Marin Software's official blog

The 2 Basic Steps for SEM Traffic Forecasting

Cong Cao
|
November 6, 2014

In search marketing, the bidding and optimization are fairly standardized across different marketers and agencies. Most search engines and bidding platforms enable users to conduct sophisticated optimizations. But for traffic forecasting, there isn’t an industry standard process. During my 7 years in search, this is the one subject no two marketers can consistently agree on.

Traffic forecasting is important, because accurately assessing the investment and performance potential could help set the appropriate expectation between different stakeholders. Secondly, it almost always needs to be determined weeks if not months ahead of the activity, in order for the clients or other budget providers to procure. Lastly, in a large agency environment with multi-channel campaigns, having an accurate SEM budget forecast allows different media teams to allocate the right relative investment and create the most optimal media mix.

In addition, most marketers have the resources to do it. But sometimes, important factors are left out of the process. In a large multi-million dollar per quarter campaign, missing a key factor during forecasting could lead to a large absolute deviation.

For me, the forecasting process should include two steps. The first is to project what you could spend, by estimating the full market capacity. The second is to define what you should spend.

Step 1


To estimate market capacity, you need three factors: historical performance, historical Share of Impressions, and the projected change in search volume.

  • Historical performance: it’s the foundation on which other factors could modify. On a basic level, you could start with spend and clicks. If needed you could also add in other metrics like conversions.
  • Historical share of impressions (also coverage or share of Voice): it is the percentage of one advertiser’s earned impressions among all possible impressions, for a defined set a queries. Past Share of Impressions is what calibrates between what we have achieved and what we could at full capacity. In our case example, I have highlighted historical performance and share of impressions. Based on these, I’m able to assess the full market capacity. But it’s not yet complete.



  • Projected change in search volume: this information could help you further calibrate you forecast, but in the temporal dimension. The projected change should comprise of Seasonality and Trend. Seasonality is defined as the repetitive and predictable fluctuation in the volume and pattern of behaviors, such as search impressions, in intervals of days, weeks, months or quarters. Trend on the other hand is exclusively non-repetitive changes. An example of seasonality is the demand for allergy medications, which peaks every springs and falls. An example of trend is the search demand for Blu-ray DVD, which has declined over the years. To project search demand, I usually use Google Trends and other industry information, to review historical patterns and to extrapolate future changes. In our case example, I project a 15% increase in search traffic Year-Over-Year. Now the first two data columns show a different set of spend and click data.


Step 2


Now we have estimated the 2014 traffic at full capacity, we move onto the question of how much we should spend. In this step, we need two new factors: Internal KPI/goal.

  • KPI/Goal: defining a KPI and setting a goal refines the forecast from what’s possible to what’s reasonable. The goal can be a number of metrics such as impressions, sales, CPC, CPA. In our case example, we have a CPC goal of $0.50.
  • Desired future Share of impressions: This is the lever that determines the equilibrium between maximum engagements and staying within the goal. In our case, we know campaign B has a CPC lower than goal, so we should prioritize budget allocation at 100%. Next, we try a few different SOVs for Campaign A. But at the end, a SOV of 8% help us reach a Total CPC of $0.50.




Using this methodology, you’ll be able to forecast accurately and set the appropriate expectations across all stakeholders!

About the Author


Cong Picture 2

Cong is a Paid Search & Paid Social Director at GroupM, leading the efforts across multiple clients and publishers. She’s responsible for driving market and performance analysis, SEM strategy, digital best practices and processes. Her industry experience spans across consumer electronics, pharmaceutical, consumer logistics, B2B technology, auto, finance, travel, and education.

Cong graduated from Swarthmore College, where she studied economics, statistics, and philosophy, after a short stint in Engineering. Before SEM took over her life, she sang in an Acapella group and enjoyed flamenco dancing. In August 2013 she climbed Indonesia’s most active volcano, Mount Merapi, and fortunately, it did not erupt that day.

Sign Up and Get Updates by Email
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.