The second agency lasted fourteen months. The first one made it eleven. Between them the website got rebuilt, the ad budget moved from Google to LinkedIn and partway back, and a freelancer took over the email list for a while. Through all of it, the complaint in the Monday meeting never changed: we’re spending real money and we can’t tell what it’s doing.
Three vendor changes. One constant. At some point the diagnosis has to move off the parts and onto whatever the parts are plugged into.
The part-swap pattern
When marketing disappoints, the instinct is replacement. Fire the agency. Rebuild the site. Try a new channel. Hire someone in-house this time. Each swap is a reasonable response to a real frustration, and I’ve watched owners run this loop for years, a little poorer and a little more cynical after every round. I ran my own business for close to a decade. I know exactly how rational each individual swap feels from the owner’s chair.
Here’s the diagnostic fact the loop keeps missing: a problem that survives replacement isn’t in the part. If two different agencies, three different channels, and a new website all produced the same complaint, the common denominator is what they were all connected to. Or weren’t.
Most lists of marketing problems won’t tell you this. They’ll give you a problem per tactic: low traffic gets an SEO fix, poor conversion gets a landing-page fix, weak leads get a targeting fix. Sometimes that’s right. But when the problems recur and rotate, you’re watching one problem wear different costumes.
Four tells your marketing problem is a systems problem
1. The problem survives vendor changes
The strongest single signal. If you’ve swapped providers and the complaint is intact, the vendors were never the variable. What stayed the same across every swap? Usually: no written strategy the vendor could execute against, no measurement the vendor plugged into, no working rhythm where results changed what happened next. New part, same disconnection.
2. Everything underperforms a little
A genuine channel problem is loud and local. The ads broke, the numbers fell off a cliff, everything else held. A systems problem is quiet and everywhere: the ads are mediocre, the email is mediocre, organic drifts. Nothing fails hard enough to fire anyone over, and nothing works well enough to double down on. Uniform mediocrity is a structural symptom. Channels can’t compound when nothing connects them.
3. Nobody can trace a lead to revenue
Ask one question: for last month’s best customer, what touched them first, and what did that touch cost? If the answer requires opening four tools and a spreadsheet, and still comes back “we think LinkedIn?”, your problem isn’t any channel’s performance. It’s that performance is unknowable. Every downstream complaint about marketing “not working” inherits this one, because working and not working can’t be told apart.
4. Fixes work, then decay
The new agency’s first quarter looked promising. The rebuilt site converted better for a while. Then the line sagged back to where it was. Early wins that decay are the signature of effort without a loop: nothing carried the lesson forward, so the system reverted to its resting state. Improvements stick when a measurement connection feeds a monthly decision. Without that, every fix is a temporary override.
Two tells it’s genuinely a channel problem
Fairness demands the other side, because sometimes the part really is broken. First: one thing is clearly failing while the rest hold. Traffic is fine, email is fine, but paid search costs tripled since spring. That’s local. Second: the signal is specific and readable. You can name the metric, the date it moved, and the plausible cause. Specific, legible failures deserve specific fixes, and a good specialist will earn their fee. It’s the vague, recurring, everybody’s-a-little-disappointing pattern that no specialist can fix, because it doesn’t live in any specialty.
What to do before you buy anything
Not a new channel, not a new vendor, not a new tool. One exercise first: pick a recent customer and trace them backward. What did they see, click, read, and receive, in what order, and which of those steps can you actually see in your data? Somewhere in that trace you’ll hit a gap where the trail goes dark. That gap is your real marketing problem, and it’s almost always one of the three connections I laid out in what a marketing system actually looks like: strategy to execution, execution to measurement, or measurement back to strategy.
The same logic applies to the tool pile, by the way. Swapping software is part-replacement with a subscription attached, and the stack usually costs more than it delivers for exactly this reason. And if the trace exercise leaves you suspecting the problem is upstream of all of it, why your marketing isn’t working walks the broader diagnosis.
Naming the broken connection is the first deliverable of every fractional marketing engagement I run. Not more activity. Activity without accuracy accomplishes nothing, and accuracy starts with knowing whether you’re fixing a part or a system.
Marketing problems FAQ
What are the most common marketing problems for growing businesses?
The visible ones are underperforming channels, weak lead quality, rising acquisition costs, and vendors who don’t deliver. The underlying one, for most businesses between $2M and $50M, is structural: activities that don’t connect to a strategy, measurement that can’t trace results to revenue, and no rhythm where results change the plan. The visible problems recur until the underlying one is fixed.
How do I figure out why my marketing isn’t working?
Trace one real customer backward through every touchpoint and note where your visibility ends. If you can’t complete the trace, the gap you hit is the diagnosis. If you can complete it, look at whether the problem is loud and local (one channel clearly broken) or quiet and spread across everything. Local problems get tactical fixes. Spread problems are system problems.
When is a marketing problem actually a systems problem?
When it survives vendor changes, shows up as mild underperformance everywhere rather than sharp failure somewhere, resists measurement, or responds to fixes briefly before decaying. Any one of those tells suggests the problem lives in the connections between your marketing parts. Two or more make it near certain.