As marketers, we’re all too familiar with putting off executing on an idea.
There’s always a perfectly good reason for this: the end goal hasn’t quite crystallized yet. The resources and budget need to be better thought through. The analytics systems in place don’t seem up to par. …you know how the rest goes.
The underlying problem?
We badly want to “do it right” so much that we end up not doing it at all. Perfectionism paralysis at its finest – and ironically, the higher the stakes are, the harder it hits.
At Ahrefs, we’re definitely no stranger to this. But when your marketing and content strategies trend towards conducting multiple experiments, you pick up certain ways to trick your mind into letting go a little.
We do a ton of experiments. Some succeed and some fail. What’s important is that they get done.
We’ve learned lots along the way, through trial and even more error – and now, we’d love to share some key points of our journey with you.
Ready? Let’s go.
3 social media experiments and their results
Our CMO, Tim Soulo, believes that a huge chunk of marketing ROI simply cannot be measured. And things get done a lot faster when the people up top aren’t pressuring you to predict ROI, set specific numbers to hit and track data for every step of every process.
In fact, we often do things because our gut feeling tells us it’s the right thing to do, rather than because the numbers say so. In short, we believe that it’s better to do something than not do anything at all – simply because it doesn’t seem perfect. If we have a cool idea and some spare resources, we take action…and quickly.
Let’s dive into some examples that illustrate this. We’ll be exploring the thought processes behind three Twitter marketing experiments that have varying degrees of polish to them.
1. The 1k retweet growth hacking experiment
Starting off quick and easy: in February, we attempted to get 1k re-tweets on Twitter in order to loosen up a data limitation on one of our paid subscription plans.
Here’s a little secret: we were going to loosen up the limit either way, but we also wanted to draw some attention to that fact. Hey, no harm having extra toppings on our ice cream.
In terms of measuring success? We didn’t even bother.
This is a fantastic example of doing something because it feels right, not because we expected or were aiming for a specific outcome. It could just have easily turned into a monster of an experiment – calculating the rate of visibility spread, analyzing data on the users who retweeted, attempting to measure the value of publicity generated…so on and so forth.
Instead, we decided to simply jump into it, then collectively nodded when the experiment succeeded and called it a day.
The story: Our initial, “raw” tweet collected about 100 re-tweets (Dmitry doesn’t have a large following on Twitter.) To give it a little boost, we pushed it to our users via Intercom announcement and also shared it in our private Facebook Group. This boosted the retweet count to nearly 700.
Funnily enough, at this point our counter leaped another 500 re-tweets in no time at all to total almost 1.3k; we suspect that someone lost his or her patience and purchased re-tweets on a service like Fiverr. So we gave up on our goal of 1k re-tweets and rolled out the update – we’d reached our goal of more publicity for this plan upgrade, and that’s all we needed.
(Looking at the tweet now, we’re back to 800. It looks like Twitter deleted the bot accounts and removed all retweets from these bots.)
2. The recirculation experiment
Sometime in March this year, we decided to run another experiment. Granted, this one’s actually more of a best practice, or highly-adopted technique, in social media marketing: content recirculation.
We have tons of well-researched, in-depth blog posts that we go to great lengths to create. Most remain highly relevant months after their initial publish date, but traffic soon starts to dwindle when compared with newer posts – mostly due to an increasing lack of visibility as they get pushed further down our archives.
Solution? We decided to dust off our previously underutilized Twitter account (due to lack of resources on our end) and kick off a more regular content schedule. This included lightly repurposing our existing content into catchy tweets.
We love to make things easier, so we didn’t set any hard KPIs or goals. Our followers on Twitter are a cool bunch; engaged, receptive and fun. We just wanted to highlight cool stuff that they might have missed and see how things went from there.
Turns out, plenty of these tweets were really well-received:
Initially, we were simply aiming to see if there were any changes in the number of people who liked, retweeted or replied to our tweets. Then a peek at our Twitter analytics a week later showed these trends:
Above: red lines mark when we began the experiment. Engagement rate, link clicks, retweets and likes all showed a noticeable upward trend.
Pretty cool, we thought! The charts instantly confirmed what we were hoping to see: a definite boost in overall engagement and link clicks.
