This is a guest post by Stevie Duffin-Lutgen, Marketing Analyst at Mobovida, a customer-driven, vertically integrated mobile accessory brand delivering fashion forward products direct to consumer. Check out our recent podcast interview with Edwin Choi, VP of Marketing at Mobovida.
If you’ve given up on display advertising as a source of profitable revenue, you’re not alone!
My predecessors in digital marketing at the heavily ROI-driven Mobovida didn’t have much success converting upper funnel banner clicks, instead wielding display for remarketing purposes only.
Mobovida owns CellularOutfitter, the largest online retailer of cell phone accessories. We sell a plethora of mobile accessories for just about every phone model, and wanted to know how we could revisit converting the cold display customer to reach a broader audience.
So when we heard that AdWords was going to start serving clicks on Gmail via Gmail Sponsored Promotions (GSP), we knew we had to jump on the opportunity. And I encourage anyone reading this to do the same!
Not only did the channel prove to be a valuable source of sales, but it allows the company to quickly test its Mobovida product line against fresh audiences. The data provided via quick-to-launch channels such as Gmail Ads has been invaluable for powering Mobovida’s agile product development model.
Although the channel started off losing $14 per $1 spent, we were able to methodically reach our tipping point ROAS (Return on Ad-Spend) of $2 per $1 after 13 rounds of testing.
Glad you asked!
If low-quality display clicks have ever given you nightmares, rest assured – surgically targeted Gmail ads with effective creative are not quite the same animal.
When Internet surfers are perusing their inboxes and see a banner for a Galaxy S7 Edge phone case (one of our top-selling categories), there’s always the possibility that they’re more annoyed with the banner interrupting their experience than they are interested in the offer (even if they actually need a case).
But GSP mobile ads are served up in the often retail-laden “Promotions” folder of Gmail user inboxes:
My drip e-mails from Nordstrom, ModCloth (can you guess my demographic?) and my other favorite shops are already in this folder, so when I access it I’m in a curious, if not full-on shopping, mood.
My job is to drive sustainable revenue at a 3:1 ROI target by harnessing Internet traffic. So when my team suggested working with Gmail ads – a branch of the Google Display pipe dream – I knew it was going to be a challenge.
But with some permission to adjust my ROI targets and experiment with a sizeable test budget, in one month I took our Gmail ad campaigns from 0-30% ROAS up to 3200%:
Here’s how it got done, in this order.
Display advertising is only as good as its targets, and one place where we struck gold in 2015 was Facebook, largely thanks to sweat equity and – you guessed it – rapidly testing endless permutations of targeting and creative.
Since Facebook ads are not triggered by search queries, we were alarmed at our ability to grow and scale the account after hammering out the right ad-demo-target formula, and truth be told it exhibited some game-changing potential:
I needed somewhere to start in Gmail and using my team’s experience in conquering Facebook was integral. I knew which products were converting for which demos and on which devices, which was a huge advantage! If you don’t have the luxury of previous team experience, consult your Google Analytics Audience -> Demographics tab for an expanded date range and look at your top contributing genders and age groups.
So I consulted some historical analytics and confirmed that older women using specific phone models converted best on our site. Thanks to Gmail campaign-level phone model targeting and ad group demo levers, I was able to carve out a solid beginning baseline upon which to test the various GSP ad types.
So I got my initial targets down; the stage was set. Then I was confronted with the myriad Gmail ad options laid out in the AdWords Gmail ad gallery:
The question was: Which to choose?
The answer? Choose them all as there is no right or wrong option.
Every ad type is going to start in the same way: a “collapsed” format at the top of your audience’s inbox. It’s going to have an e-mail subject line and description that must spawn an initial, curious click. Once clicked, your ad will be revealed in the “expanded” format, but wait! – you’ve already been charged.
A challenge for us was finding the ad type/copy/creative combo that drove enough interaction on the first, charged click, and also resulted in click-throughs to the website. You can monitor this metric in Columns -> Gmail Metrics -> Gmail Clicks to Website in AdWords for a more accurate picture of CTR and conversion rate.
