Google Analytics v. Sawmill
February 06, 2007
Running ClinPage has forced us to learn a bit more about web-based reports. Scratch that. We are not sure we know anything. That may be a reach, as some readers regularly remind us.
But we do love a web-based reporting tool described a few paragraphs down. Right off the bat, up front, we concede that this tool is not (as far as we know) something that can be applied to clinical data management. It just handles the extremely mundane task of analyzing who (or what) visited a website.
To be clear, we’re not talking about merely saving exported report data to Excel or Word. Any web reporting tool can do that. So can the two systems described here. We’re talking about using an online, browser-based system to learn, in a few clicks, a great deal of real-time status information about an ever-changing situation. Running on the web, to our surprise, both of these applicaitons are more responsive and easy to use than Microsoft Excel—or any other application running on our desktop. This is new.
We assumed, after five minutes of in-depth online research, that Google Analytics would be the best. It has a strong reputation. It’s free. It’s powerful. It’s tied into Google’s search, keyword and advertising services. As with just about everything else, the sheer excellence of what Google offers, and its ease of use, make its Analytics offering a very tough competitor. Picassa will hurt Adobe, and Google is rumored to be close to releasing its own web-based version of Powerpoint, too. These guys will own everything eventually.
Except, maybe, website analytics. To be fair, Google Analytics has two powerful features. Google Analytics is smart enough to ignore non-human website traffic. (We’ve been shocked, just shocked to learn that some websites may include nonhuman website visitors to inflate their statistics.) Our analysis shows that Google’s algorithms never let that bad data in, meaning no poor schmuck (us, your humble correspondent) has to filter it out. Kudos to Google Analytics on that.
When combined with Google Adwords, Google Analytics also has a very powerful way to globally select the date ranges for reports. We’ve got a screenshot here of our own modest effort to advertise ClinPage. It shows the Google AdWords pulldown menu to select the reporting dates. Very simple. Very clean. Very nice. Can your clinical reports do that? Let us know. Minor quibble: Google Analytics, per se, has no such dropdown menu. You only access it through Google Adwords.
But it turns out that our favorite web-based, web-traffic reporting tool is a small, obscure company called Flowerfire, out of Santa Cruz, California. It makes a nifty, web-based program called Sawmill. We have access to Sawmill through the web-hosting firm that serves ClinPage content. Sawmill soundly kicks the wide, complacent posterior of Google Analytics. What the Sawmill reports show is less interesting, for clinical readers, than the way that the application displays data. And, dear God, the ease and speed. After driving a Maserati like Sawmill, driving Google seems like a Isuzu.
Sawmill delivers a few key advantages that are extremely powerful when combined. All Sawmill reports can be quickly updated with a simple “refresh” button. Unlike Google, Sawmill gives you access to data in real time, not with a one-day delay. Through simple checkboxes, any data field in Sawmill can be included or excluded from a report.
In our case, we can quickly distinguish web traffic from Germany vs. web traffic from France. In a clinical context, you could include or exclude, with a checkbox, patient data from UCLA or Dr. John Doe. With a click, you can see all the filters, or just the ones that have been activated. It’s easy to insert a regular expression (“myocardial” or “Gilmartin”) into a filter.
Turning off a filter does not remove it; you can return to the Filters pane and click the checkbox to turn it back on at any time. Better still, all checked filters can be globally turned on or off with a separate checkbox that is right next to that “refresh” button. This allows you to see the filtered data and, a moment later, the unfiltered data. How may people visited yesterday? What is ClinPage’s busiest day of the week? Sawmill knows. Wednesday.
This option to include or exclude is quite powerful. What if you wanted to just focus on recruitment trends at clinical sites with key opinion leaders? Or exclude them? The right reporting tool should let you do that. As with any good technology, with any new instrument, there is a place for play—for experimentation with the system to see what secrets it holds.
The right tools for sifting and displaying the data make the task, for lack of a better word, fun. (Among other things, Sawmill taught us just how often a public relations firm, Bacon’s, was remotely monitoring ClinPage.)
Beyond that, Sawmill has what it calls “interlinked” reports. Sawmill provides this explanation of what “interlinked” means. Google also allows progressively more detailed reports by clicking on hyperlinks. But it doesn’t allow the data to be as easily sorted and sifted with each click.
No Manual To Read
Basically, Sawmill has a better approach to immediately refreshing your data. That refreshing is based on criteria that were specified only by the user’s own intuitive clicking through the hyperlinks in a previous report. You don’t have to set anything up. You don’t have to save anything.
If it were running in a clinical trial context, Sawmill would not have to be programmed to know that you were just looking for open queries, in Texas, more than two weeks old. You’d have navigated there in three clicks. Once you’ve run a report (i.e. clicked on a few links), Sawmill allows you to “zoom in” on any additional field in the database. That supplies more granular insight, based on the previous report. For example, if you’d run a report on a subset of clinical sites, within a subset of one data monitor, you can easily generate a bar graph of the distribution of numbers of days to resolve queries.
Pick A Date
It took us a bit of playing around, but we eventually discovered that even selecting dates for reports is far better in Sawmill. There is a separate filter just for date range selection. It allows easy, customized selection of a date range from July 31, 2006 to February 2, 2007. But if you just want to get in and out, it’s even faster. Once the Sawmill calendar filter is open, you can select a single day, a single week, a single month with one click. Bang. You’re done.
In a clinical context, this would make it dead simple to say, “OK, we launched our radio ad for this trial on June 4. What happened at our sites during the week of June 4?” Can your existing systems do that? Does someone need to take a day-long course and learn a new program to do it? Or can they click around on a browser and know what they need to know? Better web-based reports are on the horizon as tools available to the larger tech community slowly trickle down into the clinical trial landscape. Just as Google already has web-based alternatives to Microsoft Word and Excel, there will be alternatives to the leading desktop tools from Cognos, Business Objects and other firms.