Does Twitter need advertising?
Dec 3rd
Recently Twitter has started rolling out it’s “Promoted” features. Promoted accounts appear in user’s “Who to follow” suggestions, Promoted trends appear in the trending topics section, and Promoted tweets appear in timelines and searches. Obviously all of those things generate revenue for Twitter. They bring in quite substantial amounts too. A worldwide promoted trend can apparently cost as much as $100k.
Twitter using an advertising model was inevitable. They’d be mad not to.
This raises an interesting question though: Is the advertising model is actually necessary for Twitter?
Twitter have access to an incredible amount of data. That gives them the sort of real-time, up-to-date business intelligence that people will spend huge sums of money on. You just need to look at the sites buying Twitter’s firehose to see that:
- Microsoft
- Yahoo
- Jive
- Converseon
- Twazzup
- Collecta
- CrowdEye
- Scoopler
- Kosmix
- Chainn Search
- Ellerdale
- TweetBeat
- …and probably more.
The firehose is a data feed of every single tweet on Twitter. How much it costs to access depends on the size of your company and what you’re doing with the data, but considering last year there was evidence that Google paid $15m and Microsoft paid $10m, it’s not unreasonable to think the rest of the companies are paying $1m+ each. Even the “halfhose” stream (a feed of a random 50% of tweets) costs $360,000 a year. There’s a great deal of money in selling raw data, and more in selling tools to analyse that data.
Clearly, Twitter aren’t going to abandon adverts. They’re going to be pushing more out to our timelines soon (if you use the #newtwitter website client). I just think it’s worthwhile considering how else they’re making money when we think about how prevalent the ads are going to become. Twitter will never be chock full of adverts. It just doesn’t need to be.
The Right Way To Get Things Wrong.
Oct 6th
“The website is down.”
That’s an all too familiar message. Errors on websites are still shockingly common. And what the user sees when your website is suffering an outage of some description, be it a server error, a missing page, or simply the user getting something wrong should be a big concern in your life if you care about retaining their custom.
Back in the olden days of the internet, so about 18 months ago, most websites had customised “404″ pages, the page that the user sees when they go to an address that isn’t found. Giving the user something more friendly than “Not found, try again” was seen as best practice in web development. Some sites came up with amusing ways to tell their users that the page was missing. Of course, some sites still just dump the not found message on you. But that practice hadn’t spread to other server error pages. Users were still faced with an unfriendly, complicated message if the server broke.
That changed with the growing popularity of Twitter. As the site grew the developers realised that server problems were a real, and frequent, problem. Obviously they fought to keep them to a minimum but they also put in an error message that would placate and pacify the user – the “fail whale”.
It was nothing short of a stroke of genius. People largely stopped complaining about outages and problems and referred to the errors are “fail whales”. While it was just a semantic change it alter people’s perceptions and attitudes towards the site. I strongly suspect that Twitter’s enduring success is down, at least in part, to that whale. Without it people would have abandoned the site and not returned. With it they weathered the problems. It’s worth noting that the engineers at Twitter have seemingly always been more keen to keep the API server block running more so than the website. The API servers are what deliver tweets to 3rd party applications – ones without the whale. Users seeing the pretty, friendly error will wait for a fix longer than those who don’t.
This notion of making things friendly extends beyond Twitter. On an Apple Mac, particularly on an iPhone or an iPod, seeing a error message is incredibly rare even with unstable software. The application just dumps you back to your desktop or home screen, or occasionally the device locks up and needs a reset. Again, I believe that is a conscious, deliberate decision on the part of Apple. By not showing an error message (particularly a complicated one with error codes and text that’s only meaningful to the developer) the user doesn’t get confused by something they don’t understand. Everyone gets being dumped back to their desktop. They know what happened. They understand the application crashed. It doesn’t matter whose fault it was, the user knows that they need to start over. And they do. Compare that to an error on a Windows PC where it’s not so simple, where the user might see the famous “Blue Screen of Death” with it’s memory address notifications, segfaults and IRQ conflicts. It’s not friendly. It stops the user in their tracks.
