Tuesday, April 12, 2016

Develop free, lightweight attribution insights for your paid & organic channels

data-analytics-team-ss-1920

When running paid search alongside your organic SEO strategy, it’s essential to ensure you have an integrated strategy and a consistent message in the search engine results pages (SERPs). To manage this, data analysis at the keyphrase level — for search terms where both channels are active — can improve traffic and reduce inefficient paid spend.

Usually, to perform this type of analysis, you’d need to spend six-figure sums for powerful attribution insight tools from Google or Adobe or the like. Personally, I prefer to roll my own for an admittedly less deep, quick insight tool focused just on channels visible on Google’s SERPs.

To map paid search term data to organic, start by pulling comprehensive daily data on all keyphrases driving traffic to your site from Google’s Search Console API. I’ve written about this process before on Search Engine Land, so I won’t go over old ground.

One major change to the described approach, however, is grabbing the data segmented by device. Device segmentation to mobile, tablet and desktop can be matched to AdWords data and allows like-for-like comparison of SERP click-through rate performance, which lies at the heart of our analysis.

Once you have started collecting (and storing) this data daily, mapping data to Google’s AdWords API for matched time periods and devices will enable you to discover opportunities to increase the “halo effect,” wherein a website receives increased click-throughs for both channels when paid ads are visible alongside highly ranked organic results.

Mapping this data together also helps identify opportunities to reduce paid spend cannibalizing organic results where very strong organic click-through rates are occurring.

By extending the data to pull in attribution models used in Google Analytics, you can start making significant improvements to the overall efficiency of your entire search marketing budget by repurposing inefficient spend elsewhere across multiple channels.

In effect, you have a form of lightweight Google Attribution 360 (fka Adometry) insight you can apply to your paid search channel at a fraction of the cost (in fact, for free).

Building the halo tool & running data analysis

We are combining three data sources, via three APIs: Google Search Console, Google AdWords and Google Analytics. The same analysis is perfectly possible with Adobe Analytics API, as well.

Additionally, Bing Webmaster data can be used for markets where market share is large enough to be worthwhile for analysis (looking at you, America!). Search click-through rate data are specified here.

Note: I have not tried building an equivalent analysis for Yandex, as it appears not to offer the necessary SERP impression and click data, but I would be interested in any attempts made to generate the equivalent data for Yandex in Russian markets.

As a brief walk-through, pull the Search Console data as described in my previous Search Engine Land article (shown below).

 In addition to pulling the raw Search Console data, calculate the number of terms in each phrase, and compare the actual CTR to the expected for the position. Finally, also calculate the available traffic should the rank be improved to position one at the expected CTR. Use formulas to filter by brand or generic as described in my previous Searchengineland post. I also use a pivot table to show the datestamps for the data and pull that into the final reports for convenience.

(click image to enlarge)

Then, to collect the AdWords Data, pull the data shown below via the API.

Perform the same brand and wordcount calculations with your AdWords data. For ease of data-selection, I'm pulling the below conversion data using a uniquely generated key to match up my keyphrase terms.

(click image to enlarge)

To allow for attribution, we now pull revenue data from Google Analytics, which means we can actually pull data for different attribution models available within Google Analytics.

For a starting point on attribution analysis suitability and approach, I recommend Avinash Kaushik’s primer on the topic.

Here is what the Analytics data pulled looks like.

Pull revenue data via Google Analytics so you can apply an attribution model. To map this data back to your AdWords data, create a key using the campaign name, the search term and the match type. Merging the data can be done in the pivot table.

(click image to enlarge)

Now, we get our pivot tables out. I typically create two pivot tables from the three sets of data: one to allow for sorting by different priorities for the most recent time period (the last day, week, fortnight or month, depending on your preference)…

This pivot table is used to create sort conditions for the report segments.

(click image to enlarge)

… and another to look up historic data for comparison with the first time period.

This pivot contains all historic data to allow cross referencing in our reports, and time period performance comparison.

(click image to enlarge)

Then, create a new tab in Excel, and style up a report which pulls data from the first Pivot prioritizing paid spend like this:

This report can be segmented through the long tail, or by brand Vs non, device, location, etc and highlights organic performance gaps for high value paid search terms. Time to get your copywriters busy!

(click image to enlarge)

And then also create one prioritizing organic performance like this:

This report can be segmented in the same ways and highlights opportunities to test reducing paid spend where organic performance is very strong. Lets get efficient, and say hello to a free, lightweight Adometry-style insight!

(click image to enlarge)

Because we are calculating the number of terms in each phrase, we can create reports to the “fat head,” “chunky middle” and “long-tail” terms in our data. We segment by device at all times to compare like with like. And because we have spend and revenue data alongside our click-though rate (CTR), we can see where our opportunities to save spend lie.

To (very simply) illustrate this, see the report snapshot above showing brand terms with AdWords PLA spend against them. In this example, roughly £8,000 per month is being spent on those top 1,000 terms.

Given we have a greater than 50 percent above expected click-through rate for organic across 90 percent of those terms, we can test switching off that budget. We have more than 200,000 terms in our data set, so there’s plenty more opportunity to be found.

To understand if the budget was cannibalizing organic traffic, we re-run the report the next day, week or fortnight and confirm that we have the same overall traffic coming through.

When you are confirming conversion and revenue, with sound attribution models, then you should expect to run reports against individual products or categories and measure the overall revenue impact of pulling out paid spend. If your overall revenues are unchanged, then you can safely move that budget elsewhere.

In this way, you can drill through the highest spend areas of your account to find overlap or “cannibalization” opportunities to test, increasing the efficiency of your search marketing campaign.

The post Develop free, lightweight attribution insights for your paid & organic channels appeared first on Search Engine Land.

No comments:

Post a Comment