Presenters Share PPC Tactics

A group of presenters discuss a number of different techniques for understanding PPC and improving campaigns. Apologies for any logical breaks in these notes but the presenters shared a bunch of great information very fast.

Dan Theis

“You must know why things work on a Starship.” – James T. Kirk

Knowing how things work is vital in Adwords. Here are the items that Dan reviewed:

How to Grok

• Nature of space and time
• Locking phaers: match type
• Mind meld with the market
• Scanning the system

Time Sucks…

There are many ways to suck at pay per click campaigns.  Day parting and day of week effects makes a big difference. Some days convert well, some do not. Look at the conversion rates by hour through Google Analytics to identify which are the best time for average order value, better conversions. Also, look at average revenue per visit by different times of the day. Assign a value to leads to understand an ROI.

Set up a break even point – if you are selling repeatedly, then a break even point is ideal, because incremental purchases bring up that return above break even. Look at these metrics by hour. So if you calculate conversion by average sale, get dollar per visit, and compare to the cost per click. This will provide an appropriate bid level.  If a particular part of the day is less effective, then consider doing offers by time-stamp. For instance “midnight offers.” Are there profits that you are missing during the less effective conversion times that can be improved upon?

For days of week – there are different campaigns / experiences / ads that can work differently by day of week. For instance, some might work well in the week and other experiences work well during the weekend. Therefore, set up campaigns specific for day of week to serve the right experience to the right customer / searcher segment.

Space also sucks…

This means that it is important to understand where the best location is for your ad to appear. It is possible to get cheaper clicks lower on the page, but will these people convert? If someone is ready to buy, they are likely to look at top results in paid above natural, then the top one or two spots in the right rail. In short, look at buyer intention – if they are browsing, they may click farther down the paid side, but if buying, then they will look higher up in the page.

Some markets are different. BUT – do start testing at the top. It’s important to understand how well things perform at the top spot -  because that brings more sales, traffic and profits and gives you market dominance in key phrases. The auction is a “second price” auction – this is how Google works. So it’s bid * quality score – so basically you pay what the second place auction bidder has bid. Because of this, there are options about how you can do things at the top of the page that you cannot do lower down.

Matching…

If you can’t make a profit when you know exactly what the searcher typed,  then broad match is not going to work. Look at phrase match to get past the issues with broad match. One issue is that if searches are very low volume, Google does not develop a quality score, therefore your ads may just be turned off in exact match long-tail.

In ad testing, you can use multiple copies of the same ad. If you want a 90% / 10% test, then make 9 ads with control and 1 with the test. Great, simple way to test at a lower rate!

MSN Adcenter labs “OCI Tool” gives reading on commercial intent based on specific phrases. This can be valuable in identifying the right phrases to target for conversion.

Ryan Lash – ymarketing, LLC

Scenario: paid search campaign is not performing as well as it used to. There are only a few options:

1.  Conversion optimizer
2.  Implement a third party management solution (but there are challenges here)
3. Optimize it yourself – but there are times that this can be challenging.

How to build a semi-automatic bid management tool? Use a spreadsheet, web browser and adwords editor.  Here’s the process:

1. Get search reports
2. open spread sheet
3. do some math
4. analyze data
5. make changes

Calculating ideal CPCs for CPA campaign, look at existing cost by keyword. The ideal cost is conversions times target cost per conversion. The ideal CPC is ideal cost divided by clicks. The CPC sweet spot is closer to effective half net margin, but is based on actual performance.  Next set your statistical confidence. Set up the number of clicks you need to get before you make a decision. So, calculation is based on all clicks except for branded terms (because that skews results) – if conversion is 2%, take action after 50 clicks, 1% after 100 clicks.

Secret sauce: Average position, cpc, conversion volume. Don’t increase bids if position is 1, or average cpc is more than the max cpc you set for the campaign. Do not decrease cpc for top performers.

Next look at your keyword reports for the past 7, 14 or 30 days – include conversion data. Next, open Google Docs. Add in special headers for target costs. Use the various maximum values you have established to set the appropriate CPCs. However, look at the data to assure that you use the ideal CPC as a guideline so you don’t lose visibility on good terms.

In the presentation, he steps through how to take the data / bids from the spreadsheet and loading it up to Adwords through the editor. YOu can get the spreadsheet and background from their site – SearchGuyver presentation. Ryan then reviewed results of using this homegrown bid management method.

Dan Soha – Five Mills

First trick: Use keyword insertion, set the keyword insertion format to bring attention to the ad.

Quality index is at the adgroup level. Therefore, build highly targeted adgroups. They should be very similar to each other.

