![\"Dan](\"\/wp-content\/uploads\/2017\/07\/dan-thies-ppc-strategies.jpeg\")
You\u2019ll easily find Dan Thies in Google search results detailing his credentials in the PPC marketing world. For his workshop at Rocks Digital 2017, Dan focused on Google Adwords. Please take note,\u00a0many of these\u00a0same PPC strategies can be used on Bing.<\/p>\n
Here\u2019s\u00a0the focus of today\u2019s post:<\/p>\n
Dan says: This is not a beginner show, as it\u2019s AKA \u201cAround the world in 90 minutes,\u201d so let\u2019s go!<\/p>\n
Well, users\u00a0put words in the box and search, that\u2019s obvious. Then Google matches those words to bids. But in reality, Google Adwords works somewhat weirder than you might think, particularly for negative keywords. And you have to be exactly right, or Google will not be nice about that.<\/p>\n
There is an auction for Google Adwords. Winners are picked via a live Vickrey auction. Note: They are not simply auctioning off clicks \u2013 your bid per click is being translated into something else, and this is way more important than you think!<\/p>\n
In the search results, ads are shown and the searcher responds. Here are the match types:<\/p>\n
The Adwords auction is for impressions and not clicks. Every impression is auctioned off live. You pay for clicks, but you get impressions (they used to run this off stored data).\u00a0There\u2019s a downside to this. When you see \u201clow search volume\u201d keywords, that means there\u2019s not enough data to run the auction. And it especially affects local stuff!<\/p>\n
Auctioning<\/span><\/p>\n Every advertiser’s CPC bid gets translated into an estimated CPM bid. This is a forecast of how much Google is likely to make if they show your ad at the top spot. You must win the bid in order to win that top spot. But wait, it gets weirder after that.<\/p>\n Estimated \/ Effective CPM<\/span><\/p>\n Bid x Quality Score \u2013 but it\u2019s not the score that relates to your actual landing page, but the one for the ad itself. To simplify this, think \u201cBid x CTR\u201d where the maximum CPC bid is set by you.<\/p>\n The Click Through Rate (CTR) is measured or predicted by keyword. Here\u2019s an example: $1.00 x 10% CTR x 1000 = $100.00.<\/p>\n Every advertiser’s bid is different, and every advertiser\u2019s (ad) CTR is different.<\/p>\n The Second Price Vickrey Auction is designed to get every bidder to bid their full maximum value at all times. The winner pays the loser’s price. See these example bids:<\/p>\n #1 wins, and pays equivalent to $16.00 eCPM; #2 pays equivalent to $12.00 eCPM. Translating that into your actual cost per click is the same math in reverse.<\/p>\n $1 CPC x 10% CTR = $100<\/p>\n $1 CPC x 20 % CTR = $200<\/p>\n So, your eCPM Bid is doubled.<\/p>\n Using the data in #1, #2, and #3 above, at the #1 position doubling your CTR will result in twice as many clicks at half the cost per click, because you are bidding for impressions, not clicks. Your total cost doesn’t change. This is almost a fact \u2013 increases in CTR are free. You get more traffic at the same cost.<\/p>\n Keyword Strategy and Budgeting<\/span><\/p>\n Let\u2019s look at the Google Adwords account structure. A campaign contains ad groups that control geography, ad schedule, bid strategy, distribution and so on… just a bunch of stuff. Today it controls more than it used to.<\/p>\n Keywords \u2013 Don’t Use More Than You Need <\/span><\/p>\n Dan\u2019s keyword strategy starts with exact match keywords if you can, but this is not always possible. Local businesses and niche businesses want to use long tail keyword targeting. Whatever you do, be as precise as you can.<\/p>\n Look for auctions with WEAK ads. Look for what you like \u2013\u00a0and what you don’t like \u2013\u00a0about these ads, figure out what you can improve or beat, and then work to knock them out of the running. For instance, you wouldn’t charge someone for a mortgage quote<\/em>, but if people normally did charge for that, you could stand out by making yours for free.