Getting Started With Retargeting
Have you ever sat staring at a load of figures in Google Analytics or the Facebook ad manager and thought… ‘bloody hell. I suck at this.’
If you have, good news: so do we. And so does pretty much everyone else.
We’re a SaaS company. And we’re self-serve, which means we don’t have a phone team we can direct leads to. More to the point, we SHOULDN’T have a phone team. Our price point just doesn’t justify it.
(How many of you just thought ‘charge more, then’? It’s a valid argument and something we’ve thought about it, but this isn’t really the article to talk about why we went down this road. If I ever write one, I’ll link it.)
The general consensus is that you should have an ACV (that’s Annual Customer Value, kids) of at least $3000 to support a sales team. Even our top plan is just under $2000/yr, so we have to get people to sign up without endlessly badgering them down the phone #playinglifeonhardmode.
And so we’re learning paid ads.
And one of the things we’re learning as part of this is retargeting.
This is what we’ve learned from our first set of tests, so with any luck if you’re also getting started with retargeting, it’ll get you going faster.
We’re using Perfect Audience. I’d like to say this was down to a detailed analysis of the strengths and weaknesses of various platforms, but actually they were just the ones we created a test account with several months ago, so it was the path of least resistance.
This happens a lot. I’ve stopped feeling guilty.
Then we need to think about our goals. The obvious answer is that we’re after conversions – but a conversion isn’t just a conversion.
In fact, Perfect Audience give data on a whole range of different conversions your ads can achieve:
Click-Through Conversion: Someone saw the ad, clicked on the ad, and then converted on your website. Boom, baby!
View-Through Conversion: Someone saw the ad. They didn’t click on it, but at some point in the next 30 days they found your site and converted. It was the last retargeting ad they saw.
Assisted Conversion: Someone saw the ad. They might have clicked on it, might not – but either way they didn’t convert at the time. Between that point, they saw at least one other retargeting ad, which they also may or may not have clicked on.
To borrow some of Perfect Audience’s own examples:
Example One
- You are running Campaign A and Campaign B.
- A user sees (but does not click) on ad A.
- The user later sees (but does not click) ad B.
- The user then converts on your site.
Campaign B will be attributed with one View Through Conversion, and Campaign A will be attributed with an Assisted conversion.
Example Two
- You are running Campaign A and Campaign B.
- A user sees and clicks on ad A, but does not convert on your site.
- The user later sees (but does not click) ad B.
- The user then converts on your site.
Campaign B will be attributed with one View Through Conversion, and Campaign A will be attributed with an Assisted conversion.
Some of these seem like they’re stretching the definition of ‘conversion’ a bit. Someone saw (and by ‘saw’ we mean ‘it was also on the website they visited and they may or may not have noticed’) your ad, didn’t interact with it, but bought in the next 30 days.
This feels kinda like hiring a sales consultant, having them do nothing but sit in your office eating Cheetos and watching YouTube, and then claiming credit for all the sales you made that month.
But the fact is, conversion is complicated. It’s not just a case of see-ad-click-ad-buy any more, especially in SaaS where your product involves a hefty financial commitment.
Conversion journeys these days can involve any combination of organic search, blog posts, ads on multiple platforms, landing pages, social comments and more. They can all play a part in making the final sale. People need to see you more than once, and just seeing an ad – even if they don’t click on it – builds trust that you’re a real company they can do business with, not just some rando off the interwebs.
And that means while last-attribution – knowing which part of your marketing actually kicked your prospect over the line – is worth knowing, it’s important to know how they got to that point as well.
If you’re into football, you can think of it like a set-up to a goal. (I’m not into football – much more a tennis boy – but I’m also never going to pass up a good metaphor.)
In the first goal shown in that video, Sane was the one who scored. He’ll get the little number next to his name on the match reports and he’ll get the glory. He was the last-attribution. But he never would have got there if De Bruyne hadn’t managed to get him the ball from halfway up the field. And De Bruyne could never have done that if he hadn’t been passed the ball earlier, and so it goes.
