September 28, 2006
According to MarketingSherpa's new Search Benchmark Guide, consumer marketers have dramatically slowed their PPC keyword growth over the past 12 months. However, B-to-B marketers are going in the opposite direction.
Should you be? Here's a Case Study of a real-life B-to-B lead generation marketer who changed bidding tactics *and* tested landing page changes.
Here's their story, including CPA and conversion rate result charts:
Like many B-to-B lead generation marketers, Teri Wiegman was at first thrilled by the promise of paid search advertising (PPC).
As VP Marketing for Onvia, a database provider serving the government and education contracting marketplace, she authorized toe-dipping search tests in late 2004. Her search team started by buying a few thousand clicks. This blossomed quickly into tens of thousands, then hundreds of thousands; and after a few months they'd cracked more than a million clicks per month.
At first giddily welcomed, the expansion began to feel like a runaway train. In less than 12 months, a huge swathe of the budget was dedicated to SEM. Wiegman wondered whether she could reduce her cost per lead without risking missing any good leads.
Last year, the standard answer to that dilemma was to continue broadening the campaign, casting a wider net for "long tail terms" that might be cheaper per click -- so the cheaper clicks would balance the expensive ones and bring overall campaign cost per click down.
It's a great idea, until you max it out. What's next if you need to continue cutting costs?
"I'm always running a test on something," notes Wiegman. So, she and her search team tested their way through the next year. The tests fell into what seems remarkably like a carefully planned three-part strategy. "It wasn't by design. All the time we're trying to flush out patterns."
However, she came up with a strategy; it's one worth following:
Step #1. Segmentation by lead readiness
As leads came in through the search click landing pages, Wiegman gave her in-house sales team 24 hours maximum to follow up with each one of them. The goal: to gain enough further information to rank that lead in any of three buckets:
a. Nearly ready to close (aka "hot") -- needs to be convinced by the right product and offer.
b. Middling -- requires a bit of nurturing but could close in a few months or less.
c. Absolute beginner -- at the 101 stage, needs serious education about the industry before even starting to consider buying the service.
Next, Wiegman's team created a chart to cross-reference the lead bucket with the search term that lead had come from. Yes, a direct correlation existed between many specific search terms and stage in sales cycle.
As their next logical step, the team developed and tested three types of landing pages -- each tied to a different group of search terms based on sales readiness. (See link below to creative samples.) For example, the team sent absolute beginner clicks to an educational microsite titled 'Government Contracting 101.’
Although it's not normally a best practice in landing pages, the page had many navigational cross-links to more advanced offers and pages. This made sense because more serious leads could graduate themselves "up" the chain without the risk of anyone going further down by mistake.
Other test landing pages included "medium" offers such as registration forms for a free sample guide, free demo or free consultation. (Note: All three amounted to about the same thing, but the descriptive language was worth testing.)
Step #2. Optimize landing pages for lead conversion
Again, the team examined the leads generated, this time placing them in buckets by industry. Turns out certain verticals, such as construction, tended to generate both high quality and quantity leads via search.
Wiegman authorized a series of 10 A/B landing page tests for the top 10 industry verticals the team had identified. Each test ran a generic landing page versus one that was customized for the industry by having the keyword such as 'construction' placed in the headline and in repeated spots all over the page.
She suspected this would lift response rates, given data MarketingSherpa has reported frequently in the past. However, would response rates raise high enough to prove it's worth the time and cost to build custom pages for every keyword?
Step #3. Cut low-value campaigns
At last, the team again examined incoming leads with an eye to cutting campaigns that simply weren't performing against budgeted quality lead cost per acquisition (CPA). In February 2006 and again in July 2006, the team cut campaigns producing lower-quality leads -- slicing their average clicks per month bought to half of what it had been in January.
The goal: to keep the in-house sales team toped up with a solid pipeline of high quality leads.
As you can see from the results charts hotlinked below, Onvia has cut their cost per lead by more than two-thirds since they began to optimize their search marketing campaigns.
Ultimately, they decided to shelve the general 101 educational program for now -- it produced too high a volume of clicks that weren't anywhere near qualified. However, initial optimization tests for the middle-stage and later-stage leads proved well worth it. Two findings:
1. Offer wording matters -- "free consultation" wording didn't generate as many leads as "free sample guide" wording (for the same offer).
2. Plug your awards -- when Onvia won a Codie award, they tested adding the icon to landing pages although few if any prospects would know what the Codies are. The icon definitely helped response rates -- any award is better than none.
The A/B tests for vertical landing pages were so successful that Wiegman wound up authorizing the team to invest building 120 additional vertical landing pages. Despite this extra cost, her final cost per lead continued to hold steady because conversions improved so much.
Lesson learned -- If you have to choose between paying for landing page tests versus paying for more search clicks, the landing pages should win.
Useful links related to this article
Side note: If you'd like to see what Wiegman looks like, she's the tall woman in the photo on page three (no British humor here, please) of MarketingSherpa's B-to-B Fall Summit Brochure. We snapped her pic when she attended last year's conference, and she admits it was slightly startling to see herself in this year's mailer! Here's a PDF of it: http://www.marketingsherpa.com/summits/DemandGenSummit06.pdf
Creative samples and results charts from Onvia:
Point It! Inc. -- the search marketing firm Onvia relied on for help with this PPC campaign
ClickTracks -- Web analytics software Onvia uses for its site currently: