Yesterday, Andy Leonard and Jessica Moss announced that their Wednesday 3pm class on SSIS scripting will cover Twitter integration. In today’s SSIS Boot Camp by Brian Knight, I spoke with some DBAs about the Twitter idea. The initial reaction has been, “Why would you want to do that? Twitter is so unreliable.”
True: us database administrators are conditioned to working with extremely reliable tools, logged transactions, mission-critical uptime, yadda yadda yadda. Twitter is none of those things – so why would we rely on it?
After they stop to think about it for a minute, there’s been some interesting use cases come up:
Tiny daily status reports to the public – say you’re a fundraising group like United Way, and you want to let your followers know how close you are to your fundraising goals. Or maybe you’re a blood bank, and you want to tell people which branches are open or what types of blood you need most this week. Tweets would be a great, free answer that scales pretty well, and it’s a pull-based setup where anyone can sign up without you writing a subscription interface.
Mini-reports of yesterday’s metrics for global users – same thing for companies, except you’d protect your Twitter feed. You might send out a tweet each morning with a recap of whether we were on track for yesterday’s sales. Imagine an automated tweet from the sales executive recapping yesterday’s sales vs goals. Of course, keep security in mind – may just want to say, “Sales yesterday was 1.5% over goals!”, not actual sales numbers, even if your updates are protected.
“Interesting record” alerts – if you’re running a kick-ass IT question & answer site, you may end up with users who can’t wait to find out when their question is updated – but they only want to know when they’re online. I could easily see value in setting up an SSIS/Twitter integration on StackOverflow.com so that when any of my posts are edited, or when someone answers my questions, it sends me a tweet. It wouldn’t clutter up my inbox, and it’d only show up when I’m online. (Or if I was a hard-core geek, I could set up Twitter to text me when these come in, but I’m not that hard-core.)
Deal alerts for your customers – Amazon posts sale alerts when items go on sale in their MP3 store. Check out the AmazonMP3 Twitter feed for an idea. We could set up automated jobs to take the items in our inventory that have the highest in-stock amount but lowest volume, for example, and discount the price by x% to move it out. Granted, you’d want to put a lot of business logic in there to make sure it doesn’t discount something that you don’t want to discount, but it’s an interesting way to try to move product. Yes, you could do this in a separate application, too, but doing it entirely in SQL Server would be a fast way to get the job done.
Building Twitter bots for sales force automation – imagine a Tweet conversation where one of your salespeople could send Tweets to a bot connected to your CRM system or data warehouse:
- From @BrentO: “@QuestBot whois Jane Doe”
- From @QuestBot: “@BrentO Jane Doe, ABC Consult., Boston, last order 9/10/08, $10m, rep John Smith”
Of course, you’d probably want to do this via private direct messages, not public messages, but you get the idea. It’s a fast way of building a web-service-style CRM system without a lot of infrastructure.
Data mining your Tweeps – you can’t really follow more than a couple hundred people without losing Tweets, and even at that level, it takes a lot of time to digest your crowd. I would love to see a summarized dashboard of activity from yesterday with what keywords were popping up, what links were sent out, who else my followers are talking to (so I can maybe find more interesting people), and so on. I know I can do that if I can get everything into SQL Server, and this would probably be the first thing I’ll try to accomplish.
I’m way excited about this – can’t wait for Wednesday’s session. 3pm. Be there or be square.