I couldn’t find anyone to talk with about webcomic advertising on Twitter…so I went out and did my own. Here’s how it went.
(This is part of the “how do I webcomic?” series, with useful information on all kinds of comicking-related topics.)
Done a couple months before the Leif & Thorn Volume 3 Kickstarter, because I wanted to see if it was a promising way to advertise the book.
Not going to describe all the step-by-step setup, because there are a lot of generic “how to run a Twitter ad campaign” guides out there already. But I’ll start with some general notes about Stuff I Did, then move on to Useful Tips I Learned.
Stuff I Did
Some basic background:
When you set up a campaign, you can have it include “organic tweets” — the normal ones you post, visible to your followers — or “promotional-only” tweets, which you make during the setup process.
(Promotional-only tweets can still get organic views, because once the campaign starts, some of the users who see them as promotions will like/retweet them. Then they show up naturally in the feeds of those users’ followers.)
A promotional tweet can be set up like any other, or it can include a Twitter Card, which, in turn, includes an image. (Or video, but for this campaign I didn’t make any of those.) The image size needs to have a ratio of 1.91:1, so unless you make art with those exact dimensions upfront, it’s going to need some cropping.
So here’s a couple of the images I made for this campaign:
All of them were cropped from specific Leif & Thorn comics — aiming for scenes that are evocative or intriguing, even for a non-reader who’s seeing them out of context.
(Some were my own ideas; others were suggested when I asked readers, “if you were trying to get someone hooked on Leif & Thorn, but you could only show them a single strip, which one would you pick?“)
For comparison, the full version of that first strip. And here’s a version with the cropped-out parts shaded over:
And, listen, I can hear you thinking: this is a bit of a cheat, right? This is a bit clickbaity. This is the “makes you think Twitter did a bad auto-cropping job, so you click to see the full image, and instead it takes you to the advertiser’s website” trick.
…okay, it kind of is. But hear me out.
Your media gets added to a Card, which includes a short tagline — such as “Buy Now” — and a URL that the whole thing links to. Here’s two cards using the same image, but different taglines (under the image):
That preview only shows the domain name, but when you click the actual card, it doesn’t take you to the Leif & Thorn main page — it takes you directly to that specific strip.
(And I wouldn’t use the same strategy for promoting the Kickstarter, because there’s no equivalent of that kind of cropped image that really fits.)
So it’s not a bait-and-switch, and it’s not the “trick you into clicking through five screens of ads to reach a not-worth-the-buildup punchline” theory of advertising. It takes you to the thing. The punchline’s right there. If you like it enough, maybe you keep going? It’s up to you.
As for how well it worked — I’ll circle back to that.
First, your Cards need to be turned into actual Tweets.
You can make as many tweets as you want with the same card. (Later, if you change something about the card, all those tweets will be retconned to have the new version. I don’t recommend that, though. It’ll screw up your data.)
This is great if you’re the kind of indecisive person who comes up with 5 slightly-different ways to word everything, and never knows which one will land the best. You can just do all of them, and find out!
So here’s two tweets, same card, where I field-tested slightly different captions:
Note: The first version up there has more likes and retweets. it also has fewer views, so it’s not just that it got more exposure than the second version. But if you look at more behind-the-scenes stats — the click-through rate on the first one is just 3.1%, and on the second, it’s a whole 4.7%.
People say they like the first one, which is nice and all, but they actually click the second one. And my goal here is getting people to the comic.
So: I made a bunch of images. I put them in cards. I put those in tweets. In the first week or so of the campaign, I kept an eye on the stats, and culled a few individual tweets with the absolute lowest click-through rates.
Then I took a breath, backed off, and let Twitter do what it wanted until the budget ran out.
Useful Tips I Learned
In no particular order, some things I had to figure out along the way.
1) Some of your tweets will never end up Approved For Use In Advertising
I don’t know what mystical human/algorithm tag team decides which tweets are ad-ready, but it’s a finicky one.
Most of my “promotional-only” tweets were auto-approved — or at least, approved so fast they were available by the time I finished the campaign setup. The organic tweets didn’t get approved until after I actually started the campaign. And…one from each group never got approved at all.
Non-approved ones show up as “Halted” in your campaign metrics (as opposed to “Paused”, for the ones you paused yourself).
What happens if they eventually get rejected? No idea. Is the long-term halt actually a soft rejection they’re not telling you about? It seems pretty likely. How do you appeal to Twitter to (re-)evaluate a halted tweet? Pretty sure you don’t.
In the long run, this didn’t give me any trouble, because I had more than a dozen other ads in play. But if I’d been relying on One Perfect Ad, and that was the one that went to processing jail, I’d be sunk. Especially if it was an ad for something time-sensitive (like for a Kickstarter).
So make lots of different ad tweets! And don’t panic if a few are Halted forever. It’s not you, it’s the algorithm.
2) Don’t stress too hard about making targeted ad groups
Ad groups are just groups of viewers with specific characteristics, each of which your campaign tracks separately, so you can see how they compare.
