Golden Nuggets From SMX Search Analytics
Generally, I’m not much of a note taker at conferences. I will jot down the occasional URL to check out later, but I’d much rather pay full attention to the speakers and their presentations than miss things as I’m trying to write down what they just said. However, at last month’s SMX Search Analytics conference […]
Generally, I’m not much of a note taker at conferences. I will jot down the occasional URL to check out later, but I’d much rather pay full attention to the speakers and their presentations than miss things as I’m trying to write down what they just said. However, at last month’s SMX Search Analytics conference in Toronto, I probably wrote down more pages of notes than I have for the last five conferences I’ve attended combined.
I think the main reason I felt the urge to scribble so much stuff down was the nature of the sessions. This wasn’t a general info event like some of the major conferences that are held each year. This was most definitely a search marketer’s show. Not only that, the focus was on the analytics of search… highly targeted content. I think that the speakers (I was among them) were able to speak at a much higher level than they would normally because of the audience’s skill set. No one had to stop and explain what Quality Score was or define Negative match type. There were a lot of pros in the room and everyone was eager to share, even with their competitors sitting two rows away.
What follows are some of the tidbits I picked at the show. Some of them are tips and tricks, others just stats. There are some things I knew before and was reminded of them, and there were some things I had never seen before.
Favorite tip/trick of the conference: Operation Camouflage which Jamie Smith, CEO of Engine Ready Software, covered in his presentation. Feel free to visit the link above for more details, but basically, this is a paid search tactic to ultimately stay “off of the radar” of your competitors. Jamie showed us how his company identified a handful of key competitors to one of their clients and then set their PPC campaigns to geotarget outside of those areas.
In Jamie’s words, “Our goal in the PPC battlefield is to trick our competition into believing that we have opted not to enter into battle with him for prime visibility of popular keyword phrases. If our competitor does not see us on the battlefield for those keywords, he will likely alter his combat strategy into a more passive mode.” Beautiful!
Obviously, there are some issues here if your competitors are in a huge city center that you cannot afford to drop your marketing. As well, your rivals may have an SEM agency that is another city altogether and your cloaking techniques won’t work. However, I love this idea and plan to use it myself for clients that have one or two specific competitors.
Another good session was a panel that focused on report building. There a lot of things that go into building great reports from understanding the audience, what charts/graphs work best for specific uses, and even how color psychology can be utilized to effectively build your case.
Some other things to think about included:
- using visual representations to “break up” long reports
- listing tactics/goals next to data instead of at the end
- showing trend data so users can focus on one flowing line instead of each and every data point
- conditional formatting in excel to color highlight important cells
- for search pros, use more long tail charts
The most important piece here is how complicated or simple you need each page to be for the report to be successful. If you make it too simple, not enough info will be relayed. If it’s too complicated, it doesn’t matter what you show because it won’t register with the audience.
And now… the speed round:
- Baseline SEO questions before you even start: Are all of your pages being crawled? Are they all being indexed?
- Split your site into intuitive “sections” when uploading into Google Webmaster so you have an easier time with the QA process.
- Use the canonical tag so the engines don’t ding you for duplicate content! This is a fairly new process that you can read more about here
- Use the Google Webmaster tools to see which of your terms rank high in the engine, but don’t receive many clicks. It may help you prioritize your efforts.
- In a way, search engines can detect high bounced pages. How? A return visit to the SERP within a short amount of time (say 10-15 seconds) may indicate a quick click of the back button. This is important as it could (or already be) get included into the page ranking algorithm.
- On the topic of bounce rate, here’s a great post by Google Analtyics Evangelist Avinash Kaushik on tips to stop users from bouncing.
- In reports, “no metric should be isolated without context”. A rule to live by.
- For SEO analysts, try correlating all of your keyword KPIs with your organic rank. So, for example, Traffic by Keyword (by Rank), Conversions by Keyword (by rank), Bounce by Keyword (by rank). It could give you some interesting perspective into how your rankings can affect everything else.
