Over the past few weeks – months, even – I’ve been looking into ways to monitor what’s being said ‘out there’, and then measure it in a meaningful way. My conclusion? There are plenty of tools out there that can almost do the job, but most of them fail in some significant way.
Yahoo Pipes is kaput
For people like me who sort of know how to program but don’t want to bend their brains around it any more, Yahoo Pipes is a godsend. It’s a nice visual way of doing funky things with RSS, and I’ve spent quite some time creating a modular monitoring system with it: that is, change keywords for a client in one place and it reflects through all the searches, or change one search and it reflects through all clients.
Now, the system works. I know it works because I’ve tested it with my recent ‘What are people saying about…?’ series of posts. It’s a nice, elegant system that anyone can use, adapt and extend.
But even while writing those posts I noticed that Pipes was going from idiosyncratic (occasional crashes, strange error messages) to downright insolent. Today it’s a rarity that I can save a pipe successfully, or preview the results, or even run one. To my mind it’s a system that is actually falling apart as I type and whereas the pipes development team have responded telling me they’re working on this, they’ve been working on it for a good few months now and I’m not holding my breath.
Google Docs is good, but limited
Next up: Google Docs. You can actually create a dashboard system with Google Docs with the imporfeed function, and this is quite cool because you can also generate charts of these feeds on the fly. More to the point, you can share these charts as straight images, so they can be built into any web page or dashboard widget, such as Netvibes.
Everything about this is really nice. For example, you can format the text output and bring it into something like Netvibes to have a beautiful, consistent interface, with all sorts of charts showing what’s happening when it happens.
The only problems are: importfeed seems to work one day and not the other, which immediately makes it completely unsuitable for presenting to a client; you have to keep the document open for it to generate any results, so if someone accidentally closes the doc or shuts off the PC, all your clients are looking a blank screens; and, critically again, importfeed only brings in 20 entries.
So, if you’re monitoring for what people are tweeting about, for example, Eurostar, you’ll find the importfeed has already hit its limit within the first few minutes. I’ve tried ways around this, such as creating a separate feed per day, but it still fills up. There doesn’t seem to be a way to append to the importfeed results so when they’re gone, they’re gone.
Excel works, but is clunky
So, on we go. Excel does import web data but doesn’t do it very well. My Excel crashes when I plug web data into it. It does recover, staggering and coughing, to its knees, but then it’s just very clunky. If you update a web query you can make it append rather than overwrite, but then you need to add fancypants equations to dynamically tell you when you’ve brought in duplicates, which is highly likely if you’re refreshing every day for something that doesn’t update often. This is almost the opposite of the importfeed problem: too many duplicated results rather than too few unique ones. The connections management seems a bit weird too, and I still don’t think I’ve quite figured it out. Then again, while you can draw beeeeeeyootiful charts from the data, and store it for future analysis, you cannot share these charts online. So, fail overall I think.
So where are we up to?
I’ve given up on Yahoo Pipes. It’s unworkable. So I’m looking at ways to use Google Docs to help with very quickly assembling web queries – that is, a wizard-like interface that guides you through includes and excludes and then previews the results across platforms – and then using those queries to import into Excel, display in Netvibes, and bring into Google Reader so I can store/analyse results, display them, and get insights into how I’m using them in Google Reader.
This is not ideal. There are too many different apps in this system for my liking. What I would like is to have everything monitoring-related in Pipes, and then just pull that into Excel for measurement and Netvibes for display. But no. As I said, everything works, but not quite perfectly. One day Yahoo will make pipes work; or Google will extend the importfeed limit; or Excel will improve web data import. Until then, I just have to blog about them despairingly.