By Anne Holland, President
According to MarketingSherpa data, only 28.7% of online marketers have Web analytics data at their fingertips and refer to it frequently to make better decisions.
Everyone else either relies on "gut" or "guestimates" because they either don't have decent analytics systems or they are "too busy to plow through the stack of reports."
But, checking the actual data pays off. Marketers who frequently check their analytics reports tend to get around 25% higher conversion rates than everyone else does.
Last week I flew to Utah to meet hundreds of marketers who are among the lucky 28.7% -- those who take Web analytics very seriously and are able to dedicate at least part of their working days to analyzing results. Here's a quick look at what I learned from them.Three quick action items to improve site results
Action Item #1. Super-simple design rules
The most successful Web pages are BORING. Most designers would take one look at them and beg to be allowed to add something to "dress it up."
In general, more columns, more images, more razzle-dazzle (Flash), more hotlinks, more colors, more paths ... equals lower conversions. A confused visitor will leave quickly. A confused customer will be annoyed with your brand.
That said, test it. If you've got an idea that could help one of your major visitor segments more easily figure out where to click next, then test that puppy. That's what analytics are there for.
(Side note: A panel that included a marketer from Macromedia concluded "95% of Web marketing design using Flash is inappropriate -- it's not helping the visitor scratch their itch. However, one panelist said if you're marketing to Japan that may not be true.)
Action Item #2. Track cannibalization between competing links
Yeah! You tested adding a hot offer or in-your-face graphic to a page and it's getting great results. Not so fast, say many analytics experts. Are you also tracking clicks on other links and offers on the same page or site?
Are increased clicks in one place cannibalizing the results of another? And, most importantly, which is ultimately the most profitable link path to drive traffic through?
As one marketer put it, "Don't just track conversions, track anti-conversions. Use fall-out reporting."
Action Item #3. Instead of pops, send image-clicks to another page
As we've reported in the past, loads of people click on images on your site. If your images are hotlinked so clickers actually see something new, congratulations, you may get better conversions than most of your competitors.
Now test the next step -- instead of presenting a pop of a bigger image, open a new window to an entirely new page that includes a bigger graphic plus sales copy plus an action button. More pages can help search engine optimization, plus you're letting the visitor follow the conversion path they're most interested in. Segmentation is the hottest measurement standard - No traffic is "average"
After more than five years of relentless marketer-education efforts by consultants such as the Eisenberg brothers at Future Now, Inc, persona-based Web design is finally coming into its own.
In simplest terms this means figuring out what sorts of people come to your site -- perhaps dividing them into three-to-seven distinct audiences with their own unique goals and personalities.
So, there's never an "average" or "typical" user of your site. Instead you've got a few different populations. For example you might have:
-- Newbies vs. multibuyers
-- Techies vs executives
-- Parents vs kids
-- Impulse buyers vs careful researchers
-- Discount-driven vs. quality-driven shoppers
-- Service seekers vs. shoppers.
Generally, none of these different groups are remotely interested in content aimed at the other folks. (One marketer told me at lunch he was surprised to see how little overlap in pages viewed his site's five main visitor groups had.)
Obviously you need content specifically created to please each group. Plus, you need to alter your site navigation bars, hotlinked text, and even site tabs, so these serve as overt flags, each waving a particular group of folks over to look at "their content."
Site navigation and copy is not about your company, your brand, your services, your products. It's about helping each very distinct group of visitors find what they are looking for more easily.
Today's better analytics programs make it fairly simple to split your traffic by these segmented groups and follow their activities. Then you can make separate tweaks to increase each unique group's conversion rates. (Conversions can be sales, registrations generated, clicks, surveys taken, etc.)
Plus, you'll want to track acquisition costs and lifetime value separately for these groups. Anyone who talks about "average" profit per "average" customer is glossing over giant differences between groups.
If you don't know which customers are more profitable, you can't go out and get more of them can you? How to track customer lifetime value: seven key factors
Otherwise highly sophisticated online marketers often equate lifetime value to initial order size. In fact, most marketers' paid search acquisition budgets are dictated by average first-time order size.
(Or, alternatively by the size of your CEO's ego when he/she demands you spend silly money to buy generic terms he/she thinks are critical to appear under regardless of proven conversions.)
