Close
Join 237,000 weekly readers and receive practical marketing advice for FREE.
MarketingSherpa's Case Studies, New Research Data, How-tos, Interviews and Articles

Enter your email below to join thousands of marketers and get FREE weekly newsletters with practical Case Studies, research and training, as well as MarketingSherpa updates and promotions.

 

Please refer to our Privacy Policy and About Us page for contact details.

No thanks, take me to MarketingSherpa

First Name:
Last Name:
Email:
Text HTML
May 04, 2010
How To

Analyze PPC Data to Uncover Date-based Performance Trends: 6 Tactics

SUMMARY: The wealth of data available from paid search advertising contains important details about customer conversion behavior – such as specific keywords that perform better on certain days of the week, or weeks of the year. But it takes some effort to uncover these trends.

We spoke with three paid search experts and learned six tactics to help you analyze your PPC data and find performance fluctuations. Includes advice on customizing reports, how much data to gather, and how to use your time wisely.
Jeff Ferguson, Senior Director, Online Marketing, Local.com, and his team are among the biggest advertisers on Google and Yahoo!. Last year, they compiled about six months worth of response data to look for trends in their PPC ads’ performance.

Hidden within all that data were compelling patterns of searcher behavior -- particularly differences in how ads performed on different days of the month.

"We’ve did a lot of great, deep analysis that revealed when people would be looking for specific types of services," says Ferguson. "We saw some simple patterns, like the first and fifteenth of the month, which is usually payday for most people, but we noticed many more complex patterns, as well."

Using the information, the team was able to change its bidding strategy during the next quarter and "saw some stellar growth almost immediately."

This scenario is precisely why you should optimize your PPC campaigns for time of day, day of the week or month, and months of the year. But pulling together PPC timing performance data and deciding how and when to act on it is a tricky process.

We sat down with Ferguson and two other paid search experts to find out how to analyze your ads’ performance trends over different dates. Here are seven tactics they recommend:

Tactic #1. Build and analyze custom reports

Having a reliable analytics tool is absolutely necessary. Regardless of how advanced or expensive your tool is, you will likely need to play with data to view it in a way that reveals date-based trends.

Using free tools such as those provided by search networks "almost always requires pulling data into Excel and doing some additional manipulation in order to get things sorted in a way that’s actionable. It’s worthwhile though," says Mike Churchill, CTO, KeyRelevance.

- Identify goals based on key performance indicators

Before diving into your analysis, determine the outcome you want to improve through ad timing. It could be:
o Increasing site traffic
o Increasing conversions
o Decreasing cost-per-acquisition
o Increasing return-on-ad-spend

- Look for exceptional performance

Look at your campaign performance according to certain date-related variables, such as:
o Time-of-day
o Day-of-week
o Day-of-year
o Week-of-year

Watch for consistent trends of exceptionally good or bad performance. Both instances provide actionable information.

For example, you may notice your conversion rates are higher on Mondays and consider increasing bids on that day. You may also notice your cost-per-acquisition skyrockets in January, so you might consider cutting down on ad spending that month.

Although conditions will vary for every business, common time- and date-based trends include:
o Different behavior on the weekend compared to weekdays
o Early morning hours performing poorly
o Strong sales early in the week

Tactic #2. Gather good data

You must have enough data to make informed decisions -- otherwise you risk wasting money on statistically insignificant trends.

Matthew Mierzejewski, VP, Client Services, Rimm-Kaufman Group, suggests marketers have at least 10 data points before making any changes. That means if you’re targeting a certain day of the month, you need at least 10 months worth of data to constitute a reliable trend.

- Data must be "normal"

Data points skewed by holidays, promotions or other abnormal factors should be thrown out. You want your data to be as "normal" as possible.

"As you wade into November and December, you’re kind of throwing everything out the window with Cyber Mondays and such," Mierzejewski says.

This makes longer-term trends -- such as those occurring on certain days of a month, or weeks in a year -- difficult to identify. Seasonality and other outside factors tend to skew the data too much to make it useful. It’s easier to gather 10 "normal" data points for a time-of-day or day-of-week analysis.