<\/p>\n A further example: The word nursing means more than one thing, so \u201cnursing apparel\u201d can get you various types of results depending on whether you mean the medical profession, or nursing a baby.<\/p>\n With Adwords, you can take the re-marketing words, and target the people that have been to your website. In the case of Dan selling gluten-free grape soda, the CTR went up because he let people know the grape soda was gluten-free!<\/p>\n Single Keyword Ad (SKA) Groups<\/span><\/p>\n Dan has been teaching this since 2003, but now it\u2019s \u201ca new thing.\u201d Every keyword in an ad group is going to use the same ads. If you test two ads against 100 keywords, every test will hurt some and help others.<\/p>\n SKAs are essential at high volume, but semi-useless at very low volume. For Black Friday, we will split out the campaigns due to the volume.<\/p>\n Pick Some Priorities and Go<\/span><\/p>\n If your budget doesn’t cover 100 clicks a month at the market price, well, that’s not enough budget. (Dan is not saying don’t run it<\/em>, but that\u2019s really not enough budget to do a satisfactory campaign.)<\/p>\n The more clicks you can buy, the faster you can complete ad tests. And the faster you complete ad tests, the faster you get to the top spot with the lowest CPC.<\/p>\n You can’t test faster than 7 days. Since there are some ads that run better on some days than other days, you need to compare an ad running on all 7 days of the week. Plus, some ads are good to run for 2 days, others better to run for 4, and so on.<\/p>\n Example: The same ad on one day IS NOT NECESSARILY the same ad on another day. For instance, the ad on Wednesday may not be the top ad to run on a Friday!<\/p>\n It is best to pay the maximum you can to break even because you will end up not paying that much.<\/p>\n NBA tickets is a VERY competitive auction and the competition really knows their stuff. Sometimes you can’t beat the top spot, so you have to see what you can do with the next spots. Dan made a distinction \u2013 for example, instead of people picking up tickets at will call, the person could print them out online.<\/p>\n It\u2019s best to start a client on shopping ads because that\u2019s a very easy place to start and be successful.<\/p>\n Figure out the conversion rate from a product page. Do shopping first, because the results will tell you if you want to run a search ad. Know the results of conversions if someone goes to a shopping page. This will help you get the numbers you need to convince your clients on running the campaigns.<\/p>\n Start with Google’s bids, then do the math again with the actual CPCs. You can also start with half of what Google tells you and still be competitive.<\/p>\n PPC Ad Testing Tips<\/span><\/p>\n If you have a good control ad, make copies to give fewer impressions to new (test) variations.<\/p>\n Look at add to cart<\/em> and buy now<\/em>. On Monday, buy now<\/em> may lead greatly, but then at the end of the week, they became even.<\/p>\n What\u2019s the reason? Well, on Monday when anyone is buying something online they want to go ahead and get it done. By the weekend, the consumer has more time to literally take their time. But in Dan\u2019s ad tests, they also noticed that on the weekend, a number of items in the cart were higher.<\/p>\n Here\u2019s another example. In Seattle, if the weather forecast was 3 days clear, the bids shot up for a carpet cleaning business. However, the client wanted the phone to ring on the rainy days. Since Dan did not think he could do that, the client’s money was returned, and the campaigns were ended.<\/p>\n Google has so many bad defaults, you should always test with equal ad rotation for this very reason.<\/p>\n You can make as many copies of your control as you like. That way, you can run several control copy ads against your test ad.<\/p>\n Let’s Check Your Settings<\/span><\/p>\n What you can do with the organic listing is not the same as you can do with the paid listing. That\u2019s why you may want to still consider bidding on your brand name \u2013\u00a0even if your brand name is # 1 in the organic listings.