How then can we learn from Twitter, Apple, and all the others who have embraced this simplicity? I believe Twitter especially have managed to get this right so I’d recommend a straight copy of their solution. Not necessarily with a whale, but with a single, humble error screen that conveys to the user what is going on – that there’s an error. It shouldn’t say anything else. Users don’t need to know that there’s a connection error with your database server or that there’s an unexpected T_STRING on Line 93 of your controller script. All the user should see is a friendly page asking them to try again later. Use your server logs for debugging things. That’s what they’re there for.
Twitter are planning to sell your details.
Oct 3rd
At IAB Mixx tomorrow, a conference dedicated to internet and interactive advertising, Twitter are allegedly planning to make a pretty big announcement. It concerns you. Rumour has it that they’re going to start selling users to advertisers based on their interests (sort of).
It makes sense really. Twitter’s key business strength isn’t providing a platform for micro-blogging or a social network or whatever it’s saying it is today. It’s key strength is data-mining, the act of finding information in a cloud of data. Twitter have a unique window into what we’re thinking about, and they’re looking into ways they can make money from that information. Obviously the most straightforward method is to sell what they know about each of us to people willing to pay.
Now, before you panic and rush to abandon your Twitter account, it’s worth noting exactly what Twitter are selling. When the advertiser pays they’ll get pushed to all the users interested in a particular topic. For example, if the advertiser buys “Nike”, anyone who’s been flagged as interested in Nike, trainers, sportswear, etc will see them in the list of “Who To Follow” in their account page. I imagine the advertiser will be flagged as a “Promoted Account” in much the same way as “Promoted Tweets”* are flagged in the trending topics. So who advertises to you will be limited by your actions – who you choose to follow. I’m going to guess that it’ll be a hot topic across some blogs tomorrow but it’s not really a big deal. It’s certainly much less intrusive than adverts injected into your timeline or a banner advert.
If you choose to follow the promoted account then, and only then, will you see their tweets (unless they mention you directly of course). That much is obvious. It’s a standard account in every way apart from it being pushed into your suggested users list. These promoted accounts won’t enable an advertiser to send out masses of directed tweets with impunity.
It’s interested to see that Twitter are actually doing things that obviously raise revenue at last. I’ve suspected for a while that they’re selling information behind the scenes, but this brings everything out into the public eye.
* FYI, a promoted tweet costs as much as $100k.
what the advertiser is allowed to do with this information about your account is going to be limited by your actions
How much Twitter is too much Twitter?
Sep 30th
In a recent blog post Nadine Dorries commented that someone she has heard of in her political sphere has tweeted approximately 35,000 times in just 4 months. That sounds like a lot. She goes on to describe how she believes someone wasting that amount of time on something can’t possibly be employed, and if they’re not then they ought to be because chronic tweeting is a sign that someone is capable of working.
It’s an interesting point of view. First of all, I think we should examine that “35,000 tweets in four months” thing. Assuming that this Twitter user is awake for about 16 hours a day, and tweeting every day, that gives us an average of approximately 18 tweets per hour. That’s one every 3 minutes give or take. Each tweet might as much as a minute considering you’d have to read your timeline, find something to say, and tweet it. That’s a whopping 33.3% of your waking life spent on a social network. Appalling stuff.
But wait.. That would only be the case if said Twitterer’s twittering is done as disparate individual tweets. Often people will tweet back and forth as a conversation, or several times in a row, or as separate replies to several things in their timeline. Assuming 1 minute per tweet is far too much. Wouldn’t it be safer to assume 1 minute per 2, or 3, or 4 tweets (on average). If that’s the case then we’re down to checking Twitter every 10 or 15 minutes. That’s still arguably too much for someone being paid to do something else.
Should an employee be allowed to check their phone for a minute or two every quarter of an hour throughout the working day? Personally, I don’t see why not. If they’re still being productive then it’s not really an issue.
