Google AdGroups G8 - Post Campaign

Executive Summary
Campaign Overview
Through this project, we aimed to increase Snow Factory’s, homemade yogurt producer, brand awareness result more accurate sales leads. Our goal was to spend 1800NT on keyword advertising for two weeks starting from 16th of December 2008. Considering each keyword to cost 2.5NT, we planned to capture 720 clicks, and 36000 impressions (assuming average CTR=2%) by utilizing the winter holiday themes and the company’s own brand name Snow. We designed 5 different adgroups with corresponding keywords.
Key Results
At the end of the two week campaign our results were:
We did reach our target impression rate, but the CTR was too. Due to numerous trials and errors, we figured out how to place an ad on higher position. Our ads on averaged appeared at 2.5th position. Of the 5 ad groups, Ox Year – New Year related ad group collected highest number of eyeballs. Opening Biz, a Chinese ad emphasizing entrepreneurship had the highest clicks.
Conclusion
Although we had relatively large number of impressions, clicks were so low. This meant that the keywords we chose were effective ones, but not necessarily efficient ones. First of all, our average CPC was way above our original estimation. Second, low clicks, high impression usually translates to poor ad copy. We had 5 different ad groups with 5 different ad copies. Designing attractive text ads indeed was a challenge. Relevance between keywords, ad copy, and content on the website defined quality score – position – CPC. The perfect match of all three seems to be near impossible.
Future Marketing Recommendation
1. 36566 impressions in a mere 14 days are not too few. To retain 2% CTR, Snow Factory should work on the content of the websites. No more scanned materials. Google doesn’t understand what’s on photo files. This determines relevancy of information. The more relevant, the higher the position, and quality score, the lower the CPC.
2. Snow factory should add money transfer function on their website. It is almost impossible to track conversion rate, or the ROI. People go to their website to find a phone number. Since phone calls aren’t made on the browser, it is impossible to track potential sales leads. Google Analytics can only be placed on web pages. Links that automatically lead to Windows Outlook mail service, or .xls file download hyperlinks are impossible to be connected with tracking tools like Google Analytics.
3. If possible, the company should consider applying for their own domain name. Links that appear on the bottom of ad copies should be related to the main website of the company. Webpage on another portal site is difficult to the controlled and modified.
4. If Snow Factory decides to run their own keyword advertising campaign, they should consider themes that are directly related to the content on their website. If running English ads, then the website better have English version, or English contents. Seasonal holidays are proven to be useful when promoting products that can be both consumed and given to someone as a gift. Budget should be allocated more during weekdays than weekends.

