For as long as I can remember I've always completely ignored Twitter ads. It got to a point where I didn't even notice them anymore. Now, not only do I notice them more, I actually pay attention to see how they're crafted and what they're trying to achieve. What changed? We put together some promoted campaigns on Twitter over the last month. Having been on the receiving end, and now on the 'giving' end of Twitter ads I can speak from both perspectives. This process got us thinking quite seriously about what makes a good campaign and the different steps involved to ensure that any content served is relevant, useful and not spam. With this in mind we've put together a 'what we've learned' post. Hopefully you'll find it either useful, or interesting. So please tell us if you do, or don't, or have any other opinions at all.
Let's get started
The moment you hit 'Go' on your Twitter Promoted Account campaign you'll see the available budget counter start to rapidly tick downwards. The first thought that goes through your head is: Wow! This costs a lot of money. And it's entirely possible to run through your entire budget in a day if you put your mind to it. So now you might be thinking that Twitter advertising is only for those with deep pockets. Well, yes and no. As with so many things it depends on what you want to achieve and how much you're comfortable spending. You can still get decent results on a small budget and, arguably more importantly, learn a lot about yourself on Twitter. To gauge whether this is cash well spent, or not, Twitter provides two ROI metrics on the advertising platform. Cost per engagement (CPE) and Cost per follow (CPF). CPF is the measurement for their Promoted Accounts product and CPE is the measurement for their Promoted Tweets product. Something that Twitter stresses when they're making their sales pitch is that they don't charge for impressions and only for actions. For Promoted Accounts an action could either be a follow or a click on your profile. For Promoted Tweets it could be a click on a tweet (to view a pic), a reply or a RT. You'll need to decide how much an engagement or a follow means to you monetarily. If you're happy spending £3 for a follow so be it. Although this won't guarantee you any sort of quality control as compared to spending £1 per follow. What will guarantee quality is having well defined lists, intelligent bidding, being responsive and adaptable with your campaign and allowing Twitter's Implicit campaign the time to do its work.
There are two types of campaigns you can run for the Promoted Account activity. One is the 'Implicit' campaign. When setting up this campaign you just leave the keywords field blank and Twitter chooses whom best to target. To start with Twitter randomly promotes the account across a widespread set of users. As followers grow, Twitter uses this data to narrow down the sorts of users to display your account to based on topics and interests that are better performing. Although the Implicit campaign can be slow to take off it gets better and provides real value of money, and quality, as followers pick up and the data Twitter has to work with increases. The other is the 'Explicit' campaign. Here you define the keywords to be targeted. It's a good idea to have a few of these keyword lists. This allows you to tailor each one with a different subset of keywords to help monitor performance.
Getting the best out of your campaigns
Focus on the lists. Lists are at the heart of all Twitter advertising activity, whether it's for Promoted Accounts or Promoted Tweets. Well thought out, differentiated lists work really nicely, allowing a deep level of performance monitoring and aiding a flexible approach. Setting up lists based on different interests, or topics, will enable you to monitor the performance of an individual campaign, so you can either tweak a list that isn't performing well, stop using a particular list, or put a new one into play to get more traction where necessary. Apart from using keywords in lists it's a good idea to use relevant Twitter @handles as well. The idea being that your account will show up not just when someone searches for one of the keywords/terms, but also when they search for one of the tweeps on your list.
How to – Copy your list easily
When you’re setting up your lists you might possibly do so in Excel. Twitter, however, asks you enter keywords as comma separated terms. Bloody hell! Why can’t I just convert my Excel sheet to a CSV file and upload it? Twitter doesn’t allow this for their own reasons unfortunately, which they feel are valid enough I’m sure.I’m not going to spend 30 minutes copying this list to a text file and then adding the commas manually. What do I do now? From an Excel sheet: In the column beside the keywords (Column A) place a single comma (Column B) > then Select columns A and B > paste into Notepad (Windows), or TextEdit (Mac) > select all, and copy and paste this raw text into the Keywords field > hit enter.
Voila, you’re all done!
Why I would spend the cash: Analytics
Possibly the best thing about running an ad campaign on Twitter is getting access to the analytics. I’d personally say that it is worth spending even £1,000 just to be able to study these.
Admittedly the analytics feel like they’re in Beta at the moment. They’re slightly better than basic and it’s a given that Twitter will start rolling out more insights to this platform soon.