Perhaps we’re missing something else though. I know from my own Twitter usage, which is staggeringly high by most accounts*, tweeting isn’t something you do every 15 minutes for your entire day. It’s something you do a lot of every so often. I often waste an entire hour on Twitter all at once. Spreading all those tweets out over the course of a day is unusual. It’s more likely that the Twitterer in question spends a couple of hours each day tweeting non-stop, conversing with their followers. If that’s the case then there’s little reason to assume they’re necessarily doing that during working hours. That might be how they spend their evening.
If that is the case, then one might still argue that the person is spending too much time on a relatively aimless task, but to suggest that this person is unemployable or that they’re a workshy scrounger is ludicrous. I often spend 2 hours of my day reading a book or watching a movie, or, heaven forbid, listening to political debate. Am I incapable of getting a job? No, of course not. That’s ludicrous.
The fact of the matter is, while a headline number of “35,000 in 4 months” sounds like a great deal of time and energy went into generating that much ‘content’, tweeting is not actually a particularly time consuming activity and consequently running up that number of messages is easy. If anything, Nadine Dorries has only highlighted her lack of understanding of how people use Twitter by her complaint.
As an aside though, I can understand her position looking at it from her particular frame of reference. After all, this is an MP who attended just 2% of the meetings of a parliamentary committee she was on. If she can’t find the time to do her job even when she’s not twittering then she certainly shouldn’t consider taking it up.
* It’s ok though, I own the company.
Twitter’s Next Steps
Sep 15th
Confirming what I have long suspected, that I am hopelessly addicted to Twitter, last night I actually stayed up to follow the Big Tuesday announcement live online. I found it rather entertaining. Yes, I know. It’s pitiful. I should get some sort of a life.
This big announcement turned out to centre on a new web client for Twitter.com. It looks rather shiny. There’s a new design, some new features, some refinements for the old features, but not much else. Some people in the social media space are saying it’s not really lived up to the hype. But I think there was something else in the announcement, something a lot bigger than the news of the website update.
Evan Williams, in his introduction, made a comment that Twitter is “a realtime information network”. This builds on a comment by Kevin Thau (Twitter’s VP of Business and Corporate Development) at a Nokia World presentation that “Twitter is not a social network”. Within Twitter there has been a shift in emphasis away from building networks of friends to talk to and towards pushing information out to a group of people. The expectation of a two-way conversation is being reduced. Evan went on in his presentation to describe how he wants Twitter to be useful even to people who don’t tweet. There couldn’t be a much clearer indication that they’re wanting to move away from “social”.
I think this indicates, for the first time in the past four years, that they’re publicly trying to move Twitter into a position where users expect things in their timeline that don’t call for a response, making business oriented Twitter accounts less about interacting with your customers and moving more towards pushing PR and marketing messages to a willing audience. Of course, Twitter aren’t being that overt about it; they’re saying it’s for “news”. But they would.
It’s almost a given; for Twitter to roll out an advertising sales platform that users are going to accept the users will need to accept those sorts of breaks in their timeline without complaining. At the moment it’s uncertain if they would. If you suggest twitter adverts most of the feedback is understandably negative. This is just the first of many necessary steps.
Just briefly it’s also worth noting another sideline announcement that Twitter made yesterday – that the new web client is built entirely on top of their existing public APIs. That means that there is no longer any difference between what is displayed on Twitter.com as there is in Twitter for iPhone, Tweetdeck, Osfoora, Hootsuite, and so on. This is another key factor in rolling out an advertising platform. Ads won’t work if they’re only displayed on Twitter.com – the users would just shift to a client that doesn’t display them. With every client using the same APIs every client will get the same content. Should they ever start pushing out adverts that means every client will get the advert data. They’ll be inescapable.
More positively, it also means that there’s a level playing field between Twitter and 3rd party app developers. There’s no reason why a talented team of developers couldn’t build an even better Twitter client than Twitter’s own official offerings.