- Volume needed

The data you’re analyzing also should have enough clicks to be relevant.

"When there’s a keyword that gets five clicks a month, it’s hard to be actionable on that," Churchill says.

Looking at aggregate data, Mierzejewski suggests 50,000 clicks as a benchmark.

"That’s a point when you start getting more confidence in the numbers. Certainly you can make decisions before then, but you’ll be more confident at that point."

Tactic #3. Monitor slow-converting clicks

Many people click on paid search ads when they’re researching a purchase, but actually make the purchase several days later. When attributed to non-branded keywords, these clicks are valuable. They helped drive the sale and should be counted as such.

If your data only reports conversions that occur immediately following a click, then you’re not giving these slow-converting clicks enough credit. Including them in your analysis will help you make better decisions on which clicks are valuable, and which are not.

- Cookie PPC visitors

You can gain insight into which clicks lead to delayed conversions by tracking visitors after they click on one of your PPC ads. You can do this, for example, with cookies.

Unfortunately, many consumers surf the Web on one computer at work during the week and on a different computer at home on the weekends, which means you’re unlikely to catch everything.

Tactic #4. Test bidding changes carefully

Once you’ve identified a trend you’d like to exploit, don’t dive in head first. Make modest adjustments to your bids.

For example, if you notice campaigns consistently performing 10% higher on a Monday, you may be inclined to increase bids for that day by 10%. But start smaller by increasing bids half that much and monitoring the results. You can always increase bids later.

There are so many factors in PPC, from consumer confidence to your marketing schedule, that you should not assume the performance you saw in historical data will hold up under increased bidding. Anomalies and fluctuations abound.

Tactic #5. Invest time wisely

You should start by testing your high impact keywords. Doing so will ensure your time and effort will be well rewarded.

"It would behoove you to focus in on the big spenders first before you wander off down the weeds on the long-tail terms," Churchill says. "If your core spend is not operating effectively, you’re not going to make it up on the long tail."

- Start broadly, drill deeper later

Ferguson’s team started by adjusting bids across their entire account to help take advantage of noticeable trends. Once they achieved success, they looked at each individual campaign and adjusted those bids accordingly. They continually drilled down, adjusting bids at an increasingly granular level.

Tactic #6. Regularly revisit the data

Search performance changes over time. This is why finding actionable trends beyond time-of-day and day-of-week analysis is so difficult, and why PPC management is an ongoing process.

"You’re never really done. If you want to continue to make the account better and make it perform better, you have to keep your foot on the gas pedal and continue to make those adjustments," Churchill says.

Seasonal trends should be analyzed at the end of a season to check how your campaigns performed, and to see where you can adjust for next year. As for shorter-term trends, don’t bother checking every day. The fluctuations will be unnerving.

"We would advise no fewer than eight weeks for the day-of-week type of point, and ideally longer than that [for an average size campaign]," Mierzejewski says.

Useful links related to this article

Members Library -- Attributing Conversions to Assisting Keywords Lowers CPA: 5 Steps to Optimize Bids

Members Library -- Week-Parting, Copy Tests & Keyword Trials: 3 Tests that Doubled Search Leads

Members Library -- Improve Attribution: 8 Steps to Measure the Impact of Your Marketing Efforts

AdQuants: Advertising Volume for Google's Top 100 Advertisers in March 2009

KeyRelevance

The Rimm-Kaufman Group

Local.com


Comments about this How To

May 10, 2010 - Rachel Simeone of ZetaBlue Marketing says:
This is a very helpful and insightful article. To tactic 2, I would also add that companies also should consider major changes in the way that they do business, or special offers when they gather data. A major site redesign, a hugely popular offer, an overall shift in average unit offered can all affect data collection.



Post a Comment

Note: Comments are lightly moderated. We post all comments without editing as long as they
(a) relate to the topic at hand,
(b) do not contain offensive content, and
(c) are not overt sales pitches for your company's own products/services.










To help us prevent spam, please type the numbers
(including dashes) you see in the image below.*

Invalid entry - please re-enter




*Please Note: Your comment will not appear immediately --
article comments are approved by a moderator.