Thanks to our vast, SKU-level analytics data, I was able to break each phone model down by its top products, so I knew which items I wanted to serve on each phone model’s campaign.
A good example is screen protectors for the Galaxy S5 campaign: we know that our S5 users love screen protectors and especially tempered glass, so we started with simple single promotion ads:
We experimented with similar visuals and language across the available ad types, but the return was not there. We spent over $600 to make $35 on the ad pictured above, so something wasn’t working.
I didn’t know what was wrong, but my hypothesis was that our current aggressive ad types might’ve been too abrasive for Gmail users, who are opening what looks like an e-mail.
While there’s no true A/B tool available within AdWords for display at the moment, I had to test a different ad vibe. I created a toned-down version of my sales pitch and tried for straightforward instead of sensational. However, I still didn’t see the return I wanted with screen protectors no matter the ad type or my copy.
Here’s when a search marketer bangs their head against the wall and cries out, “I am done with display!” But I’m lucky – I have a lot of products to test, and I was determined to test them until something worked. So I switched to phone cases.
My original poster boy for phone model-targeted campaign success was the Samsung Galaxy Note 4.
I started with a toned-down version of our single promotion ads to test the waters, paired with a top selling Note 4 case:
And all of a sudden I was breaking even – I spent $375 and made $377, but that pesky “Ad Copy CTR” was driving my costs up significantly.
Is it possible as a PPC marketer to see a 25, 35, 45% CTR and see red?
In Gmail it is.
Earlier, I warned against the dangers of the collapsed versus expanded GSP ad formats, because advertisers are charged on the first click. That first click opens the e-mail ad in all its full, expanded glory – it doesn’t give you an URL pointing to your website. Consequently, Gmail ads are technically charged per impression.
And our flashy, monster-percent-off e-mail ad subject lines were attracting way too much attention. I needed to reduce junk or purely curious clicks significantly. I brainstormed with another member of my team and it was clear that we needed to heavily qualify every incoming click.
As a possible solution, we tested product price in the e-mail subject line:
Our Ad Copy CTR was still hovering above 30%, and volume was building, but we were still breaking even.
Now, I don’t know about you, but I’m only willing to pay so much for a customer. I don’t want to pay for tons of clicks from accidental swipes or rubberneckers.
So when my tragically effective subject lines were burning holes in the company’s pocket, my co-manager suggested I turn on target CPA bidding at a realistic level, just to give it a whirl. If I could not sacrifice good old-fashioned compelling copy for lower costs, AdWords needed to give me predictably higher-quality clicks based on its learnings from my unprofitable, proliferating conversions.
Our costs went down overall, but so did our impressions and conversions, perhaps from a lack of sufficient conversions learnings since our volume-per-campaign was still low on average.
From inception, we’d seen a pretty steady 5% average conversion rate, and target CPA results did show conversion rate lift (scroll down to see charts), but since our impressions were down to boot, our Gmail Clicks to Website metric was looking especially anemic.
The next thought was: could I turn manual CPC bidding back on for low conversion campaigns, create highly compelling expanded ad copy, and increase ROI by getting visitors through my ad and letting the PDP landing page do the rest of the work?
How do you qualify a customer? You tell them what to expect.
We’d been doing that with our subject line copy by discussing price, but we’d been repeating it inside the expanded ad as well. Could it be that seeing price again was discouraging? Does our audience not want to see dollar signs? Do they want more information than we’re giving them and we’re wasting the expanded headline space with fear-inducing redundancy?
I decided that it could certainly be one or many of those things, so we tested a new ad format – what came to be called the “WasNow2b” format, in which we still list the “Was $X, Now $Y” in the subject line, but add fear- or doubt-reducing value propositions in the expanded ad:
We started to see days of high return interspersed with those of low return, with the Gmail Clicks to Website metric increasing.
This wasn’t too surprising (gaining some traction only to hit shaky ground) because admittedly, our mobile experience isn’t the most flawless, and we’re driving pure mobile traffic from these Gmail ads.
