Agency return on investment is usually discussed as a reporting problem. It is really a definition problem. Most businesses cannot tell whether their agency is working because nobody agreed in advance what working would look like, and by the time the question gets asked seriously, the baseline is gone.
Which Number Actually Matters?
Marketing reporting is full of metrics that move without meaning anything. Impressions, reach, and engagement rate describe activity. They are useful as diagnostics, the way a mechanic watches engine temperature, but they are not the outcome on their own.
TSN always provides metrics, and the right ones depend on the business. For some clients that means qualified leads, conversion rate, or cost per acquisition. For others, something simpler, like growth in social media followers, is a perfectly good place to start. What matters is agreeing on two or three numbers before the work begins, not building an elaborate list.
Record the Baseline
This is where most engagements fail before they start. If you do not record where those numbers stood in the month before the agency began, you will spend the following year arguing about attribution from memory.
Record them. Even rough figures are far better than none, and an agency that does not ask for baseline numbers in the first month is not planning to be measured.
Agree on the Time Horizon
Different work pays back on different schedules. Paid search can show meaningful signal within weeks. Conversion work on a website shows up within a quarter. Content and search optimization generally need two to three quarters before the curve turns. Public relations and brand work operate on the longest horizon and are the hardest to attribute cleanly, which does not make them less valuable, only harder to prove.
A program judged on a schedule shorter than its natural payback will always look like a failure. This is the single most common reason good marketing gets cancelled.
Lower Your Expectations on Attribution, Raise Your Rigor
Perfect attribution does not exist. A buyer reads an article, forgets it, sees an ad, searches your name three weeks later, and arrives through direct traffic. No model captures that cleanly.
What works better than chasing a perfect model is combining three imperfect signals:
- Platform and analytics data for the channels that report.
- Self-reported attribution, which means actually asking every inquiry how they heard about you and recording the answer.
- Correlation over time, comparing the shape of your pipeline against the shape of your marketing activity across quarters rather than weeks.
None of the three is sufficient alone. Together they usually produce a picture good enough to make decisions with, which is the only standard that matters.
Do Something With the Answer
A reporting package that arrives monthly and is never discussed is theater. The value comes from the conversation about what the numbers changed. If your agency reporting does not end with a decision about what to do more of and what to stop, the reporting is not doing its job.
A Reasonable Structure
Agree two or three outcome metrics before work begins. Record the baseline. Agree a realistic horizon for each stream of work. Review monthly for diagnostics and quarterly for outcomes. And insist that every quarterly review ends with something being stopped, because a program where nothing is ever cut is a program where nobody is really reading the data.
Frequently Asked Questions
Which metrics actually indicate agency performance?
It depends on the business. Qualified leads, conversion rate, average deal size, and cost per acquisition are common outcome metrics worth tracking. Simpler metrics, such as growth in social media followers, can also be a legitimate starting point, since TSN always provides metrics suited to the goal.
How long before a marketing program shows results?
It depends on the discipline. Paid search can show signal within weeks, website conversion work within a quarter, content and search optimization generally within two to three quarters, and public relations and brand work on a longer horizon still.
Why is a baseline so important?
Without recorded figures from the month before the engagement began, any later discussion of results becomes an argument from memory. Rough baseline numbers are far more useful than none, and an agency that never asks for them is not planning to be measured.
Can marketing attribution ever be fully accurate?
No. Buyers encounter brands across many touchpoints and arrive through channels that credit none of them. The practical approach combines platform analytics, self-reported attribution collected by asking every inquiry directly, and correlation of pipeline shape against activity over quarters.
What makes agency reporting useful rather than decorative?
Reporting is useful when each review ends in a decision about what to increase and what to stop. A report delivered monthly and never discussed produces no value regardless of how well it is designed.