A goal only works if it starts from a real number.
That's why GPS doesn't simply assign the same benchmark to every campaign.
Depending on the information available, we generally establish the initial performance baseline in one of three ways.
1. Historical Performance
If we're working with an existing account that has reliable historical data, that data can provide the strongest starting point.
We may look at:
Historical lead or purchase volume
Cost per lead or acquisition
Conversion rate
Close rate
Customer acquisition cost
Revenue or ROAS
Other relevant business outcomes
The goal is to understand what the account is actually producing today before establishing where we want it to go.
2. Business Math
For a new campaign, we can also work backward from the client's economics.
Depending on the business, useful inputs may include:
Average sale value
Gross margin
Lead-to-customer close rate
Average order value
Customer lifetime value
These numbers help answer a much more useful question than "What's a good CPL?"
What can this particular business afford to spend to acquire a lead or customer?
A $150 lead could be incredibly profitable for one business and completely unsustainable for another.
Context matters.
3. Industry Benchmarks
Sometimes a new client simply doesn't have enough reliable information yet.
That's okay.
When historical or business-level data isn't available, published industry benchmarks can provide an initial baseline. Those figures should be treated as estimates, not permanent truths.
As the account generates its own performance history, real data becomes more valuable than the benchmark we started with.
GPS currently uses a 90-day performance assessment to replace estimated assumptions with the account's own numbers where appropriate.
What If My Client Doesn't Know Their Numbers?
Start with what they do know.
Even a conversation about average sale value, margin, close rate, or desired customer volume can uncover information that makes the media plan stronger.
And if some inputs aren't available yet, that's exactly what benchmarks are for.
We would rather begin with a clearly identified estimate and improve it with real data than pretend an arbitrary target is a fact.