My “default” ad group only specified “English-speakers” (because there’s no point advertising a comic to people who don’t use the language it’s written in).
Early on I tried to pair that with a Highly Targeted ad group. Multiple relevant interests! Same age group as my existing readership! Lookalike audiences for the followers of creators I admire!
…it did miserably. Fewer impressions than the default group, lower rate of link clicks for the people who saw it, twice the cost for the ones who did click.
I backed off, re-evaluated, and tried a group that was just “English-speakers, ages 18-49.” Still really broad, right? And it trims off the age groups that are less likely to appreciate my jokes in the first place. Should be good.
The more-limited group had a sliiiightly higher link-click rate (5.21% over 4.56%), and each click was only two-thirds the cost. Looks good!
And yet — it only got 18% of the campaign spending. Because I also told Twitter “try to spend $2 per day”, and apparently there were so few people in the better-performing audience that it left most of the budget unused.
…so Twitter spent the rest of the money getting me almost 100,000 clicks from the general audience, which is nothing to say boo to.
If your budget is low and your schedule is long, it might be worth the effort to refine your ad groups so they get more limited results. But if not, leave yourself open to a general-audience ad group.
Although “general” is kind of a misnomer, because:
3) Twitter will automatically do better selecting/refining/targeting than you would anyway
When a campaign starts, yeah, it’ll only be restricted by the limits you give it. But as it runs, it’ll optimize itself using its own generated data.
Here’s a graph with the number of clicks my campaign got per week:
First week removed because I was still manually tinkering with the parameters; last week removed because it only included 3 days’ worth of budget.
So the way it goes up throughout July, and then takes a flying leap to stay high for all of August? That’s not anything I did. That’s just The Algorithm, retargeting to “more people who are similar to the people who’ve clicked on these ads already.”
Twitter’s campaign dashboard has other graphs, and all of them show the same pattern. Views take a leap. Cost-per-click takes a dive. The click-through rate has a subtler change, but every week pre-August brings the overall average down, while every August week pulls it up.
Okay, enough encouraging points, here’s a more sobering one:
4) Don’t do this unless you’re prepared to block, hide, and report
Unlike with traditional banner ads, users can leave comments on your promotional tweets!
Unfortunately: they will be Twitter comments.
I did get lots of likes, and a fair amount of friendly retweets, on all the promotional tweets. Your work will be exposed to plenty of nice and generous people. The comments, though? An exhausting mix of “your art sucks and you suck for making it,” “I’m trying to piggyback on your promoted tweet with an ad for my unrelated product,” and “masks are a scam.”
So you have to look inside your heart and decide, honestly, can you shrug that off? The insults especially. Can you hit Hide Comment and Block User and not give them another thought?
Or is it going to get to you, get inside your head, crush your mood and genuinely ruin your day when some internet stranger announces that he hates your comic?
And I say “when”, because it happens to everyone. It’s not personal; it’s nothing to do with how good your work actually is. You could literally be Vincent Van Gogh and there would be people on Twitter lining up to tell you that your art sucks.
But impersonal or not, if it’s going to ruin your day anyway…don’t give them the chance. You deserve better. Find other ways to share your comic with the world.
The Takeaway (so, how well did it work?)
Here are the past few months’ worth of Leif & Thorn stats, week-by-week, with the period of the advertising campaign in the darker blue. (Stats courtesy of the Jetpack plugin.)
Overall site traffic: definitely up! Every week after the campaign started had more hits than any of the weeks before, except that one fluke where I had a shockingly good week mid-April.
(Not sure what to make of the detail that there’s not a further mid-campaign jump in hits, to match the mid-campaign jump in clicks. Possibly I got a bounce in June from something else, and that was tapering off just as the Twitter clicks started to ramp up.)
Traffic boost compared to other ad services I’ve tried: much, much better. At the end I was getting 3,000+ clicks for my daily budget of $2. Take a site like comicad.net — listen, I’m using it, because it has advantages in other areas (see the earlier section about Twitter comments), but you’re lucky to get 30 clicks there for the same price.
Traffic to the individual strips that got linked: massively up! Not counting the main page, they’re all the most-visited pages during the past month.
Traffic from people who got hooked by those strips and decided to keep reading: harder to nail down, but promising. In the past month, several thousand people landed on the Kale strip from the examples above; several hundred clicked forward to the next strip; and several dozen kept reading for at least a week.
Traffic from people who landed on the individual strips, were intrigued but wanted the backstory, clicked “First”, and went forward from there: …no idea. Wish I knew, but at this point, none of my site analytics are detailed enough to say.
Traffic from people who will stick around and keep reading future strips, even while the ads have stopped bringing new people in: also no idea. And the only way to get the data is to wait and see.
But overall? Happy with the results so far, feeling good about the future, plan to try doing this again.
If you’ve ever advertised on Twitter — for a webcomic or anything else — drop a comment and tell us about the experience! And if you have questions about some area I didn’t cover in the post: please don’t hesitate to ask.