Twifficiency, OAuth and You
Aug 18th
Yesterday there was a little Twitter drama centred around a site created by a young web developer who goes by the name of @jamescun. He built a website that rates your efficiency on the social network Twitter. It’s one of those “Who’s the best at Twitter?” things that people want to be top of without really understanding, or caring, what that might actually mean. Twifficiency.com is a measure of ‘mechanical efficieny’, a ratio of how much you point in compared to how much you take out. This has the advantage that it’s not a simple measure of how much you tweet or how many followers you have, so, in theory at least, anyone can “win”. That’s great. It’s a good way to get people interested.
Following it’s wildfire-like spread across the UK side of Twitter many people came out in praise of it. As I understand it James has received web development job offers on the back of it. That’s pretty awesome. James clearly has some coding ability. What he did was admirable in many respects. Getting your website trending is the ambition of many Twitter API users.
However, that’s not to say what he did was right. The problem with the site, and the reason it spread so quickly, was that it tweeted from your account when you authorised it using Twitter’s OAuth API without telling you it was going to. That’s essentially spamming the user’s timeline. What’s more, it’s against the API terms and conditions (Part 2, “Principles”; Section 1 “Don’t surprise the user”; Point b “get the user’s permission before sending Tweets”). Had Twitter wanted to they could have revoked James’ API key on that point. James went on to correct that later in the day.
This highlights, for me, a basic problem with OAuth, and in particular the user’s understanding of what it means. Once you authorise a service using Twitter’s OAuth provider, the owner of that service has pretty much free reign over your Twitter account. The service can post tweets. The service can send Direct Messages. The service can add and remove followers. All this can be done relatively easily, and at any time after you authorise the service. Twitter’s OAuth tokens never expire. Did you sign up to a service a year ago, and forget about it? There is nothing stopping that service spamming your followers today except the owner’s morals.
To that end I tweeted that people who have used Twifficiency.com should pop into their Twitter.com account settings and revoke it’s access (my tweet). I think it hit a nerve. It was retweeted more than 600 times. It drew some attention from James’ admirers too. I’ve had lots of replies telling me that what James made wasn’t any danger, it wasn’t spamming, and that it’s perfectly safe. I know all that. I wasn’t accusing James of any nefarious intent. I was simply advocating good Twitter security – don’t give anyone access to your account without good reason, and if you do in order to give something a try, remember to revoke that access afterwards. Otherwise you might inadvertently hand your account over to someone who’s less upstanding than James.
As for my Twifficiency rating, it stands at 33%. To be honest, I would have thought it’d have been much, much lower.
The Twitter Effect
Aug 2nd
Earlier today I ran out of milk. It happens all too frequently. It’s not very exciting. But that doesn’t stop me tweeting about it. Anyone who follows me on Twitter will be acutely aware of the fact I tweet some incredibly inane rubbish.
Today though, someone was actually reading me tweets. Cravendale Milk, or more likely someone from their internet/PR/social media provider, has clearly set up some sort of listener to watch for people who mention milk, and then they reply with a little advert for Cravendale and a link to an online discount voucher. It’s quite an obvious thing to do but it’s fun and entertaining at the same time. I don’t imagine it requires much capital investment either.
As I was amused by the fact they replied I posted again, this time to tell my followers what they’d done. And that’s when Twitter’s networking effect started to become quite noticeable. There were retweets. And retweets of retweets. And retweets of retweets of retwe.. well, you get the idea. I thought it might be fun to do a little investigation to see just how far Cravendale’s offer of a 50p discount voucher might have spread.
The original message was reposted by 6 people, and then subsequently retweeted by a further 4 people. Adding together the number of follower than I have, plus the 6 people who retweeted me, and the 4 who retweeted from them, gets a total reach of 23,712 Twitter users. This ignores the number of people not following those 10 accounts directly but who would still have seen the tweet on Twitter lists that follow any of the accounts involved.