- Phrase of the conference: “Genetic Freak” is the best result from multivariate testing. As in, “I thought the combination of our ‘free shipping’ offer and the new logo would be the best recipe, but the genetic freak was an offer of 10% off combined with the product close-up image.”
- Use the new(ish) Google Analytics advance segments tool. Create some “canned” segments for all campaigns such as “Bouncers”, “High engagers”, etc. That way you can start pivoting around immediately without having to find your own segments.
- Walk in your customers’ shoes. Use your analytics pathing tools as a road map for a site visit; actually go through some of the common paths your users take to see load times, drop off points, obvious navigation issues, etc. Even try to spend the same minutes/seconds on each page as the users do.
- In many multivariate tests, local phone numbers perform better than toll free ones. “800” numbers perform better than other toll free numbers such as 866, 877, etc.
- When looking for paid search ad text ideas, don’t just look at other paid ads, look at organic ad snippets as well. They can offer some good creative direction too.
- Good stat to remember: “89% of consumers research products online yet only 7% of sales happen online.” (Yahoo’s ROBO study)
- There was an interesting product from RevTrax to help bridge the measurement gap of online research to offline purchase. They have a platform that embeds customer data (campaign info, even keyword) into promotional programs such as coupon bar codes, loyalty/reward program codes, etc.
- Google stat: “1 in 5 queries in Google are unique (from the last 90 days)”. That’s amazing to think that 20% of all queries are new (going back 3 months). It speaks to the difficulty of coming up with all of the keywords you need for your search efforts.
- Sneaky tactic of the year: Gain analytics insight on publishers by buying a homepage ad and embedding Google Analytics code in it. From that, you should be able to see visits, bounces, referral site, referral keywords, geographical spread, etc.
- Keep a timeline of everything that happens during a campaign so you can include that series of events next to the final campaign reporting to gain context into spikes/dips that occur. This has can be invaluable for explaining seemingly unexplainable data changes.
- Jordan LeBaron, an SEO guru at Omniture showed off some nifty tricks using the platform. Using API integration, you can pull in external data to correlate to site data. One of the examples he showed was to pull in SEO MOZ data such as “links to page” and “Moz Rank”. That was something I hadn’t seen before but makes a lot of sense. When you’re looking at various pages on your site, it’s good context to see how many back links point to the page—it can help to explain why certain pages have more traffic or keyword referrals than others.
- Use site visits as a stat against every other KPI. Show visits to conversions, visits to revenue, etc. Don’t ever discount the power of total visits.
- The “utm_nooverride=1” parameter can be used to make sure existing referral information is not overwritten by visits that will muddle the data such as a return visit from a direct visitor. If that visitor originally came from an ASK.com link and then they return via a bookmark and convert, the conversion gets attributed to the direct link. The nooveride parameter can make sure the ASK.com credit is still visible. Really great three part article can be found at ROI Revolution.
- Don’t forget to pull reports a few days after you reach the end of a reporting cycle to account for lag conversions.
- Links on the first page of Google get clicked 45 times that of subsequent pages (93% of all clicks come from the first page). This is very important to remember as your biggest organic keyword opportunities are current rankings of 11-15. If you can do something to “bump” them up to the first page, chances are you will reap major rewards.
Based on landing page testing, text under photos are some of the top read text on most pages.
- Good rule of thumb for multivariate testing: For statistical significance, you should aim for 100 conversions per variation. And definitely at least 2-4 weeks of testing before pulling the final results.
- Interesting resource for web usability: Five Second Test.com. This site shows users various web pages for five seconds and then asks them to list every element they can remember. It’s an interesting spot to see how users view page pages.
- Sometimes it’s more important to show clients how much they’d be losing if they don’t take the recommended action versus what could be gained. The impact is more intense that way.
- Even tests that show failure provide valuable data.
- Don’t rely on competitive data too much. There’s a good chance your competitors aren’t very good at what they do and you could be copying “worst practices”.
Thanks to everyone who attended SMX Search Analytics and made it a really fun and content-packed production. See ya’ there next year.
Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.