How We Drove Awareness for Starbucks' CloudBar Pop-Ups at $0.01 per Engagement

By Kevin Veitia

Client

Starbucks

Brought in by

Crosby Steiner, through WeGrowth, Kevin's agency at the time

Markets

Nashville and Seattle

Channels

Facebook and Instagram Ads

Timeframe

Summer 2018

The challenge

In the summer of 2018, Starbucks was testing a Summer Series of new drinks at CloudBar pop-ups in two test markets, Nashville and Seattle. Crosby Steiner brought Kevin in through WeGrowth, the agency he ran at the time, with a simple question: can Facebook and Instagram ads get people to walk into a pop-up?

It was Crosby's project, and unusually free for a brand that size. There were no messaging guidelines and almost no corporate oversight. Everything reported to him.

It also broke the usual playbook. These weren't Manhattan or London. They were mid-sized cities, where tight geo-targeting can send CPMs through the roof. And there was nothing meaningful to optimize for online: no purchase, no sign-up, no click that proved anything. The goal was people in a store.

What we did

Tested geo-targeting models before scaling

The first job was finding the balance between targeting tight enough to reach people near a pop-up and broad enough to keep costs sane. We used the same method as on every account at the time: one control ad, run across competing targeting models, so the winning setup was clear before money went into creative.

Optimized for engagement, with a hard cost target

With no purchase or sign-up to optimize for, the campaigns optimized for engagement, and Starbucks' one standing instruction was to keep pushing cost per engagement down. It reached a penny.

Let organic content point the way

Organic posts helped show which drinks, visuals and messages people responded to, so paid budget went behind content that had already earned attention instead of untested guesses.

Ran in flights, so lift could be measured

Campaigns ran in flights: on for a set period, then off, so Starbucks could measure brand lift, recall of the new drinks and store traffic in between. Because the pop-ups rotated between partner locations, a store's traffic could also be compared before and after a CloudBar arrived.

Gave people a reason to show up

An in-store coupon bridged the feed and the counter.

Results

• $0.01 cost per engagement
• About 80% awareness in the test markets. This is our own model, not a Starbucks study: we compared Meta's reach and impression data with each city's population, and counted someone as aware only after about five exposures.
• Starbucks' feedback throughout the test: it looked good, so keep going and keep lowering the cost per engagement

What we never saw

Starbucks ran its brand-lift, recall and store-traffic studies internally and never shared the figures with us. That's normal with large brands, and it's why the awareness number above is labeled as our own model, not a client report. The clearest verdict we can point to is the review Crosby left afterwards, below.

Would we run it today?

Engagement is one of the metrics we no longer optimize for, because bots and click farms can fake it cheaply. Starbucks was different in one way that matters: the brand measured lift in its own studies, so the engagement number was a cost control, not the proof. Without an independent check like that, we wouldn't run an engagement campaign today.

"We had a very vague idea of where we wanted to take our project and OuiGrowth was instrumental in helping us build a strategy and then executing it. They are willing to ask "what if" and always thinking how to go bigger and better. They have a great feel for social platforms and advertising and how to deliver the most value from an impact and cost perspective- I would be happy to work with them again."

Crosby Steiner, Starbucks

What you can take from this

• In mid-sized markets, test geo-targeting models first. Too tight, and CPMs explode.
• When there's no click worth measuring, run in flights so the brand can measure lift between them.
• Label modeled numbers as modeled. Our 80% awareness figure comes from Meta's reach data, not a lift study.
• Engagement is only safe to buy when someone else can verify what it did.

Want a second set of eyes on your account? Book a free 30-minute consultation and we'll walk through what's working, what isn't, and where the next win is. Curious what it costs first? Our pricing is public.