Unless you sell only one thing and there's no other way to ever make another dime from that new customer, lifetime value is a glorious many-flavored thing. Seven key factors to weigh in customer valuations follow:
o Profitability -- what's their average order size, cost per unit, amount of handholding required (some demographics may require lots of service or sales rep assistance), and likelihood of cross-sell/upsell?
o Renewability -- do they order things you can sell to them repeatedly (updated versions, refills, subscriptions, etc.)?
o Recency -- when was the last time they ordered from you? Those who ordered more recently are invariably more valuable because they're more likely to convert for additional offers.
o Frequency -- how often over an average course of time do they order from you? Are they one-shot buyers or do they return on a fairly regular basis that you can predict and encourage?
o Offer preferences -- do they leap for anything that's "new"? Are they only motivated by discounts? Are they clearly gift-shoppers buying for others on a predictable schedule? Do they view you as a commodity and would be easy for a competitor to poach from you?
o Evangelism/advocacy -- do they adore your brand and will refer other new customers to you?
o Repeat visit bounties -- once they've found your site, how much will it cost you to get them to come again? Will they come directly to you (or via an email link you send them)? Or are they addicted to shopping engines, paid search engine ads, affiliate links, etc.?
Once you have this data for each of your main groups of visitors, you will know which to focus more visible site navigation elements and copy for. Three tactics for measuring consumers across multiple channels (online and off)
For most marketers, lifetime value also includes offline activity. Merging online and offline analytics is a massive headache that started in about 1998 and shows no sign of letting up yet. (Heck, just getting your email data tied into the Web is hard enough.)
Three clever tactics I heard folks mention at the Web analytics conference:
Tactic #1. Campaign-specific inbound phone numbers
Depending on how prominently you display a phone number, calls can run from 2%-20% of your responses. 10% is a good bet for sites with complex offers and very visible phone numbers.
Several sites at the conference (and others I've heard of) have asked the phone company for as many phone numbers as possible so they can use them to count inbound responses to particular campaigns. You're limited by the amount of numbers you can buy, but it's better than no data at all.
Tactic #2. Warranties or product registrations with gift online
If you sell via resellers or offline retailers, test offering a gift of some sort in your packaging to incentivize buyers into registering their purchase on your site.
Unlike general name acquisition campaigns, you can make the offer as soft and sexy as you want and still get good names. After all, these are *proven buyers*. It's the most valuable list you'll ever acquire. So don't be chintzy with the offer. Plus, use your analytics system to carefully test the landing page to maximize registration conversions.
Tactic #3. Personal offers printed on paper receipts
If you sell via brick and mortar, consider adding an offer onto all customer receipts. Perhaps it's a quick survey, "opinion poll," or a free digital gift (a printable coupon, a music download, a PDF, etc.).
The key is each entrant must type in their unique ID code from the receipt. That way you can tie the offline buying info together with their online registration data, creating a truly powerful customer database.Three challenges for integrating multiple data sources into your Web analytics stream
Everyone in the audience gasped when HP's customer data team announced they have collected 18 terabytes of data so far, tracking everything from Web activity to email responses by customer.
Macromedia added they have 60 million customer records and counting. But then AOL's marketers trumped everyone by casually announcing they are at 100 terabytes, including offline and online marketing activity.
However, the size of the database needed to house all your records isn't one of the big challenges anymore. Server space is relatively cheap. The three biggies I heard were:
Challenge #1. Picking a common denominator per customer
Once you've decided to track customers via every media channel they interact with you, how do you pick the one unique item in a record that will identify that user no matter what channel they are in?
Most consumers have multiple (and changing) emails and multiple (and changing) credit cards. Plus, they wipe cookies, shop from different PCs, visit different store locations, and if they obey the IT department's urging, they've got a few different user names and passwords, too.
If you want to track customer touches -- inbound and outbound -- inevitably you'll get a lot of duplicate records, which makes your data less useful. Just consider everything you could consider a touch:
o Site visits
o Email responses
o Online advertising responses
o In-store visits
o Phone calls
o Sales rep meetings
o Webinar attendance/white paper readership
o Direct (postal) mail responses
If you can tie this together with one common denominator per customer, you'll really know which customers and demographics are hugely interested in your brand and which are ignoring you completely in every media.