Industry Component
Introduction
Our client, Snow Factory, promotes its homemade yogurt on 3 sites: bid.yahoo.tw, snowfactory.pixnet.net/blog, and a webpage liyihui.myweb.hinet.net/newstore2. On/offline combined monthly sales revenue of the company averages around 400 K NT$. Considering the state of the company’s scattered information over several websites, we proposed keyword advertising to show its effectiveness beyond W.O.M generated by local personal blogs, which is how the company has been pulling customers.
Campaign overview
Our goal was to increase Snow Factory’s sales and brand awareness by identifying critical keywords for the company and couple them with attractive text ads. The two week campaign started on 16th of December 2008 under Google Adwords account ID: ntumarketing, and ended on 30th of December. For our campaign named: Snow Factory, we designed 5 ad groups (General (CH), Opening Biz (CH), Winter Present (CH), NTU MBA Program, Ox Year). Considering the fact that an overnight effect is quite impossible, we monitored the campaign once every two days to make necessary adjustments to the keywords or the text bodies based on quality scores, clicks, CTRs and cost. We used keyword recommendations relating to our campaign and those with high quality scores, and sometimes our own imagination to come up with new keywords. CPC per keywords we selected cost around 2NT at the time of pre-campaign development (health, beauty related keywords were 15NT). Daily budget was planned to be allocated as 2/3 of the weekly budget on weekdays, 1/3 on weekends. Assuming average CPC as 2,5NTD, our projected clicks were 800 (720) clicks. Return on investment was to be recorded by Google Analytics, but the transactions method of the website didn’t allow us to keep track of the actual sales leads.
Evolution of the campaign strategy
1. Until 23rd of December, we ran 3 Chinese ads (General (CH), Opening Biz (CH), Winter Present (CH)). On 23rd, end of the first week, our cumulative statics were 44 clicks, 14,353 impressions. We believed that additional ads about the coming winter holidays in English would attract more locals and foreigners. Also, 0,3% CTR at this point was too low, hence more attractive texts needed to be developed. Considering the background of the company’s managers, also increasing number of MBA enrollment, we came up with 2 more English ads (Ox Year, NTU MBA). Although NTU MBA didn’t create any CTR, it added 1200 more impressions. Ox Year turned out to be one of the higher click adgroups.
2. At first, we couldn’t divide our ads into separate groups with separate pools of keywords. It linked certain keywords with not so specifically related ads, leading to a problem of incorrect identification of keywords’ efficiency. Soon we learned how to separate ad texts to specific keywords and separated our text bodies to different adgroups. Doing so not only increased impression rate and added a few clicks, but enabled us to increase quality score and stabilize our ad position.
3. Originally, budget allocation for Max keyword CPC was set to Google’s default CPC bidder of which it averaged about CPC=35NT. On 23rd, we changed the default setting to manual bidding. The Max CPC was lowered to 25NT considering the high expense on few clicks. We also experimented with content network by allocating 10NT for content network CPC, but soon realized that not many impressions originated from content network. With better quality score and lower Max CPC, we accumulated about the same amount of clicks at cheaper CPC and total cost.
4. For the first week, we used the system’s default setting to optimize our campaign. After the 1st week, we had only 712.32 NT (1800-1087.68) left in our account. Hence, we lowered our daily budget limit to 128 NT (1800NT/14days).
Campaign Result
Date Days Impressions Clicks CTR CPC Cost Position
16/12/2008 Tuesday 82 2 2.44% 17.25 34.5 1.6
17/12/2008 Wednesday 3614 12 0.33% 12.92 154.99 1.9
18/12/2008 Thursday 2045 7 0.34% 26.33 184.34 2.3
19/12/2008 Friday 2268 5 0.22% 26.82 134.08 2.8
20/12/2008 Saturday 1525 4 0.26% 31.87 127.46 3
21/12/2008 Sunday 2169 5 0.23% 30.96 154.81 3.4
22/12/2008 Monday 1270 6 0.47% 31.23 187.37 2.9
23/12/2008 Tuesday 1469 3 0.20% 36.71 110.13 2.6
24/12/2008 Wednesday 4261 6 0.14% 16.14 96.81 2.7
25/12/2008 Thursday 4244 8 0.19% 12.64 101.1 2.6
26/12/2008 Friday 1474 10 0.68% 10.29 102.93 2.8
27/12/2008 Saturday 3404 7 0.21% 14.34 100.35 2.3
28/12/2008 Sunday 2203 10 0.45% 10 100.04 2.5
29/12/2008 Monday 3309 7 0.21% 14.24 99.66 2.4
30/12/2008 Tuesday 3229 4 0.12% 17.62 70.49 1.8
Totals/Avg 36566 96 0.26% 18.32 1759.06 2.5

Result of our 2 week campaign is as above. There is couple of points worth mentioning here.
First of all, the average number of impressions of the whole campaign seems to be increasing along time. One might argue that the seasonal effect is the cause of such shift. Indeed, impressions increased during Christmas and before New Year.
Second, the number of clicks seems to follow a pattern. It increased in the middle of weeks (Wednesdays) and decreased over weekends. We believe that people click on advertisements during weekdays, when they are in office, behind computer desks. Therefore, it might be wiser idea to allocate more budgets during week days.
Third, starting from the second week, average CPC decreased dramatically. This coincided with the separation of adgroups and keywords. Our careful selection of keywords based on quality scores, more detailed identification of specific keywords facilitated by grouping increased our relevancy rate; hence average position of 2.5, lower overall cost structure.
Starting from the 2nd week (23rd of December), we set Max CPC on our own (not following Google’s default) to 25 NT, restricted daily budget to Max of 120NT, all of which resulted higher rate of clicks at lower CPC.
Adgroups and Keywords analysis


The combined result of our 5 adgroups is shown above. According to the data, Ox Year generated the most impressions, as we chose more general keywords. With better relevancy and higher quality score, google charged us less CPC on Ox Year than any other adgroups. NTU MBA was deleted after 3 days as it was not creating a single click.
Each adgroup with its corresponding keywords is summarized as follows:

