on Google. I generally do not like this scattershot approach to traffic and publicity, so I cannot speak to the method any further.

Referral Traffic: This is the one you care about. We can break down exactly which sites are sending us traffic. It should come as no surprise that our biggest referrers are Facebook and Twitter, given that I have a large following on both channels. We use these channels to fire off deals, specials and to announce new rare items added to our site.

Knowing your referral sources is crucial, since you can easily target the effectiveness of different strategies. When I first started QS, I did nothing but put a link to my site in my forum signature on assorted MTG boards. That generated bout 100 hits a day, enough to bootstrap me into bigger and better things.

Coupon Codes for Tracking

Sometimes you can't track the referral source, as with the business card example above. In this case, you can opt to add a coupon code to the mix. Your eCommerce package should allow for this. In the biz card example, we could hypothetically include a coupon code on the card itself, like "BCARD001", that gives 5% off. (And no, that code is not active. Don't bother!) This way, we can check redemption logs to see how many of the biz cards converted into sales.

Derive the cost of a biz card, multiply by the number you give away, and compare that to the number of coupon redemptions. You've just figured out the ROI on business cards! Don't forget to account for the 5% hit on your retail price too.

What Are These Numbers?

Here are a few numbers that you'll want to focus on when learning the basics of your analytics package:

Visitors and Unique Visitors: Google Analytics can tell when you've been to our site before, and while we want to know how many of our visitors are returning vs new, that's a separate metric. The Visitors number tells us how many times our site is navigated-to. Unique Visitors removes all the duplicates and tells us, approximately, how many individuals are coming in. While it's harder to account for visitors using multiple devices, the number is accurate enough.

When you correlate your unique visitor count to your sales numbers, you can start to get an idea of what the average visitor is "worth", how many of them reliably convert into sales, and if you set up your Analytics right you can even see the path the user took through your site, from entrance to the closed sale.

New vs Returning: I personally like to keep this number close to 50%. The farther it deviates from 50/50, the worse, because we are either not getting enough new eyes on the site or we are not retaining enough customers. I'm glad to say that MTG Card Market is hitting this target very effectively on an ongoing basis!

Bounce Rate: Ever get to a site, finish the article and the X out? That's a bounce. Defined as a visit that leaves the site instead of viewing other content, a high rate of bounces is not disired. Bounces happen, but if it's over 30% you should be worried. You might want to make sure that other content on the website is easily accessible and has some solid incentive to encourage readers to keep reading.

Mobile vs Desktop: We designed MTG Card Market to be very mobile-friendly. I just spend an evening at a lecture on UI/UX design, and I learned that the average eCommerce site has a 90/10 split on desktop/mobile. This number is considerably higher outside North America, since most of Africa and Asia visit websites almost exclusively on mobile devices. As this number grows in North America, the importance of your mobile site grows too.

You also want to take into consideration that, in my experience, mobile users spend between 50%-80% less time on your site and visit fewer pages overall. Considering most mobile users are engaged in short, 2-minute bursts while waiting in line at the bank or on public transportation, you want to remove as many finger-taps as possible between the customer's first navigation and the point of sale.

Tying It All Together

While the world of analytics is vast and the dimensions you can track are almost limitless, this should give you a baseline level of competency when you fire up your Google Analytics account for the first time. Get creative and try to think about different ways you can mix and match dimensions to try to predict user behavior.

There are so many variables: perhaps you want to know how iOS users in Canada are finding your website. Maybe you're curious to see how many non-mobile users came from a link on QuietSpeculation.com. The rabbit hole goes far deeper than you can imagine, but if you're a creative scientist, there's no end to the correlations you can find.

As Barney Harford, the CEO of Orbitz.com, told me, you want to capture every single click of the user's mouse if you can. There's no downside to gathering data from Day 1, even if you have no idea how to use it. That's the beauty of data: there are usage cases you can't even imagine until you've already got the data.