Your ads aren’t just a set of goal-scorers. They’re a team that works together.
So view-through conversions and assisted conversions are important metrics.
That said, we do have to be careful in using them. We can say they played a role in the user’s conversion journey – but we can’t say how much. Maybe they would never have got there if they hadn’t seen our ad sitting like a beacon in the corner of their Yahoo Mail inbox, or maybe they would have found us anyway.
If you’ve ever felt like marketing SaaS is basically the 21st-Century equivalent of casting bones, I’m with you.
But this doesn’t mean we don’t have levers to pull. View-through conversions might not feel directly related to a sale but each one still means a sale happened somewhere along the line. We’ve got numbers, so by God we’re going to make them go up.
So we tried 12 ads in total – 3 different messages across 4 different sizes.
Our ads:
Ad 1 – SPEED: ‘Less than 3 seconds’ is a meme for people who care about page speed. It’s from a widely quoted study and is often treated as the benchmark: less than 3 seconds = good. More than 3 seconds = bad. With this ad, we’re trying to focus directly on a specific metric our audience cares about.
Ad 2 – BEST: Probably the most generic concept in all of SaaS marketing – and all other marketing, come to that. Everyone else sucks. We’re the best. Of course, when we use it, it’s true 😀
Ad 3 – CONVERSIONS: Maybe people don’t actually care about page speed so much as what they get from page speed – the extra conversions. We’ve had some great results comparing Convertri pages with equivalent pages in our own tests, so this line uses one of them.
We tested each of these in the following sizes:
- 728×90
- 468×60
- 160×600
- 300×250
I chose these because a bit of reading suggested they were the most popular sizes. There’s far more options, but some sizes won’t be displayed often because there’s just not many sites that have those banner sizes available. One of the things we discovered was that this may have been a lie.
We spent about $300 on these ads in total, over the course of 3 weeks. We didn’t attempt to make any adjustments to ad placement, they just went wherever Perfect Audience saw fit to display them. The target page was our main site: www.convertri.com
And we got these results:
| Angle | Size | Impressions | Clicks | CTR | CTC | VTC | AC | Spend |
| Speed | 160×600 | 1950 | 22 | 1.13% | 0 | 0 | 0 | $15.96 |
| Speed | 300×250 | 2913 | 21 | 0.72% | 1 | 1 | 0 | $25.34 |
| Speed | 468×60 | 206 | 4 | 1.94% | 0 | 0 | 0 | $1.50 |
| Speed | 728×90 | 2635 | 50 | 1.90% | 0 | 6 | 0 | $24.85 |
| Best | 160×600 | 1986 | 16 | 0.81% | 0 | 0 | 0 | $16.10 |
| Best | 300×250 | 2902 | 17 | 0.59% | 0 | 4 | 0 | $25.50 |
| Best | 468×60 | 213 | 2 | 0.94% | 0 | 0 | 0 | $1.51 |
| Best | 728×90 | 2741 | 36 | 1.31% | 0 | 4 | 0 | $25.53 |
| Conversions | 160×600 | 1996 | 8 | 0.40% | 0 | 0 | 0 | $15.91 |
| Conversions | 300×250 | 2907 | 31 | 1.07% | 0 | 0 | 0 | $25.35 |
| Conversions | 468×60 | 228 | 4 | 1.75% | 0 | 0 | 0 | $1.60 |
| Conversions | 728×90 | 2751 | 49 | 1.78% | 0 | 0 | 0 | $25.31 |
So… are these good?
Honestly, I have no idea, and I don’t intend to worry about it much. To get any kind of meaningful comparison we’d need to compare stats with other SaaS apps doing a similar job to us, and I don’t imagine they’re going to be inclined to share.
But let’s brush off my sadly-neglected maths degree and focus on what we can tell from these numbers.