1. Our General (CH) adgroup used the most keywords of which our keywords were anything related to Snow Factory’s yogurt as well as some related to the holiday season. The ad copy of the General adgroup included information such as: 100% natural, ½ calories compared to conventional yogurt, helps lose weight, and maintain beauty. Among the top click through rated adgroups, General (CH)’s top performing keywords were: 優格10 clicks, 578 impressions and a quality score (QS) of 7/10. On 18th we also added “food gifts” 食物禮物, which Google recommended to be relevant to our campaign and had a high predicted QS. We deleted Christmas Present 聖誕禮物 because it only got 2 clicks, 370 impressions at a high cost of $44.71. We thought some keywords like Taiwan yogurt, Taipei yogurt, Taiwan desserts, and Asian yogurt would generate more clicks, but to our surprise they did not. On 26th with only 300NT left, we used Google’s keyword tool to find some more relevant keywords. We added words like: natural, fruit jam, fruit yogurt, strawberry yogurt, blueberry yogurt, peach yogurt and grape yogurt. We also deleted “natural” 自然 due to very low performance.
2. For Opening business (CH) our keywords were related to creating a new business, entrepreneurship. The result of the adgroup in detail is as follows:

Ad copy of this ad was: Entrepreneurship dream of 3 Graduate students in NTU MBA, who created high quality yogurt (translation from Chinese). We found out that different links put on the bottom section of an ad creates different results. We created two identical ads with different URLs (SnowFactory.pixnet.net/blog and liyihui.myweb.hinet.net). From the two graphs, it is evident that both impressions and clicks of the ad with liyihui.myweb.hinet.net are much higher than the other one. Our original rationale was that showing the company’s name on the URL will result more clicks, but it all occurred conversely. After the first week, looking at the campaign result, we decided to take off the 1st version of Opening Business adgroup. The impression rate went down along with clicks after the first week. We figured that there was a seasonal effect (holidays were approaching). This was the most expensive adgroup we had in our campaign. The better performing keyword was: 創業 20 clicks, 2569 impressions, no given QS. During second week, we added words related with world economy to pull more customers’ attention. We added words like: job opportunity, creativity, salary, income, business opportunity, entrepreneurship, slow economy… Although this ad group created the highest CTR, but the irrelevancy of the keywords and the site content, decreased our quality score, hence increased CPC, making this ad the most expensive one.

3. On 24th after which we separated keywords by ad groups the adgroup only had 1 click. Our maximum CPC for this ad was set at 25.00/day. 3 keywords we used throughout the whole campaign for this adgroup had QS’s 4 (甜點 sweet snacks), 5 (雪 snow) and 7 (冬季禮物 winter gift). Snow was the only keyword that actually got clicks, 7. We thought winter gift would create clicks due to the time of the year, unfortunately this wasn’t the case. On 18th, we deleted Christmas Present 聖誕禮物 because it only got 2 clicks, 370 impressions at a high cost, bid 44.71NT.

4. From the 2nd week, we added 2 more English ads to attract more foreigners. Ox Year’s keywords were related to the Chinese New Year. It read: Start the Ox year in great shape; BUY Taiwan’s only healthy yogurt. Impressively Ox Year created the most impressions of our campaign. The key words used for this adgroup were: 2009, Chinese New Year, fitness, healthy food, homemade, natural food, New Year’s, New Year’s resolution, Ox Year and yogurt. The keywords that performed the best were: “fitness”, 11 clicks, 2319 impressions and QS = 3. Next was “2009” of 10 clicks, 7653 impressions and QS of 2. We added the negative word “artificial flavor” which was recommended by Google. We added the keywords Christmas dessert and food gifts on 28th because of the holiday season. From the start, this ad did quite well and had an increase in impressions and clicks except for one sharp drop on 26th which could have been caused by Christmas ending. Although this adgroup was created to relate more to the Chinese New Year, adding a few Christmas holiday keywords helped.