<\/p>\n In the Adwords menu, there\u2019s an option for campaign type<\/em>. Turn on All Features (all the features and options available for search). Note that this is not a default and has to be chosen.<\/p>\n In the case of networks, if you choose network partners, make sure it\u2019s a conscious decision. That can work if you\u2019re trying to get leads for cheap.<\/p>\n Mobile Devices<\/span><\/p>\n They aren’t forcing you to bid for mobile traffic, they’re forcing you to bid on desktop\/tablet traffic! This could be a problem if your market is VERY likely to use a mobile phone to look up a keyword. For instance, no one pulls out their laptop to find a tow truck company when they really need one.<\/p>\n Locations\/Geographic Performance <\/span><\/p>\n Google is self-serving in their favor, which is bad for you, but you can run quick reports to learn where your users are.<\/p>\n Lie # 1: Location Settings<\/span><\/p>\n DO NOT CHOOSE\u00a0\u201cPeople in, searching for, or viewing pages about my targeted location\u201d \u2013 and this is a recommended setting with Google AdWords, so pay attention. Instead, choose \u201cPeople in my targeted location.\u201d<\/p>\n Bid Strategy<\/span><\/p>\n Settings to use: Focus on clicks (Dan will manually set his bids for clicks).<\/p>\n Settings to consider: Enable Enhanced CPC. Focus on conversions (Conversion Optimizer).<\/p>\n Delivery Method: Choose Accelerated<\/em>. Do not optimize delivery of ads, because you\u2019ll go through your spending budget more quickly.<\/p>\n Note: If you\u2019re running on a limited budget, you may have to use Standard delivery.<\/p>\n Lie # 2:\u00a0Ad Rotation<\/span><\/p>\n DO NOT USE “Optimize for clicks.” Choose\u00a0“Rotate Indefinitely.”<\/p>\n Truth be told, you should do some ad tests to know which one would be better, but rotate indefinitely is the best way to go.<\/p>\n Hour and Day Report (24×7 = 168)<\/p>\n Example: If conversions are 487, then 487\/168 would mean your average = 3 per hour. It\u2019s worth a periodic look once you get some numbers piled up.<\/p>\n Shopping Campaigns \u2013 How They Work<\/span><\/p>\n With Google shopping campaigns you have better control of the keywords than you do in search ads due to low search volume. Here\u2019s how shopping campaigns work in brief:<\/p>\n Once you get how this works, you will NEVER go back.<\/p>\n One Big Catch-All<\/span><\/p>\n Using one bid for EVERY product or service is not an optimal way to do things!<\/p>\n Shopping \u2013 Level 2 <\/span><\/p>\n This is working with ad groups and product groups. Splitting that one campaign up into a bunch of campaigns doesn’t help. And yes, some queries are more valuable than others.<\/p>\n Example: engraved, engravable, personalized, for sale \u2013 these terms would let you know the person is DEFINITELY looking to buy a watch. Also, people looking for a specific brand name is a buying trigger.<\/p>\n Shopping \u2013 Level 3 <\/span><\/p>\n Let\u2019s break this into high vs. medium priority:<\/p>\n High Priority<\/p>\n Medium Priority<\/p>\n What\u2019s the net effect here? The budget shifts to higher value search queries and more ROI. You can now use keyword lists to structure things in all sorts of ways.<\/p>\n Robots that do bids don’t shift their bidding structure, which means you can beat the robots with this strategy.<\/p>\n Take the High ROI, Leave the Junk<\/span><\/p>\n For \u201cBig Brand\u201d robot bidding systems, what you just did will LOWER their ROI and cause them to drop bids.<\/p>\n On Black Friday, Dan did really well against the big brands for snowboards, but it didn’t happen in the first few hours of the day. For any keywords that were specific Dan bid high enough to own the keyword, which caused the robots to lower their bids. Once that happened, then only Dan\u2019s client’s ads were at the top! After that happens you can change the bid, and it is all about controlling the bids<\/em>. You bid high to get rid of the competition and once they are gone, you can lower your bid.<\/p>\n\n
\n
\n
\n
\n
\n