Making big claims to your clients is gutsy, so we document evidence of every single advertising claim made on SpyFu.
We keep cached pages of every Google SERP we’ve ever captured.
Every piece of information you see on SpyFu — every ad, every organic rank — can be traced back to a specific Search Engine Result Page (SERP) that we saw on Google. For any single keyword search: we gather the first 100 organic results and note the domain and page for each one. We also gather ads from the first page of results and note the advertiser, ad copy, and landing page for each one. All of that data goes into our database, and we repeat that across millions and millions of keywords.
Further, we keep a screenshot of each one going back over 15 years to when we first started collecting data.
When we say that some domain was advertising on a certain keyword at 3PM on July 12th 2019, we have indisputable visual proof that they were. Same goes for organic rankings. This is iron-clad 100% fact.
From that foundation, we can carefully calculate other roll-up metrics like Clicks per Day and Estimated Budget.
Where We Get These Stats
For some time, all of the findings published on SpyFu came as we searched Google repeatedly (across millions of keywords) and used Google tools to arrive at spending estimates.
Since then, we have dramatically grown our data. We now use multiple sources and more frequent updates to come up with the strongest, most reliable data points possible.
Any new data is published instantly instead of waiting to be processed every month.
The figures like costs, search volume and number of clicks are estimates that come from other sources. This video gives better details about how we get that data.
In their keyword tools, Google tells us how many searches each keyword gets per month, and how much each one costs to buy in Google Ads.
There are a lot of variables that go into each part of the problem — take the organic click through curve, for example: the number one search result normally converts about 21% of searches into clicks. But, when there are shopping results on the page, it drops to about 8%. And what about Ads, and videos, news, images… we take all those things into account — for every single keyword — for every single domain.
Over time, we’ve made many improvements including the example about taking shopping, video and other results into account. Realistically, these days, you can expect our roll-up estimates like Ad Budget and SEO clicks to be accurate about 90% of the time.
That said, we have a good grasp of scale. If we say a domain is spending $10,000. There’s no way they’re spending $500 or $100,000.
And When We Get it Wrong…
Here’s even better news: When we’re wrong, we’re consistently wrong in an industry. So, if you’re trying to size up your competitors all you have to do is calibrate their numbers against your numbers. Make sense?
Let’s say that your domain is MeatYums.com and we think your Google Ads budget is $10,000/mo. But, you know that you’re really spending $20,000/mo. When you look at your competitor MeatyProducts.com and we say that they are spending $18,000/mo, then you should calibrate our estimate based on your inside knowledge to $36,000.
Being off consistently within an industry comes directly from the noisy CPC and search volume data we get from Google. You’ve probably seen it yourself using the Google Keyword tools. We’re working with the same data. When we’re inaccurate it’s because that input data is inaccurate, and it affects all advertisers or organic rankings across the board.
UPDATE: Thanks to some feedback, I’ve added a screen shot of exactly where you can get to the cached SERP page screen shots for any keyword.
Even when someone stops running a campaign, we still have proof of the ads. History is locked in place, and if we said it happened, we can back that up.
A note about past cached pages in 2021.
In some past instances, our linked cache pages displayed keyword data that had been collected earlier in the month and had not yet caught up with the rest of the keyword-related calculations on the page. It caused some users confusion. This has been corrected as of December 26, 2021.
The cache page in action
In this next example, the cache page explains exactly why a competitor showed up on a word that didn’t seem right.
Barnes & Noble, for example (barnesandnoble.com) advertised on “Book flights to Corfu”, a luxury destination keyword that draws advertisers like Kayak, Travelocity, and Lufthansa. Barnes and Noble was attempting to match on the “Corfu Book” element since they likely sell books on this Greek island. Sure, a large advertiser might lose track of a few words, but they would easily scoff at the notion of having advertised on something in a heavily competitive (not to mention unrelated) industry.
We can prove it, though.
Let’s find that keyword and go look at the screen shot of the SERP we saw on that date. (“View cached page” is available at the top of all SpyFu keyword stats pages as well as a link inside Ad History individual ads.)
Where Barnes and Noble attempted to cash in on travel books, they were matched to the expensive keyword that would have left a site visitor totally stranded.
An advertiser this big might never know how mistakes like this act as a slow leak in their budget. But it’s not just limited to large advertisers. Any domain is a candidate for a broad match misfire. If this is you, build a double-check into your ongoing SEM research. It gives you better understanding of why a competitor might have a particular term in their history.
Cache goes back as long as we’ve been around.
The cache gives you assurance in what you’re seeing. We stand behind our data-absurd or not.
Leave it to those whose profession is to dig for facts to put our cache into action. During the 2008 United States Presidential campaign, The Washington Post turned to us on word that the main website for candidate John McCain (www.johnmccain.com) was advertising on the keyword “hot wife.”
By the time a visitor could type that term into Google, the resulting ad could be long gone. Without our cache, there was no proof of such a seemingly absurd claim.
However, captured and locked in time, is the result page. July 2008, page 2.