First of all, sizes:
| Size | Impressions | Clicks | CTR | CTC | VTC | AC | Spend | CPC |
| 160×600 | 5932 | 46 | 0.78% | 0 | 0 | 0 | $47.97 | $1.04 |
| 300×250 | 8722 | 69 | 0.79% | 1 | 4 | 0 | $76.19 | $1.10 |
| 468×60 | 647 | 10 | 1.55% | 0 | 0 | 0 | $4.61 | $0.46 |
| 728×90 | 8127 | 135 | 1.66% | 0 | 10 | 0 | $75.69 | $0.56 |
728×90: Definitely the best one. High impressions and a high click-through rate. If we were only going to do one ad size, it’d definitely be this one.
300×250: High impressions but a far lower click-through rate means we’re paying over $1/click. That’s still not a massive amount of money, though, and this size did generate both some view-through conversions and our only click-through conversion.
468×60: Lol. You are not the popular ad size I was told you were. Good click-through rate, though. Cheap clicks mean this one would probably still be worth it if we were running ads at enough scale, but at our current spend I’m just not convinced it’s worth the extra design and setup effort. I don’t think I’ll bother with this size on the next run.
160×600: A reasonable amount of impressions, but a far lower click-through rate and 0 conversions – even view-through – suggest this one isn’t really worth it either. It seems strange that this ad should strike out on any kind of conversion given the fairly loose definition when the messaging was converting fine on other ad sizes. My current working theory is that the type of websites that have this size ad-space just don’t have our target audience. This feels like something worth experimenting with further, but not now.
Google’s AdSense guidelines suggest 336×280 and 300×600 are also popular sizes. 300×600 isn’t available through Perfect Audience, but 336×280 is, so we’ll add that to our sizes on the next run.
One thing I found interesting is that there’s 2 distinct bands in the click-through percentages. 468×60 and 728×90 both got significantly higher CTRs than the 300×250 or 160×600 banners. Maybe this means horizontal banners are more eye-grabbing. Maybe it means sites that tend to have banner space in that size are a better match for us. Maybe it’s just a coincidence, though the numbers suggest something is going on.
How about messaging?
This ended up being a bit of a complicated picture:
| Angle | Impressions | Clicks | CTR | CTC | VTC | AC | Spend | CPC |
| Speed | 7882 | 92 | 1.17% | 0 | 0 | 0 | $68.17 | $0.74 |
| Best | 7704 | 97 | 1.26% | 1 | 6 | 0 | $67.65 | $0.70 |
| Conversions | 7842 | 71 | 0.91% | 0 | 8 | 0 | $68.64 | $0.97 |
Ads 1 and 2 – SPEED and BEST – got a significantly better click-through rate.
Ads 2 and 3 – BEST and CONVERSION – got a significantly better conversion rate.
The decision of which copy to run with is pretty easy. Only Ad 2 got both clicks and conversions, so it seems by far the strongest performer. What’s more interesting is why the other ads didn’t perform at the same level.
The figures suggest that more people find ‘speed’ a more attractive hook than ‘better conversions’, which probably means there’s plenty of people out there who specifically want page speed – or at least, the claims made by the speed ad were more interesting.
But why no conversions, even view-through?
One possible reason is that at time of writing, our main sales page is not particularly fast to load. It’s also not particularly slow – which says quite a lot for Convertri’s optimisations on a single enormous page with 9 videos and an animated GIF – but if you click on an ad where you’ve just been promised ‘fast’ and you don’t get fast, you’re probably not going to think too much about the reasons.
We’re in the process of developing a more traditional website with a far smaller home page, which should help if this was having a negative effect. If that wasn’t the reason – and this is, of course, entirely speculation at the moment – then we’ll come up with another idea.
For now, we’re going to stick with the message that’s working, start trying out new designs, drop the 468×60 and 160×600 sizes and add 336×280.
In theory, that’ll help us push the numbers higher. Whether or not that makes more sales… well, we’ll get back to you 🙂