We’ll probably pop into Google Analytics in a couple of months’ time to confirm that our blog is indeed seeing increased traffic from Twitter. But in the meantime – success, and on to other things!
The custom Twitter header experiment
Now for a more complicated example – we went into this one with more planning in place.
A header image is the first thing you see at the top of any Twitter profile: 1500 x 500 pixels’ worth of it, to be exact.
Given their prime position and reach potential, we were seeing companies invest efforts into making custom twitter backgrounds for every occasion, from event announcements to highlighting important news, to more targeted offers and call-to-actions.
Here are a couple of examples:
It got us wondering – just how effective are these header images? Do many people see them? And if they do, do they actually act on them? Most of all: since creating cool images can be costly in terms of resources, does it make sense to start investing in them?
The concept was simple: use our Twitter header image to announce a flash giveaway, then monitor responses and engagement levels. We kept the mechanics easy. The first 10 people to send in an email with the message we specified would be granted a free month’s worth of access to Ahrefs.
We had ~19.3k followers at the time and were excited to see how this would translate into entries for our giveaway. Before we began, we checked where the Twitter header could be seen from:
Results page upon searching for “ahrefs” in Twitter’s search bar;
Ahrefs’ Twitter profile page;
On hover of a direct mention (@ahrefs) or on Ahrefs’ name on Twitter timeline (not applicable on mobile devices)
Pretty promising, right? We also came up with a couple of potential cons to the experiment:
A single person could abuse this pretty easily by sending in multiple entries to get many accounts;
Or, a person could spread the word to others through other (non-Twitter header) channels, effectively defeating the purpose of the experiment.
What we learned
We ran the header image on January 16th and reached our 10th entry on January 24th, over a week later.
Here’s the breakdown:
Overall, the response was curiously slow for an instant win giveaway like the one we were running. In fact, although we were hitting an average of ~45k impressions per day, we only saw our very first entry two full days after launching.
The lukewarm response continued until there was a spike in entries on the 24th. After some digging, we discovered two big occurrences on that day:
1. We experienced a power outage for a period of time. People were coming to our Twitter profile to send us direct messages and tweets asking about the downtime. Presumably, many spotted our header image along the way – one existing user sent in an entry;
2. Ahrefs picked up a direct mention and article plug from the @growthhackers Twitter account with 187k+ followers. We can assume that some of their followers decided to follow Ahrefs on Twitter (or visit our profile to learn more about us) after reading the article. These people would have been exposed to our Twitter header and we suspect this is the main reason why we had an influx of entries.
Going by the trend seen in the final spike in entries, we can conclude that new leads are probably generated from our content marketing efforts: Twitter users enjoy an article, decide to follow us on Twitter, see our header and decide to write in for a free account.
On the other hand, entries from existing Ahrefs users are likely a result of users visiting our profile to either tweet at us or send us direct messages – whether to lodge a complaint or ask a question. While on our profile, they may have come across the header and had a “why not?” moment.
In other words, without big external factors, our presence on Twitter seems to do an equally good job of engaging both new leads and existing users.
Overall, the response to our giveaway was poor until some ‘special circumstances’ occurred.
Our takeaway: customizing multiple high-quality header images and expecting huge results is out of the question. But considering that header images can be designed pretty quickly or outsourced for cheap, even a really low conversion rate might justify the cost of creation in order to complement an existing marketing strategy.
In short, we decided that investing our efforts in customized banners is a pretty low-priority effort. It may produce some results, but they aren’t likely to be exceptional. Experiment concluded!
Over to You
I hope the contrast in these stories help to highlight what we’ve learned:
The more lightweight you keep an idea, the quicker it gets executed and the faster you get a feel for whether or not you should continue down the same road.
Keep things simple! Run experiments for the sake of learning, even if you haven’t quite nailed down how to best track results.
The more effort you put into the planning and analytics stages, the more resources you’ll need to execute. It will always be “cheaper” to just throw things at the wall.
Put cool ideas on hold if you don’t have the resources available at the time. You never know when you can work on them, or if an alternative execution method will spring to mind.
I’d love to hear your thoughts – how do you strike a balance between shipping experiments while avoiding getting mired down by the technicalities involved? How long do you keep tracking your results, and how do you use this information moving forward? Are there any other tips, tricks or hacks you’d like to share?
Read more: feedproxy.google.com