Challenge #2. Getting everybody in your company to agree to mingle data
Larger organizations at the conference told me horror stories of various divisions and departments all using different Web analytics systems that didn't talk to each other (not to mention email vendors, ad agencies, etc.).
Politics can be hell -- especially when it comes to everyone agreeing on one common method and system to measure and store data. The most technically advanced companies often have the worst "data silo" proliferation because each department leaps to buy technology quickly rather than waiting for an organizationwide consensus.
The key to success? Top management must pick one exec to head the data project and empower that exec with the budget and power to bring everyone else under a common system umbrella. Sounds impossible? HP did it.
Challenge #3. Training everyone to use the mingled system
Everyone agreed the best way to get marketers to actually use the new system is to have one report "front" even if the database has many mingled streams on the back end. That way everyone (except for true experts in the data department) only has to be trained to use one system.
Do your darndest to make that one single front as easy to use as possible. It's all about usability.
Then you must routinely seek out and integrate "Excel spawn" -- spreadsheets that various marketers invented on the side to track something they couldn't use the official report front for. These spreadsheets are like weeds; there will always be new ones.
Given that this was a Web analytics conference, most marketers were talking about tweaking their analytics dashboard to serve as a front-end display for other data streams as well. So they might integrate any of the following into their regular Web analytics reports:
-- Sales lead management and CRM systems
-- PPC campaign results
-- Email campaign results
-- Offline results data
-- Appended demographic data from outside databases (Acxiom, etc.)
-- Survey results/visitor feedbackBiggest learning? Reality equals multiple touches, multiple media
When you begin integrating all data into your Web analytics stream, you learn the universal truth that most of the time no single marketing campaign or media is solely responsible for a sale.
These touches are happening in multiple media and each needs to be credited. For example, your retail store may have gotten the sale, but your Web presence helped educate, build awareness, build trust, etc., to enable the offline conversion.
Yet, when Web Analytics Association President Jim Sterne asked the audience, "How many of you have produced a report that produced change in your company offline?" a tiny number of hands rose up. Pretty sad.
Aside from the obvious stuff -- improving and measuring your Web site -- Web analytics can provide invaluable research data to improve offline marketing campaigns.
Suggestion: Perhaps your analytics department should schedule a quarterly presentation to the offline marketers and agencies to reveal the top data points you've learned about customers and prospects that might help offline marketing.
Have you discovered a particularly profitable customer segment that offline ads don't target yet? Are there any powerful keywords they should add to offline copywriting? Are there any particular SKUs or offers they should test in offline promotions due to unexpected online success?Small budgets, eager VCs, and industry m&a
Like marketers everywhere, attendees complained about their budgets. Hardly anyone was pleased with the amount that management thought analytics could make due with. (Even big household name companies were complaining.)
That said, according to a budget survey MarketingSherpa ran in partnership with AD:TECH in December 2004, 52.7% of online marketers plan to invest in new site analytics software in 2005.
Which explains why I met several VCs (venture capitalists) at the conference -- felt like 1999 all over again.
It also explains why Google's snapped up a big analytics firm (Urchin); NetIQ's management team decided to take the company private so it's no longer part of WebTrends; and every email service provider on the planet is seeking more Web analytics system integration agreements.
I expect to see loads more business news with firms merging, revamping, partnering, etc.
Which makes it harder to buy analytics in the short-term because you don't know what the vendors you're talking to now will look like in six months or a year. Be sure to ask about deals on the horizon that might affect you, but don't expect an entirely straight answer. Reps may not know everything or be allowed to tell you.
Good luck! Useful links related to this article
Omniture -- the Web analytics firm that hosted the conference I attended in Utah last week: http://www.omniture.com
Web Analytics Association -- note: this is a new organization and I'm on the advisory board. Definitely check them out and let me know what you think: http://www.webanalyticsassociation.org/
MarketingSherpa's Buyer's Guide to Web Analytics: http://sherpastore.com/c/a.pl?1150&p.cfm/2146
Web Analytics Demystified: A Marketer's Guide to Understanding How Your Web Site Affects Your Business -- useful book by conference speaker Eric Peterson: http://sherpastore.com/c/a.pl?1150&p.cfm/2148
Future Now, Inc -- Persona-based Web design experts I mentioned above: http://www.futurenowinc.com
Emetrics Summit - a useful event you might consider attending if you're a Web analytics groupie: http://www.emetrics.org