5. NTU MBA’s keywords had to do mostly with university education and career. The ad emphasized MBA alumni success BIZ, Launching one’s career immediately. What’s more, NTU’s college of management was placed at number 43 on the world arena; hence pulling customer in this way seemed to be feasible. Unfortunately it didn’t create any click. Therefore, we deleted it on 27th.
Key Results Summary
The changes we made that we found helped the most were deleting keywords that weren’t producing any results, adding more specific keywords and using simple rather than complex keywords. We also found combining words like Christmas and present, holiday and gift and Taiwan and yogurt didn’t produce the results we were expecting. Single words seemed to have better performance. We actually used more holiday related keywords under our General adgroup. We should have instead put more holiday keywords under the Winter Present adgroup. The adgroup with the highest impression was Ox Year. We definitely could have done without the NTU MBA program adgroup.
Learning Component
Learning Process
Our learning objectives were to master the basic techniques of Google AdWords platform. During the campaign, the largest problem we faced was during the first week, when we couldn’t divide our keywords into different ads. Later on we realized that the three text bodies we had created were placed under one adgroup as three versions of an ad. Hence, a keyword not so directly related to a particular text body still led to that text, creating a problem for identification of which combination (keyword X + text Y) led to what result. From 23rd (end of the first week) on, we divided the three text ads into three separate adgroups with corresponding keywords to each ad group.
Also, since our group was made of students of multi ethnic backgrounds, minor lingual barriers existed throughout the whole project. For example: One of our teammates added some keywords that were translated by Yahoo but supposedly Taiwanese don’t search using these words so they didn’t produce any useful results and they were deleted a few days later. Those keywords were: Taiwan yogurt 台湾酸奶, Taipei Yogurt 台北酸奶, Taiwan desserts 台湾点心 and Asian yogurt 亞洲酸奶. Our communication with the client went quite smoothly. They replied to our requests immediately, though, we couldn’t effectively measure conversion rate due to the original design and operation of the website.
Things we learned
-Relevant keywords are not 2.5 NTD, much expensive
-People still refer to Google as more academic search engine. Taiwanese don’t use Google for advertisement purpose. -Content network didn’t generate anything
-Google doesn’t allow ads differing from the site’s contents – ads are a way of pulling customers through any means, but Google restricts advertisers’ creativity
-Compared to our click through rate, our impressions were not few, but the actual CTRs were so low. It meant that our choice of keywords were right, but was not attracting customers. We tried different versions, but still were not good enough. Designing attractive text adds is difficult. Or maybe perhaps, people just ignored ads.
-The system requires some learning curve. The interface is too crowded.
-It took a while to apply changes to our existing ads. We couldn’t see the result immediately. Also, we had to log out of the system, and use different IP address to see the result on the actual Google site.
-Google Analytics ”html” is only allowed to patch to a site, but not to a link (e-mail, or download link). The site we chose let customers download order forms on their site. One should download the order (xl) file, fill it and send it to the company’s site. Most obvious way of counting conversion rate is to connect the tool to that hyper link.
Or people can click on a hyper link that diverts surfers to directly write e-mail to the company’s site. If the tool could be attached to that hyper link, it would have been another way of counting ROI.
-Most people in Taiwan directly call the company rather than send mails to them. The phone numbers are on the site. Google’s analytics should consider this point.
Recommendation for the company:
-They should include many topics on the site. Words on picture files (scans of newspapers) are not recognized by Google. When one writes a text ad, Google searches the site to see the relevancy of the words. If the words appear on picture files, Google basically says the words are irrelevant.
-They definitely have to consider adding internet payment method to their site. It will enable keywords advertising ROI measurement a lot accurate.
-If possible, the company should consider buying their own dotcom site. We found out that putting Google Analytics tools on portal hosted sites are almost impossible. First the company has to ask the site operator, and only if the operator’s company rules allow users to put bugs, can he plant the device.

About Google Post Campaign

First of all, I would like to thank Andrea and Deniz for providing me such detailed information about how our campaign went. I have all the necessary files for the report.
Second, due to simultaneous projects and reports I am preparing for this finals, I still have not finished the last version, or the combined version of the report. Please don't worry, I will finish it before dawn, possibly earlier than that.
Since, it is impossible to make slide show without the final version, I'll have to do that part too.
I'll be ready to present for tomorrow morning according to the slides, but should anyone wish to present certain parts of the report, please feel free to do so.
I am just trying to pull my own weight, not trying to impress anyone or anything.

Regards

Amraa