In Yahoo, don’t know what the broad match is, use their keyword suggester to pull similar phrases. Put it in the general area, add it to the “what must be in each phrase”, then it generates all associated keywords, then it will bring all the new keywords – and only NEW keywords, so that it will not duplicate as you do iterative searches.

Divide and Conquer:

• If the adgroups are well targeted, but the quality index isn’t good, then divide up the Adgroups. First look at CTR by keywords and note their relative position.
• Pause all low click through keywords. Then duplicate the adgroup.
• Next, delete the active keywords, unpause the paused keywords – then all you have there is a low click through keywords – separate from high click through keywords.
• The lower CTR adgroup will have a lower quality score.

Yahoo dynamic keyword insertion increases quality index. In fact, in direct comparison of the exactly same ad with the alternate text being the same as the static ad, the one with multiple keyword insertion elements will always score better in quality score.

Ad competition: Let Google do the work for you.

Step 1: make certain bids are targeted to current success metrics

Step 2: Duplicate the adgroup running 1 ad in each

Step 3: Let ads run until they have ample conversion history

Step 4: Rebid to current metrics of success

Step 5: Wait and repeat step 4

This means that the better click through and quality score will bring down the costs for the better performing adgroup, and you can eliminate the one that is not working as well.

Non-rules based bid management tools can cost their advertisers because they make big changes quickly. This means if someone is managing manually, and all else is equal, if the bid is pushed up in position one, then they may lower the bids and dropping the position down the page.

Addie Conner – Course Advisor, Inc.

Yahoo min CPC was put in April 2008, and prevents some keywords from showing in the auction at all. Has hit a large number of keywords within some accounts.  There are instances where 30-50% of the keywords are affected. Quality score in Yahoo mainly lives on the ad/adgroup level. There also appears to be some kind of account level score. First, ask Yahoo to create a new account, unlinking your current one. Unlike google, they do not port history. This will get rid of the problem

Gaining Competitive Insights – Google data is saved in tehir system and is ported from various accounts. This is important so you don’t lose history. This is also a good way to look at competitive metrics (these are theoretical practices and CAN be done….)

First, Google a keyword that you are interested in. Capture the ad, then click on it to get the destination URL. (Watch to strip out any dynamic parameters from the URL). Upload to a new Google account the competitor’s ad, keyword and copy. Then observe…. Look at quality score and min CPC. You can see what is happening with competition. This can also be run on your own keywords to see whether you have some campaign level quality score issues. Based on the data you get, you can build a bid to click and position to click model for your competitor.

Since others do this, look at other clicks that might be coming from someone else running this type of test. Look for ads you don’t write – there may be instances where affiliates or others who are running against your ads, keywords, or target URL. Keep an eye open for changes in quality score – this can also pose a challenge.

Dominating the page – There are instances where people are advertising multiple ads to the same landing page. Sometimes people run the same ad to different URLs. This is legal – because parent companies may operate multiple sites and the engines may see that different companies may run specific user experience for different ads – so they are OK with this process. This can be used to sandwich a competitor’s ad with identical text, which has been known to kill the value of the sandwiched ad.

Launch with a high Quality Score – First, given keywords whose overall CTR is about the same, does it matter whether initial CTR is good to give a good quality score? Yes – high initial click through rate keywords end up with a better quality score. In the test presented, this was true by about 2.5 points in the quality score.

There are instances where a position ends up causing bad quality scores – especially if it is in a position that doesn’t have adequate data for Google to calculate the best score.

Tested different groupings, by semantic, quality score and random keyword groups. Semantic keyword groupings were most effective at bringing costs down and attaining a higher quality score for the adgroup.

• Start with an inflated bid
• Run creative that is a bit more aggressive than you would normally run
• Start with high quality / traffic keywords
• Once keywords have a good quality score, then add other keywords.

Take Home Messages

• Use broad match campaigns as a farm team to find winning keywords, then change winners to phrase and exact match. Winners should go into their own campaigns to better manage budget for those keyword sets.
• Bid more highly on winning keywords in exact match to get best positioning on those phrases

Tip: Use negative exact match phrase when it is a very competitive term. So bid on broad match for the phrase, add a negative exact match for the broad term – then identify winning related terms around it.

Tip: Set up campaigns by match type. This helps control budget. The challenge is that if the campaigns are not properly funded, the exact match won’t pace properly, or will run out of money. Also be careful not to set up fragmented campaigns. Also, campaign level gives you impression share metric.

Here is summary from each presenter:

Craig Danuloff of ClickEquations – How to Hunt

Match-type is query hunting. As both an agency and software provider, they see a bunch of different choices of match type use. We are way too keyword obsessed in the paid search world. Keywords and bids are the shiny objects that they want us to pay attention to, but the search engines choose which queries to serve our ads on. We should decide which queries to be shown for – not have it done “randomly” by the search engines.