On our site, landing on a PDP can be a damning experience for the older demographic, since navigating to our many other products or categories can be cumbersome. And our site visitors like to see options, especially in a wealth of colors.
I bolstered our concerns using Hotjar visitor recordings, and witnessed many Gmail visitors landing on the PDP we provided, only to scroll around the site in what looked like a baffled mess.
We hypothesized that if we offered more case color options in our Note 4 ads, we might see lift.
The Gmail Catalog Ad is the Single Promotion Ad’s more versatile sibling. Their formats are comparable: both have a top image, headline, and description available in the expanded format, but the Catalog ad allows for more items to be featured beneath the main product image. If our Single Promotion ads had been gaining traction, my learnings would’ve been derived from not only their targeting successes and failures, but those of their formatting.
So – plain and simple – I chose to experiment with the Catalog ads to mitigate the strain of variability:
After running the Note 4 Catalog ads, we were closer to our 3:1 ROI target than ever before:
The good scientist keeps a thorough lab notebook, and so should you.
Maintaining strict documentation of our ad copy, targets, and results was paramount to success in Gmail, even though it took time to develop the format that worked for us (and it’s still evolving!). We tracked progress in a Google doc with tabs for new ad buildouts:
…and our results. Our “Results” tab listed the testing round, campaign, ad group, ad name, a link or screenshot to ad creative, KPI data per ad, and, most importantly, a column for hypotheses derived from the round and a column for next moves.
For each round, we paid special attention to the average conversion rate and average “true” CTR. Note that “true” CTR is not collapsed ad clicks divided by impressions, but expanded ad clicks – Gmail Clicks to Website – over collapsed ad, or subject line, clicks.
To see how this worked, take a look at the evolution of our conversion rates and true CTRs over the course of four rounds involving pivotal changes:
Across the first and second rounds in this series, we were funneling in target CPA bids across campaigns with relatively higher conversion counts, and still mostly using a mix of single promotion ads. You can see lift in conversion rate on average, but true CTR decreased.
From the second to third rounds, we’d begun to test more and more catalog ads, which boosted our true CTR as predicted, but did not result in a lift in our average conversion rates.
Moving into the fourth round, we’d scaled the use of our fear-reducing expanded ad copy across all campaigns and introduced target CPA bidding to campaigns with 20 or more conversions. This combination was powerful, illustrated by the jump in both true CTR and conversion rate.
None of our progress would have been possible without this weekly ritual of recording KPIs for each ad, followed by synthesizing new hypotheses.
As the conversions rolled in, I gradually migrated campaigns from Manual CPC to Target CPA, at about $2 above that campaign’s average CPA. This approach, combined with our winning ad and copy formats, was enough to garner sales, but there was more that could be done at the ad group level since I can control who sees my ads at the level of gender, age and parental status.
Over time we’d see trends evolve – like 35-year-old women who love to click a particular ad in a particular campaign, but rarely convert. To save on costs, I removed those demo groups. You can do this by clicking on your campaign in AdWords, navigating to the Display Network tab, selecting Age, Gender or Parental Status, and hitting the “enabled” button on your demo group:
Narrowing in on winning demo groups while diversifying our offerings in creative refreshes has become more important moving forward, since we’ve noticed that hitting one demo with one phone model with the same ad for more than two months slows impressions and exhausts sales. But if you’re willing to venture into the unknown with display ads, Gmail is a potentially lucrative (and far less painful!) place to start when compared with the broader display network.
Play to your strengths, rigorously control your testing and documentation of both the collapsed and expanded ads and targeting, mine your efforts for learnings from your analytics platforms, and fine-tune your targets using the Display Network tab in AdWords. You may find yourself with a new Google revenue channel, and no keyword bids are necessary!
Have you tried Gmail Ads? What’s been your experience, good or bad? Let us know in the comments below!
Check out another great Gmail Ads post right here on the Growth Everywhere blog.