Large Scale Bid Management

The PPC programs from the Search Engines have grown more complicated and robust over the years. The amounts of variables have increased as it relates to ad text, bidding, budgeting, targeting, positioning, and distribution. While some small and well-structured campaigns can run with little or no maintenance, the challenges for managing large scale PPC campaigns across multiple channels compounds daily. This session examined real life examples of large scale bid management, techniques, and tools.

Moderator: Ken Jurina
Speakers:

Jon Kelly, President, SureHits
Kevin Lee, Executive Chairman, Didit.com
David Rodnitzky, Vice President of Advertising, Mercantila

Jon Kelly, First Speaker:

When managing a large scale PPC campaign, there are 3 key actions to look at:

1. Calculate click value –if you don’t know what its worth, don’t know what to bid
2. Reward the users choice
3. Watch your campaign data.

Click value: probability of conversion, value of conversion.
The higher the probability the click turns into a lead, the more value the click has.

Model the Market – geography, product and request.

The wrong way: buckets. For small, OK but the larger the size of the site, the more punishment. For example, put every phrase with Kansas in one bucket. You can also do a bucket with all product phrases. Or, a bucket for requests, like applications, rates, etc. The search phrase is the window into the user’s mind to see where they are in the process.

Series of tags: “Topeka Kansas home equity loan rates” (very tail end). We can then construct a model to predict the probability of a conversion on that site. City (minus), State (plus) home equity loan (plus) rates (minus). “Rates” is still in a shopping mindset.

What is the value of the lead? Here, Kansas might be a great state b/c not as many competitors, etc.

Note – Click value does not equal bid.

Second important point in a big campaign is “reward the user’s choice”. You are throwing away value by not responding to that. We know the person is looking in Kansas, home equity loans, and rates. Best practice is putting out the offer the consumer is looking for with the search phrase.

Finally, watch your campaign data. In many cases, there are specific issues as to why phrases in a large campaign might not work as well as others. Someone typing in “Houston Mortgage” might be looking for the company of that name, not a mortgage loan in Houston, and the traffic will not convert otherwise.

A search phrase is a consumer making a specific request.

It is important to note any unusual amounts of traffic from tail end phrases. Also look out for homonyms, i.e. Mobile, AL home loans may give you mobile homes in the state of Alabama.

Missed geo-targeting: look at data to see how good it looks. Go through the data and look for outliers to increase conversion. In a larger campaign need more tools and ways to look at data.

David Rodnitzsky, Mercantila (specialty retail stores)

Large scale bid management for the rest of us: a lot of people who use paid search but don’t have the right staff to look at everything.

Assumption: you are not Ebay or Amazon, don’t have a lot of resources.

Step 1: Assess your expertise.
Assume you are not an expert with paid search. Advice is to outsource it, don’t try to solve something other people have already solved. If you are an expert, but don’t have a huge campaign, get 3rd party bid management software. If you do have a team, do in-house.

Step 2: Define the tail: If you are an expert but don’t have a team, define difference b/w head and tail terms. For the stuff that you will not focus on, either outsource or use management software.

Step 3: Simplify the Tail
– Don’t try to get super complicated with position basing
– Use standard campaign negatives
– Generic ads that work across all campaigns
– Search and destroy bad keywords weekly

It is great when a tail keyword converts, but not if they lose you a lot of money. Make it simple, and if it doesn’t work, get rid of it.

The tail fairy tale: what worked 3 years ago with broad match doesn’t work today. Google has broadmatched out the tail keywords. When looking at tail keywords, focus on simple rules.

Step 4: Focus on the head, once you’ve got your tail. Focus on the right keywords, the right message, the right bids, and the right URLs.

Expand top keywords: misspelling, plural/singular, prefixes and suffixes etc.
Implement advanced bid rules: optimize for position, day-part, geo-target, IP/Site target, factor in assists and latency (Yahoo).

Improve your ad text: Ad text war – take 3 best ads, run for a week, see which one works, remove the others, keep going etc. A lot of ppl focus on CTR, not conversion rate which is wrong. CTR is great, but if they are not converting, we are just giving Google money. (In one example, an ad for Findlaw read Findlay and it drove a lot of paid traffic to the site.)

If you want to keep it simple, 4 thing ppl always respond to:
-greed
-variety
-exclusivity
-fear

Test landing pages: use Google website optimizer, look at bounce rates etc. make sure the pages have keywords that match the ads. Make sure landing page works.

Conclusion: keep it simple. Don’t need the most advanced algorithms, don’t need 10 million keywords, just focus on what you know (outsource the rest) and what works.

Kevin Lee – DidIt.com

Focus on what matters – top positions when it matters to campaign. The engines are not as concerned with inefficiencies in campaigns because they will make more money. So it is important to manage your campaign. Key is to understand objectives…be in a top position to get the scale when it matters, and to not be there when it doesn’t matter.

Head vs. Tail
There is a keyword tail. You can generate almost unlimited numbers of keywords. Is that really how you leverage the power of the head and the tail? The size of the campaign is driven by segmentation so adding more keywords may not necessarily add value. Segment the head by every possible lever. The audience is different. There are different people behind the clicks! The person searching for a mortgage at noon in NYC is different than the person in the Bronx at 3 AM (day-parting). But there is a lot of opportunity to take the head and slice it up into manageable clusters – analyze head keywords to determine if a segment exists.

If you have existing customer data (offline) that can help, take advantage of that. Look at what you already know about your customers, talk to your colleagues about what you can learn from your off-line data.

Segmentation, modeling is good for the search engines because it results in better targeting, which will get a higher CTR, and then you can pay less for the click. From a targeting perspective, it’s good for everyone. If you look at geographies with critical mass, that can be particularly helpful. Start at the large metro areas and see how the data differs from your average nationwide campaign. If you can get a higher CTR within a specific area, you will get a higher position for less money! Also, you need to think about your creatives because your audience is different! If the clusters are large enough, may be worth it to create separate landing pages as well.

If you have immediate value you can measure (shopping carts), you can run analysis across different segments and see which ones bring higher value, and bid more on your campaigns when it counts for that segment.

A lot of people ask – how much difference is there really, this is a lot of work! Can’t I just set and forget?
It matters! You need to build a model of the changing value of a click over time.

Certain segments are large, like Los Angeles, and bring high CTRs, conversions, etc. which brings higher immediate relevant value, and those people will become the most valuable so it’s worth testing different creatives, URLs, etc.

Conclusion: stay educated on Best Practices, pick the right partners for technology and strategy development, never stop testing, never stop segmenting. Concentrate on the head but turn a little bit more to the tail.

Q & A:

Q: What bid management tools would you recommend?
A: Kevin: I’m in the process of testing a bunch. My experience is there a lot of impressive power points that will prove one tool is amazing. You should do a bake off. Give similarly sized campaigns to 2 or 3 different tools, set a time period to run it, set expectations, and see what happens. Also, you need to think about the ease of limitation and implementation. Many bid management companies want you to use specific tracking and use certain technologies; if you change your tracking URLs you will lose ad history or keyword history. Takes about 7-10 days for history to recover. Just need to test on a small scale, make sure you have a 30-60 day out clause in your contract.