Of course, that doesn’t mean nearly 24,000 people saw Cravendale’s brand. A percentage of Twitter accounts are dead, some others aren’t viewed regularly, some people will have ignored the tweet, and so on. Conservatively though, it’s not unreasonable to think that Cravendale’s PR team managed to get their message out to several thousand people this morning for the cost of watching Twitter and replying to someone positively (and a 50p voucher).
I still need to go out and buy milk though. Someone should invent a way of delivering it by the internet.
5 (and a bit) Interesting Ideas
Aug 1st
There are millions of completely crazy ideas around the internet. Sites that you see and think “Really? REALLY?”. I’m not going to name names but I see them almost every day. Frequently they fade away without a trace, occasionally they grow to be incredibly successful, and I’m usually left scratching my head thinking “What did I miss?”.
What’s out there at the moment that’s piquing my interest?
Much as I hate blog posts that consist of lists, I’m going to sacrifice that notion to bring some of the ideas I’ve seen recently that I believe are doing great things (or will be doing great things in the future);
GetGlue – http://getglue.com/ – It’s check ins for media. Rather than checking in where you are, you check in what you’re consuming (tv, dvd, books, etc). It’s a great idea. The statistical profile data of who is doing what and how that relates to what other people are doing will be fascinating. There are rivals (Philo for example), but GetGlue seems to be the best one at the moment.
Zong – http://www.zong.com/ – It’s kind of like Paypal for mobile phones. To buy something you just use your phone number. I’ve been talking about the idea of a “walled garden” for mobile sites for a little while now, and I think this sort of service is exactly the kind of thing that needs to be in place before that can happen seriously. There’s another option in the form of Vento, but Zong appears to be a more mature product at this stage.
PlacePop – http://www.placepop.com/ – Another Foursquare “rival”, PlacePop have taken the idea of GPS enabled check in services and applied it more directly to businesses using a “virtual loyalty card” idea. If they can get traction from retailers I imagine they’ll be huge. Another FourSquare alternative that’s looking interesting is SCVNGR ( http://www.scvngr.com/ ). They’re taking the check in model and applying it to gaming allowing users to build games on top of their service rather like geocaching. It’s hard to see SCVNGR failing considering they’re backed by Google. I fully expect to see SCVNGR games appearing in Google Maps and Earth soon.
Hunch – http://hunch.com/ – Hunch takes a list of the things that you like and builds a “taste profile” of you enabling it to recommend other things that you might like too. It’s quite an obvious idea but the graph technology behind the site makes things quite exciting. “Taste engineering” seems to be something that’s cropping up a lot recently. Local start-up and Difference Engine veterans wishli.st ( http://wishli.st ) have something similar running as a beta.
Jumo – http://www.jumo.com/ – Jumo is a site that will, once it launches, aim to bring together volunteers with volunteer organisations. I don’t yet know much about how they’re planning to do it, but it’s definitely worth watching as it’s been founded by Chris Hughes. Hughes was a roommate of Mark Zuckerberg and a co-founder of Facebook and the brains behind MyBarackObama.com, Barack Obama’s online presence during the 2008 presidential election campaign.
Canv.as – http://canv.as/ – This one is a complete mystery. There are no clues to what it is, or will become on the site whatsoever. Like Jumo though, it’s of interest because of the person behind it – Canv.as is the brainchild of Chris “Moot” Poole who founded 4Chan (the site that gave the internet “rickrolling” amongst other things). Poole is clearly good at community building, so I’m keen to see what comes from his next venture.
What the user sees first.
Jul 29th
Imagine the scenario. Someone out there is using Google. They’re looking for a company. They want to buy what you sell. Have you optimised your website for the key words they’re looking for? For the sake of argument, let’s assume you have. So your business is in the first few results. That’s brilliant. Except for some reason you’re not getting their clicks.
All this happens before the user has even visited your site. Before they’ve seen your fabulous design, read your professionally written content, browsed through your product pages with the wonderful photographs and seen your temptingly competitive prices. All this has happened on Google’s search results page. You aren’t helping your business get sales if your business listing in Google doesn’t look as good as it possibly could.