In broad match, there are many types of searches that might have different economic values. But if you pay for them all and send them all to the same landing page, we end up managing our campaigns by “averages” – where some are great performers and really appropriate, and others are very inappropriate and won’t convert.

Craig continues to compare match types to different weapons:

• Exact match is a rifle – the exact words
• phrase match – shotgun – get mostly the people we want
• broad match – bomb – get a bunch of people, but many we don’t want
• automatic – black hole. The collateral damage is immeasurable

If you just use broad match, and don’t decide which queries where your ad is shown, then you don’t decide who will see your ads, the engines will. So, competitors can grab the queries through match type, and you get left out of good converting terms. So – in ads being shown: exact match beats phrase match which beats broad match. Therefore, will lose positioning to a competitor who might use a more tightly controlled method of selecting phrases.

Goal:

• Control search queries. Target the good ones, and filter out the bad. Select and prioritize those we really want to be seen on.
• Value search queries as appropriate

• Filter out the loser

The Keyword Trap

IF you buy keywords in all three match types, then you can (theoretically) get the best likelihood to be seen on the really important terms, then others that are less important, you get the broader keywords. When bidding on all three, you can bid different levels on each match type, so you get higher positioning on the really good phrases and save money on the longer tail ones.

Broad match keywords are the farm team – they are the ones that help you find out which keywords are winners, and you promote those to phrase or to exact match.

10% conversion in exact match – $1 CPC, then $10 per CPA

5% in phrase – $0.50 CPC, then $10 CPA

1% in broad – $0.10 bid, then $10 CPA

Once the construct is built through “farm team” work in broad match, this is the way you would proceed.

Separate match types by different AdGroups – this is really important for reporting and then move good keywords to exact and phrase match Adgroups for better bidding control.

Brian Kaminski – iProspect – Match Types in Action B2B Case Study

There are a lot of complex issues in paid search that lend a lot of complexity to managing campaigns. In short, match types can help simplify managing campaigns.

Company is a multi-billion dollar client, driving a lot of clicks. However, conversion and CPA goals, negative brand image, bounce rates were bad. Broad match was 98% of clicks.

The three key areas that the campaign was failing at was that the ads were showing in inappropriate queries, there was irrelevant traffic and strong product overlap lead to user confusion. People were arriving at the wrong product when they searched.

Step one was to focus the campaign. In this situation, match types was the best way to simplify this campaign.

1. 1. Added mutliple match types on brand
2. Hundreds of negatives to clear up confusion
3. Added thousands of terms
4. BUilt out smaller ad groups
5. Used ad copy to get the right people
6. Refined AdGroups

This brought CTR up, CPC down, CPA fell and overall coonversion volume increased. THere was a tighter relationship between query and landing pages. This better experience did have some positive effect on the negative brand impact because the user experience improved.

Keys to Make PPC Match Types Work

1. Evaluate match types based on budgets. Sometimes it’s important to fund exact and phrase match campaigns more highly, so that you don’t lose visibility on these valuable terms.
2. There are different best practices for different engines. Modify campaigns differently by each engine.
3. Branded terms should also be set up as broad match to assure good visiblity for longer tail.
4. If the product lifecycle or overlap is a challenge, it’s important to set up match types for the best converting terms and have landing page experience match the query.
5. TEST.
6. Look for opportunities and openings in competitive landscape. Are there opportunities to get better positioning there?
7. Use matchtypes to simplify your life, not make it more complex
8. Match types are not a set it and forget it approach – leverage the data going forward.
9. Do NOT just go halfway – complete the expereince and test with specific ads and landing pages based on match type.

Start out with clearing up match types, then work on ad copy, then landing page experience. It’s important to get the winning keywords and then improve campaign conversions from there.

Seth Barnes – Edmunds.com – Observe and Report – In-house Strategies for Traching and Tweaking Keyword Match Types

Since Edmunds provides information on all different types of vehicles, they can play in a very broad marketplace for keywords. To do this correctly, though, it’s important to have the measuring tools in place and look at them all the time to assure that you are on the right track.

If you don’t play with multiple terms, you risk losing market share.

Don’t be over-zealous with match type – Deploy, Observe, Report and test, test, test.

Within the vertical, if you ar ebig player, until you have deployed your campaigns, you cannot know what is going to happen. For instance, there are issues with the economy, etc. It is not possible to just listen to recommendations from others – it’s important to get into the campaigns and try things.

Seth is a big advocate of automation. However, it takes a lot of effort up front to build out the method by which you manage the campaigns. Setting up a template and methodology to deploy on new product launches, so taht it simplifies getting ads out quickly. Rolling out organized campaigns makes life easier.

