--- title: "How to Know Your Marketing Is Working When the Data Cannot Prove It" url: "https://marketermagazine.co/insight/how-to-know-your-marketing-is-working-when-the-data-cannot-prove-it/" author: "Orkan Arat" published: "2026-09-25" updated: "2026-09-25" --- # How to Know Your Marketing Is Working When the Data Cannot Prove It ### The problem every marketer has stopped admitting Every dashboard I have ever opened promises a clean story. Spend here, get sales there, connect the dots. In practice the dots do not connect anymore. Browsers block cookies, phones block identifiers, customers switch devices mid-decision, and platforms report their own numbers using their own rules, which conveniently make every platform look like the hero of your growth. I have run performance marketing budgets since the direct-response days of satellite TV and home security, back when a phone call and a name were the whole tracking stack, and I have watched the industry go from too little data to too much data that nobody fully trusts. Most marketers respond to this by pretending the problem does not exist. They pick the platform-reported number that flatters the campaign and move on. That is not a strategy. That is picking the answer you like best off a menu of wrong answers. ### Stop trying to track everything and start testing things The teams that make good decisions under bad tracking are not the ones with the best dashboards. They are the ones who have replaced tracking with testing. The distinction matters. Tracking asks a platform to tell you what it did. Testing asks reality to tell you what happened. The simplest version of this is a holdout. Take a channel you already run, whether that is paid search, a retargeting campaign, or an influencer program, and turn it off for a defined period in one region, one audience segment, or one time window, while it keeps running everywhere else. Watch what happens to total revenue, not just the metric inside that channel. If revenue barely moves when the channel goes dark, that channel was mostly claiming credit for sales that would have happened anyway. If revenue drops noticeably, you have real evidence, not a platform’s self-reported click count. Small teams resist this because it feels like giving up guaranteed volume for an experiment. I understand the discomfort. But a two-week holdout on ten percent of a channel’s spend costs far less than a year of scaling a channel that was never doing the work you credited it with. ### Triangulate instead of trusting a single number When you cannot run a clean holdout, and sometimes you genuinely cannot because volume is too low or the business cannot tolerate the risk, triangulate. Look at three separate signals and see if they agree before you believe any one of them. First, platform-reported conversions, with the understanding that these are almost always inflated because platforms count anyone who saw an ad and later converted, even loosely. Second, a simple pre- and post-comparison at the business level. Did total orders, total leads, or total revenue actually shift when you turned spend up or down, independent of what any one platform claims? Third, ask new customers directly how they found you. A short survey question at signup or checkout, phrased as an open text field rather than a dropdown of your marketing channels, produces surprisingly honest answers. People will tell you they saw a friend mention you online or heard the name somewhere, and that kind of answer tells you something no pixel ever will. None of these three signals is reliable alone. Platform numbers are biased toward the platform. Pre- and post-comparisons get confused by seasonality and competitor activity. Self-reported answers suffer from people misremembering or not knowing. But when two or three of them point the same direction, you are on solid ground. When they disagree, that disagreement is itself useful information, because it tells you exactly where your measurement is weakest and where you should test next. ### Choosing where to spend when you cannot measure perfectly This approach changes how I think about picking a first paid channel, which is the question I get asked most by founders moving from zero marketing spend to their first real budget. My advice runs against the common wisdom to test everything at once and let the data decide. With imperfect tracking, testing five channels simultaneously just gives you five sets of numbers you cannot trust, competing for credit. Instead, pick one channel with a plausible mechanism for reaching your actual buyer, commit a budget large enough to see a real signal over four to six weeks, and run it in relative isolation from other new experiments. Resist the urge to add a second channel until you have a baseline read on the first one. This is slower than launching everywhere at once, and it is the only way a small team without a data science function can actually learn something. The uncomfortable truth is that marketing measurement was never as precise as the industry made it sound, even in the years when tracking worked better than it does now. What has changed is that the imprecision is now visible instead of hidden behind clean-looking dashboards. Treating that visibility as a reason to test more carefully, rather than a reason to give up on measurement altogether, is the actual skill. This is also the operating model I apply through Plondo, where I work with direct selling companies and small businesses building acquisition programs from scratch, usually with far less clean data than a large company would have. The constraint is not unusual. Most businesses are making channel decisions with incomplete information, and the ones that grow well are the ones honest about that instead of hiding behind a number that happens to be convenient. --- Orkan Arat is the Chief Executive Officer of [Plondo Network Inc](https://plondo.com), an agentic technology and marketing partner that builds software and runs marketing programs for direct selling companies and small businesses. Learn more at Plondo, https://plondo.com.