This is best illustrated with a couple of examples.
What the user sees first when they’re searching for a company to buy from is something like this:
Nothing in Abacus Media‘s listing tells you very much about the business. While Abacus have a great looking website that clearly sets out what they do, a search engine user would have to click on their rather obscure and wordy listing to read that content. Few people will – they’ll move down to the next listing to see if that better matches what they’re looking for. Compare Abacus Media’s listing to another Newcastle company:
This listing is for Union Room, a web development company. The difference is immediately apparent. Union Room’s listing clearly defines what they do, where they are, and even how to contact them. It’s a better description of the company by far.
Union Room have achieved this by making use of the META “description” tag in their page content. In contrast, Abacus Media have chosen to omit the tag so Google’s spider has had to lift the text for it’s listing page from the page content of the website. In situ the sentence might be fine but in Google it doesn’t work.
The description tag is the only META tag that Google pay any attention to these days as the others were abused by spammers in the early days of the internet. Keywords, authors, and so became worthless as measures of a page’s content so one by one Google started to ignore what they contained. Recently though there have been moves inside Google to start using page content in search listings again, just not META tags.
“Rich snippets” are well defined blocks of HTML content that Google look for in your website content. They’re a variation of a technology called “microformats”. What a microformat does is enable a machine such as Google’s search spider to understand what the content of a page really means. You, or more likely your web developer, designs the code for the page in such a way that when Google’s spider encounters it it’ll be understood better, and consequently displayed in Google’s listings in a much better form. Currently Google understands 4 different types of snippet; “reviews”, “people”, “products”, and “businesses”. You can find out how to use each one here: http://knol.google.com/k/google-rich-snippets-tips-and-tricks. If you decide to implement any of them in your website you can test that Google understand your code using the new Rich Snippet tool in Google’s webmaster suite: http://www.google.com/webmasters/tools/richsnippets.
For search engine traffic the face of your business is what they see in Google’s listing page. Make sure it’s not ugly.
Can SEO analytics backfire?
Jul 27th
For those readers of this blog who don’t happen to read the Boston Globe I’d like to share a fascinating link: http://www.boston.com/bostonglobe/ideas/articles/2010/07/11/how_facts_backfire/?page=full
The article is about how readers of political articles, when faced with the corrected facts about the article they’ve read, actually become more adamant and entrenched in their misinformed views rather than less as one might expect. Far from admitting defeat, they insist they’re right. Human nature makes us incredibly unwilling to admit we’re wrong.
The problem is known as “backfire”. No one yet knows why backfire happens. It’s pretty strange. And, as with most of the articles and stories I read, I got to wondering how this might apply to web development.
Anyone who’s been in a client-facing role in web design or development, or any other creative industry I imagine, will have met at least one client who is insistent that they know best because they’ve been misinformed by an article, a friend or a previous developer. It’s especially prevalent in the field of Search Engine Optimisation. There are many, many myths about what effects change in your website’s ranking, and persuading a client that what they think is good for their site is actually quite harmful can be a considerable challenge.
Presenting a clear and concise set of web and search analytics results that show how a change has made a positive improvement to the website, either in terms of search traffic, conversion ratio, or customer feedback doesn’t always lead to the client agreeing that the change should stay. The “knowledge” that the client is working from trumps the numbers in front of them in apparently indisputable black and white.
The answers to the issue of client knowledge backfire are given in the article – the direct, blunt approach of flatly refusing to accept that the client is right (because they’re not) while repeating the evidence to them might get through. Alternatively, complimenting the site and highlighting the positive aspects before presenting the aspect that’s backfiring might work instead. It depends on the client. What’s clear though, from my experience and from the literature about the problem of backfire in other aspects of life, is that simply giving a client the facts and expecting them to come to the ‘obvious’ conclusion isn’t always going to work.