Start out – prune back on broad match to keep your campaigns sharp.  Look at AdWords data or inhouse data to better understand what is going on. Engines expand and contract match tyueps for quality, coverage, and revenue purposes. It is important to look for click spikes, low ROI keywords. Use the search engine reps to your advantage. THey can help navigate these changes.l

Recent developments includes a 20% increase in daily queries for the first time in 90 days. Yahoo is setting up market reserve pricing, and MSN has launched Bing with will increase relevancy of search, so may not show search ads as frequently, but there is a lot of buzz to happen with MSN.

Yahoo’s market reserve means that if you lower your bids too far on exact match, they will then switch to broad match or they will go dark completely.

Here’s what to look at to implement good campaign assessments:

• Reporting and Analysis
• Rev iew log files and query tracking
• Build an ongoing negative keyword lists. Cross pollinate negatives across engines
• Build a dashboard
• Look at search query reports to understand what keywords are being shown for a query
• Clean up the campaign structure to allow good reporting and develop an action plan for better fine-tuning

Last, Seth shares a case study about how they handled issues about buying keywords for Chrysler during the bankruptcy issues. They researched what phrases were being used, and negative matched “bankruptcy” assocaited keywords, but left those phrases that in place that had purchase intent.

Google recently announced a change in their policy about using trademarks in AdWord campaign text ads. It’s an interesting change – and will be valuable for companies that sell parts, accessories, for franchisees and for affiliates. However, it puts brand holders in a potentially precarious position.

Here is the pertinent information I pulled from Google’s policy statement:

As a provider of space for advertisements, please note that Google is not in a position to arbitrate trademark disputes between the advertisers and trademark owners. As stated in our Terms and Conditions, the advertisers themselves are responsible for the keywords and ad content that they choose to use. Accordingly, we encourage trademark owners to resolve their disputes directly with the advertisers, particularly because the advertisers may have similar advertisements on other sites.

As a courtesy to trademark owners, however, we are willing to perform a limited investigation of reasonable complaints about use of trademarks in ads. In the US, our policy permits use of the trademark in the ad text in the following circumstances:

•  
  • Resale of the trademarked goods or services: The advertiser’s site must sell (or clearly facilitate the sale of) the goods or services corresponding to a trademark term. The landing page of the ad must clearly demonstrate that a user is able to purchase the goods or services corresponding to a trademark from the advertiser.
  • Sale of components, replacement parts or compatible products corresponding to a trademark: The advertiser’s site must sell (or clearly facilitate the sale of) the components, replacement parts or compatible products relating to the goods or services of the trademark. The advertiser’s landing page must clearly demonstrate that a user is able to purchase the components, parts or compatible products corresponding to the trademark term from the advertiser.
  • Informational sites: The primary purpose of the advertiser’s site must be to provide non-competitive and informative details about the goods or services corresponding to the trademark term. Additionally, the advertiser may not sell or facilitate the sale of the goods or services of a competitor of the trademark owner.

 

How I read this policy change is that they are allowing companies that offer compatible products to use brand terms, so “discount Toyota parts” or “iPod accessories” can now be used. Also, affiliates can use brand terms, so “Best deals on XYZ product.” I can see a lot of affiliates that base their model on “coupons” will use the heck out of this approach.

Here’s where it can get really hinky for brand owners:

Affiliate ad showing in a brand placement – since it has an offer, discount statement AND the brand phrase, a searcher is more likely to click on that ad rather than the brand holder’s ad.

Google requires that an affiliate have a landing page or something equivalent – which gives the affiliate carte blanche with user experience. If they are good at closing sales, then the searcher will buy from them. The brand owner sees the effectiveness of their paid campaigns diminish, they are paying out more in affiliate fees, and (to add insult to injury) the affiliate starts getting lower AdWord fees because they are getting the clicks – so the brand owner loses positioning on their own phrases.

I do not wish any ill-will on affiliates - they are a great source of revenue for any company. However, they start affecting brand-based search campaigns, which are usually where brand owners close the sales that they have fostered through more expensive, broader phrases.

Additionally, there are ways for companies to get around the requirement regarding not advertising with competitors’ brand phrases in the ads – especially with the “informational site” clause. However, I think this is downright unethical.

It will be interesting to see what happens in the longer run with the change in this policy – and whether we will see an update soon from Google about how this is handled.

Quick addendum – Did you know that Google does not pursue trademark violations when they appear in the visible URL of the ad? Keep your eyes open because your competitors CAN get around Google’s trademark policy by using your product / brand name in their visible URL. You’ll have to reach out directly to the competitor about that one.