One of the most interesting set of stats for me was the Timeline. It shows you the level of engagement over a set period (Follows, Unfollows) and also how individual tweets have been performing. This is a great analytical tool to gauge not just what your followers engage with, but also tone of voice, time of tweet and the type of engagement. It is really insightful to see how certain tweets will elicit certain responses.
The stats for Followers are broken down in different categories: Interests, Location and Gender. There could be a lot more done here to make these more useful though.
(There’s a small list below of what else I’d like to see from Twitter Analytics. All other ideas are more than welcome.)
The interests are partly useful because you get a sense of how to tweak your campaign’s keyword lists to target better. As you click through each of the top interests you can also see where followers are based, what gender they are, and who else they might follow. This is pretty basic really.
Unfortunately the list of Interests and the accounts that ‘Your followers also follow’ are fixed. They display just the top five (interests) or 10 (also following) percentage values.
Location isn’t that useful if you’re outside the USA at the moment. Given that Twitter only allows ‘metro level tracking’ in the USA you can see that your followers are in the UK and that a small proportion of them might be in London, Manchester, Birmingham or Glasgow.
Granted that this data is partly dependent on people geo-locating their tweets, but there must be ways to break down the location data to display it more effectively.
Any thoughts on this would be very welcome.
Gender is semi-interesting from a targeting perspective. And you can see how wildly the gender balance swings on certain interests.
This is an initial list of what else would be great to see on Twitter analytics.
The ability to see how your followers’ interests and the other accounts they follow change as you click through each variable.
For example, if I clicked on a certain interest, apart from seeing the gender breakdown, it would also be great to see what other interests that subset of followers have and who else they follow. Now that would be some serious insight.
There also needs to be a different type of visualisation and data set for global and local, or regional campaigns. The way a global or a local campaign are run are so different in nature it seems unfair to lump them with the same sort of dataviz and provide the same stats for each one.
Having access to slightly improved analytics would provide some really rich insight into follower patterns. It would also allow for some really niche targeting, rather than Twitter’s ‘specifically scattered’ approach that we’re stuck with at the moment.
Now, although Twitter claims that it shows results in real-time it can take up to three days for the campaign budget to calibrate and stabilise to show the exact spend and CPF/CPE.
For example, you’ll see that a particular campaign has overrun the set budget and is at –£3.71, but check back three days later and the balance on the campaign is actually +£0.28.
A more responsive spend calculation would be useful, especially for advertisers with smaller budgets.
From an analytics standpoint it can take up to a week to show your new followers on the Followers graph. You can see the numbers ratchet up on your Twitter page, but looking at the graph you’ll be excused for thinking that performance has actually been really slow.
But we did appreciate the relatively real-time nature of the platform. It allows you to have a general strategy with a lot of flexibility to get the best out of campaigns that are performing well, while winding down or modifying others that aren’t.
You can quickly move around budgets, change daily spend caps, increase or decrease bids, add or remove keywords, and add new campaigns to see what kind of difference any of these variables might make.
Mainly, you can learn enough in an hour or two to effectively manage time and budget to get the best out of any campaign. For high volume, fast moving campaigns I’d say that you could learn enough in about 20 minutes to stay with your current strategy or tweak as necessary.
You can also tweak bids at certain times of day to see how your campaign performs ‘out of hours’. You’ll be surprised to see how you can generate a high number of impressions, and hence conversions, for very low bids at different times. Especially during times that others might consider to be ‘dead’ and consequently either pause or stop their campaigns.
The Twitter advertising platform is changing constantly and new features are being added on a monthly basis. There are pros and cons as I’ve outlined.
More excitingly, there will some serious opportunities for real-time promotion on the platform. Especially if you’re able to move fast, willing to be flexible and can adapt at internet speed.
Also, I think it’ll get cheaper sooner rather than later (economies of scale).
It’s been a great learning experience running these campaigns and playing with the platform. I appreciated the seriously granular control over campaigns it allows. I learnt a lot through the analytics, but was left feeling that I could have learnt so much more. Hopefully this will happen sooner rather than later.
In the end, if you want to get your follower numbers from 0 to 60 in the equivalent of 3.5 seconds, or run reactive, real-time, of-the-moment campaigns this is a very good way to go about it.
P.S. There will be a separate post on Twitter’s Promoted Tweets product soon. We’re in the process of playing with it and want to collect some more data before we post.
P.P.S. We’ve also been running some experiments using Reddit Advertise and StumbleUpon Paid Discovery. These obviously work differently, but they get some very interesting results. There will be more on these two platforms in the next few weeks.