Large Scale Bid Management

Nov 11, 2008 • 8:13 pm | comments (1) by twitter Google+ | Filed Under WebmasterWorld PubCon 2008 Las Vegas
 

Moderator: Ken Jurina Speakers: Chris Knoch, Principal Search Consultant, Omniture Jon Kelly, President, SureHits Gerry Bavaro, Executive Chairman, Didit.com

Chris from Omniture:

Economy of Efforts - acknowledge your resources - determine key priorities and tactics.

Take a step back and understand what you are working with. Large scale keyword campaigns cause SEMers to consistently overestimate their resources and build unrealistic expectations.

Perform and honest review of your resources.

What kind of team do you have?

Internal, agency, engine?

Determine the true head count and capabilities. What is there expertise? What kind of tools do they have? Don't want to create an organizational structure without the ability to execute. Don't try to bite off more than you can chew.

What is most important to your campaign? What are the key priorities and tactics? Can't do everything at once. Do what gets you closest to your goal.

Targeting example: People thing more keywords is better. How much volume and cost savings are you getting from a massive build out? Sometimes the mentality is to do it as a badge of honor, or for pride. Think if it's actually going to improve performance.

Automate bidding: Most bid management systems look at click to conversion rates of keywords or portfolios of keywords. Most automated bid keywords would bid keywords downward. But is that the real priority?

Need to take a step back and dig into the numbers. Large keyword lists face resource constraints.

Make sure you are not following a new optimization trend because others are doing it.

Tactic: Organize your mess. Keyword classifications in bid management. All these people faced with campaign restructuring. What's the best way to do it? Do you want to organize by volume, performance, budget? Let financial performance be the key goal.

An example of how to do it in Omniture Search Center. Basically, labeling keywords with metadata such as "high volume". Find your keywords, and apply labels to them using Excel. Manage via these classifications. Keywords that are tagged as "high ROAS" are the ones to focus on first. Slice and dice data, and apply the right optimizations to them. Change the labels as necessary. Use different techniques for each classification. Set rules - such as position rule, performance rule, or portfolio rule. Create product families.

Focus on financial performance!

Gerry Bavaro from Did It:

Large Scale campaigns - beware more keywords doesn't necessarily mean more success. Don't forget, you pay for your keywords!

Your budget waste goes to the engines bottom line. Inventory may drive long tail, but what's the real gain?

Keyword additions reaches diminishing marginal returns fast.

Maximum returns at the top - the trick is affording it. Power terms get massive clicks, good conversion rates, and good CPC's. It's your go to list.

Segmentation - testing which keywords perform at the highest ROI.

Efficiency: Many companies and people spend more than they make. Efficiency is the new creativity. Creative people in marketing are the ones that know how to effectively pull the levers into the media buys and get the maximum output. With search, you have many levers. Lots to work with.

Most large campaigns are defined by how many keywords. Should ask how many segments you are running. A single keyword can break down into 5, 10, or 100 segments. Can have different values, desires, profitability. Recognizing the power keyword can have a geo overlay. Need to look at all the factors and figure out how to leverage them. Segmentation levers are there, more than ever, beyond keywords to marketing fundamentals.

Pick the clicks you want most and target to that human segment's predicted profitability. Your levers: Syndication, convents vs. search. Syndication site targeting in content networks (display ads?). Products purchased (keyword bid on vs. what was purchased). Age - (five age ranges on MSN). Gender (on MSN).

This data translates your business into search. Why do certain keywords get such high click through rates? Have to look past the keyword - you are buying audience - not just clicks.

Trend noticing is that average order value is going down. People are spending less, tightening the belt. How do you find the keywords that are giving you high order value and keep them running? You might have to drop your ROI. Where are people going on your site? Make sure tracking is in place, analytics is extremely important.

Segmentation drives better targeting. Breaking out the campaigns and figuring how to buy a higher value audience. Brings you closer to the holy grail. Want to maximize revenue per page. Tightening up the campaigns gives you inherent QS increases.

What's possible at large scale? Geographic segmentation is the easiest to do. The levers are there in front of you. A campaign running at a 3:1 aggregate ROI. Find out that NY, TX, and FL have 6:1 conversion rate. Can make better decisions based on this data.

Lifetime value segmentation: Crunch your customer file. Do certain geographies predict higher lifetime values?

Do certain initial purchases lead to latent revenue? Look at the full trail. When you segment - you can segment the segments. How do you tune the creative and landing pages?

In conclusion- It's the data from the high spend clicks that are the most valuable.

Jon Kelly, from SureHits wraps it up. Jon is one of the sharpest minds in the industry.

SureHits Manages large PPC accounts as an agency and for in-house products. Looks at 3 key areas:

1. Calculate: Click value

2. Reward: The user's choices

3. Watch: Your campaign data

Calculating click value: Profitability of conversion. Probability that a person will turn into lead, policy holder, or closed mortgage. Whenever starts a campaign, looks at a market model. Must think about the key drivers that drive probability of conversion in your market. In financial services - its geography, product, and request. Every market is different.

Wrong way, or old way - bucketing. Bucketing by geography, product, or request. The right way to manage the campaigns is by tagging keywords. Similar to a tag cloud on a blog. Tag them based on concepts of market model. Is it a state or city phrase? Mortgage product, refi product? Is it a request? What was it that the user was looking for?

Tags predict conversion of these keywords. In the mortgage market - a city phrase has negative value to them. It generally has a lower conversion rate than someone who doesn't. State has higher conversion rate. Someone who types in "rates" is less likely to convert than someone who types in "quote".

Reward the user choices. Search engine users give you a window into their brain. Searchers tell you what they are looking for. If you track that, and serve a page based on that, you'll do better converting. Need to serve back to the user what they are actually looking for.

Watching your data: City phrases don't work well in financial services market. Lots of brand phrases. "Tampa Bay Mortgage" might be a company name vs. the city modifier. If you are not that company, won't convert well. "Fake" tail phrases: Parked pages are a big part of search. We see in our campaign data lots of phrases that shouldn't have much traffic. The conversion rate will not be good.

Missed Geo-Targeting - lots of data at your disposal outside search engines. Does lots of geo-targeting. Compared two different states and noticed that TX conversion rate was abnormally low. Overlaid with US population. TX population is 30% greater than FL, but with roughly the same campaigns, 74% more traffic came from TX. That was strange. Investigated, and found that the traffic wasn't really from TX, and the conversion rate went up.

Another example, homonyms. "Mobile Alabama" is a classic example. "Mobile Homes" might refer to the noun (homes in Mobile AL) or the adjective (homes on wheels). Human looked at it and was easy to spot. You need automation for calculating keyword click value, day-parting, landing page testing and tuning. But the human intervention side you cannot overlook.

Exception analysis: Humans are much better at it than machines.

Follow Jon on Twitter - @jonkelly

Live coverage provided by Avi A. Wilensky of Promediacorp, a Manhattan based online marketing agency.

Previous story: Discover Techniques Used by Enterprise-Level SEOs/SEMs
 

Comments:

Claire

11/13/2008 02:24 am

Great wrap-up. I agree that customer lifetime value should be taken into account when calculating your optimum keyword bid. Following up with back-end products is a powerful yet often overlooked strategy, especially in SMBs.

blog comments powered by Disqus