Case studies / eCommerce
Startup paid media · Google, Meta & AdRollOur first recommendation was not to spend the budget.
Bluejay Bicycles hired us to run Google Ads for a several thousand dollar e-bike with no sales history, no audience data, and no brand search. We told them the honest answer for month one was a folding table at a local event: and that we'd start spending properly once we knew something.
A new brand with a great product and nothing else yet.
Bluejay Bicycles makes vintage-inspired electric bicycles, a design-led product at a price point where buyers take weeks to decide. When we came in, the company was new enough that almost none of the inputs a paid media account needs actually existed yet.
No sales history, no customer list, no brand search. What the company did have was a founder already running local events where people rode the bike for free, real demand happening in person that nobody had thought to record.
You cannot optimise toward data you don't have.
A startup with no sales history, no customer list, and no brand search is the account most agencies take and start spending immediately. The budget clears, the reporting looks busy, and when it doesn't work the explanation is that the market is hard. The client pays for a month of learning and receives none of it.
Buy the data with media budget.
Launch broad, let the platforms find an audience, and pay for every wrong guess at auction prices. On a several thousand dollar considered purchase with a long decision cycle, it takes months of spend before the signal is strong enough to act on: and the budget is gone by then.
- Month 1 spend Full budget
- Audience signal None yet
- Cost of learning Paid at auction
- Runway consumed Before it works
The data already existed. Nobody was capturing it.
The founder was already running local events where people rode the bike for free. Real demand, real reactions, and none of it recorded. Structuring that capture produced a seed audience and, more valuable, an understanding of what the product actually was to the people who wanted it.
- Month 1 to 2 media spend Deliberately reduced
- Contacts per event ~70
- Signal quality They'd ridden it
- Cost of learning Near zero
A great product, and nothing to point a campaign at.
Bluejay makes vintage-inspired electric bicycles, a design-led product at a price point where people take weeks to decide. When we came in, the company was new enough that almost none of the inputs a paid media account needs actually existed.
With a $4,000 monthly budget across every channel, there was no room to buy that information at auction prices.
Four things were missing at once.
-
No sales history to model from
No conversion data, no purchaser profile, no idea which audiences converted. Every targeting decision would have been a guess, and platform learning phases eat budget while they resolve.
-
No brand search, and generic terms are expensive
Nobody was searching for Bluejay because nobody knew it existed. The alternative, bidding on generic e-bike terms against established manufacturers, is one of the more expensive places a small budget can go.
-
The product has to be experienced
A several thousand dollar bike is not an impulse purchase from a static image. People need to sit on it, ride it, and feel the difference. No ad format replicates that.
-
Limited budget meant limited guesses
At $4,000 a month across search, shopping, and social, we could not honestly guarantee a strategy would work. So rather than promise one, we said which parts we were confident about and which we weren't.
Five phases, in this order.
The first one involved almost no advertising at all.
Turn the events into a data source
The founder was already running local events with free test rides. We advised holding back most of the media budget for the first month or two and putting the effort there instead, then built a structured capture process: full name, email, phone, address, and what each rider thought of the bike in their own words.
Around 70 contacts per event, every one of whom had ridden it.
Learn what the product actually is to buyers
The open-ended responses mattered as much as the contact details. What people said after riding it told us which benefits landed, which language to use, and who was actually reacting, information that would otherwise have cost months of creative testing to establish.
Messaging built from what buyers said, not from a brief.
Demand generation on Meta, infrastructure on Google
Facebook and Instagram carried the demand generation, because the product needs to be seen in motion. In parallel we set up Merchant Center and built Shopping, search, and brand campaigns: infrastructure that would only pay off once there was demand to capture.
The first few weeks were quiet. We had said they would be.
Month three: the data caught up
By month three the accounts had enough conversion signal to optimise against properly. Google Shopping reached 3.5x return on ad spend, and the search and brand campaigns began capturing demand the events and Meta activity had created.
Paid media started paying for itself.
Follow the data to retargeting
Testing across Meta showed something we hadn't planned for: retargeting consistently outperformed prospecting. On a considered purchase that takes weeks, staying present through the decision matters more than reaching new people. We shifted budget hard toward retargeting, then extended it across the web with AdRoll.
AdRoll retargeting reached 6 to 7x return on ad spend.
And one number we're not going to claim.
Once the data existed, Google Shopping was returning 3.5x and AdRoll retargeting was returning 6 to 7x, on a total budget of $4,000 a month across every channel.
The brand search campaign returned around 13x. It's the most impressive figure in the account and it's the one we'd caution you about.
Brand campaigns almost always report spectacular returns. They're cheap, and they reach people who are already searching your name. But those buyers didn't come from the brand campaign: something else created the demand and brand search collected it.
In this account, that something else was the founder's events and the Meta demand generation running alongside them. Presenting 13x as a marketing achievement would mean taking credit for a client's own work, which is exactly the kind of reporting that makes agency numbers hard to trust.
Brand campaigns are worth running. They protect your name from competitors bidding on it and they close people who are already convinced. They just aren't a growth channel, and any agency leading a pitch with a brand ROAS figure is showing you the easiest number they have.
- Total monthly budget $4,000, all channels
- Google Shopping 3.5x
- AdRoll retargeting 6 to 7x
- Brand search 13x, see above
- Contacts per event ~70
Four things worth taking to your own launch.
Find the data you already generate
Events, trade shows, showroom visits, sampling: most early-stage companies create real customer interactions and capture none of them. That's a seed audience sitting on a clipboard.
Ask what happens in month one
If an agency promises results from a cold start with no data, ask how. The honest answer for a new product is usually that the first month is for learning, and you should know that before you're billed for it.
Considered purchases live in retargeting
When the decision takes weeks, being present throughout beats reaching more people once. We expected prospecting to carry this account. Testing said otherwise, so we moved the money.
Discount brand ROAS in any pitch
It's the cheapest, easiest number in any account and it measures demand someone else created. Ask what the prospecting and shopping numbers look like instead.
What went into it.
Six of our services, on a $4,000 monthly budget.
Launching something with no data behind it?
Tell us what you're selling and what you already know about who buys it. We'll tell you honestly whether paid media is the right first move, or whether your budget goes further somewhere less obvious for a couple of months. Findings are yours to keep either way.
No obligation · We'll say if paid media isn't the answer yet