In a post yesterday I looked at the current accuracy of Google Insights for Search data. Further.co.uk had indicated that the data presented by Google could be less valuable today than it was in 2008. I agree, it does appear that the data has been modified (it looks like the data has been smoothed).
I believe the data produced by Google Insights for Search is more valuable than Google Adwords; and this blog post should walk you through the workings that led me to this to conclusion. Google Insights doesn’t give you accurate monthly search volumes ‘out of the box’ (you have to run some multipliers) but it does give you more variability in the data ranges from week to week (you can see how clipped the Google Adwords tool data is in the screenshot below).
I read the increase variability in Google Insights for Search as a more honest reflection [by Google] of actual search trends; on a weekly and monthly basis the data has a level of noise that is indicative of real fluctuations in search volumes.
For this post I have extracted data sets from (1) Google Insights for Search, the (2) Google Adwords keyword tool and (3) actual data from a PPC account. Note: because Google Adwords only gives you the previous 12 months of search volume I have only looked at a 12 month date range for comparison.
As mentioned in my previous posts, you have to amend the raw data that exports from Google Insights for Search; you need to multiply the data up by a factor to get it to map onto the Google Adwords data. In the two charts below I have plotted Google Insights data (INS), Google Adwords tool data (ADW) and actual data from a Google PPC account (PPC). I used multipliers of 150 (chart 1) and 200 (chart 2) to map from Google Insights onto the Google Adwords Keyword tool extract data. Also included is EXACT match data from a PPC campaign – this affords the ability to ground the whole data-set against an actual known monthly search volume.
Key to the charts:
INS = Google Insights for Search data extract
ADW = Google Adwords Keyword tool extract
PPC = Google PPC data extract (monthly impression x impression share)
“ “ = Phrase match
[ ] = Exact match
Chart 1 – A 3-keyword phrase
.jpg)
Chart 2 – A second 3-keyword phrase
Observations:
I make the follow observations. I’d be interested to catch your opinion too – leave a comment and I’ll follow it up.
1. The data from Google keyword tool does appear to be a close fit with actual PPC data – it doesn’t have the same levels of monthly fluctuation but it does trend in a similar manner.
2. The Google Insights for Search data follows the trends too. The added monthly fluctuation in the Insights data appears to track the actual PPC data more closely.
3. It is difficult to determine if any one of the three keyword match-types maps better than the others.
4. Google Insights for Search only shows trend data for broad and phrase match (not exact) so the trending tool is not so strong if you are planning an exact-match keyword campaign.
Summary
In summary, you can use either Google Adwords tool or Google Insights for Search to forecast keyword volumes.
– The Adwords tool data gives you relatively accurate volume data (the data spikes are removed to give you smooth estimates) for the last 12 months but it has no forecasting data.
– Insights for Search gives you data for longer date ranges (the data goes back to 2004) and gives you data forecasting. However, you have to put in some work to get at the data (you need to benchmark the data first), you can only compare 4 additional keyword phrases against your benchmarking phrase and the tool does not provide data volumes for exact-match search phrases.
The data provided by Google Adwords is clearly easier and quicker to work with but I feel the additional effort for using Google Insights for Search has its rewards.
I am keen to share learning on keyword volume analysis and forecasting tools – please give me your feedback and findings from your own number crunching. Drop me a note in the comments section (with your name and email address) and we can pick this up via email.