Founder anxiety and marketing drift: why most "marketing" is anxiety management

A solo founder I know logged every marketing action they took for one week in April 2026. The list ran to 41 entries. Tweets drafted and discarded. Three landing-page hero rewrites. Two competitor pricing pages opened in five different sessions. An hour and a half on a logo revision nobody had asked for. One cold email sent. Zero customer conversations. When asked what each action would change if it worked, the founder could answer for three of them. The other 38 existed to lower their anxiety, not to move pipeline.

This essay is about that gap. It is not about productivity. It is about the specific way anxiety dresses up as marketing work, the cost of letting it, and a 14-day audit method that surfaces what to cut without requiring you to feel better first.

// 01What marketing drift actually looks like

Marketing drift is the founder version of what behavioral economists call activity substitution. When the load-bearing work is ambiguous and the feedback loop is long, the brain seeks adjacent work that produces immediate, legible output. Refreshing the analytics tab produces a number. Tweaking the hero copy produces a visible diff. Reading three new articles on positioning produces a sense of momentum. None of these are marketing. They are anxiety regulation, billed to the marketing column.

The tell is uniform across founders. The actions are small, plural, and ungrouped. They happen between the things the founder calls “real work.” They produce no artifact a customer will ever see. The founder finishes the week unable to point to a single completed marketing experiment, and yet describes themselves as having “done a lot of marketing.”

The taxonomy of drift, in five shapes

• Optimization without traffic: Editing landing-page copy, hero illustrations, and CTAs on a page that gets fewer than 50 unique visitors a week. The fix has no addressable audience. You are rearranging a room nobody has entered.
• Research without a decision: Reading competitor pricing, opening seven Demand Curve essays, sending yourself five YouTube videos titled "how to find PMF." Research without a pending decision is consumption with a marketing label on it.
• Posting without a plan: Drafting tweets in a doc you never publish. Writing a long LinkedIn post and then deciding the timing is wrong. The work happens in private, where it generates no learning. The unposted draft is a fossilized version of the anxiety that produced it.
• Dashboard refresh as a verb: Opening Google Analytics, Plausible, or PostHog more than once a day on a product with under 100 weekly active users. There is nothing to learn at that resolution. You are reading the slot machine.
• Asking without testing: Polling Twitter about whether the headline should be "for solo founders" or "for indie hackers." Posting in 4 Discords asking which logo is better. The poll is a stand-in for the experiment you do not want to commit to running.

// 02Why anxiety builds work, not rest

Marketing for a solo founder has two properties that combine badly. The feedback loop is long, often 4 to 12 weeks before any meaningful signal. The information environment is infinite, with new tactics, new tools, and new comparisons appearing every day. The combination produces what researchers at the Frontiers in Psychology imposter-syndrome program call “anticipatory uncertainty.” The brain does not tolerate the open question well, so it generates closed micro-tasks that feel like they are addressing it.

Drift is the felt-sense of doing those micro-tasks. The relief is real. So is the cost.

This is also why “just work harder on marketing” is the wrong prescription. Working harder without a structure means producing more micro-tasks, which means more relief, which means the founder reports feeling busier without producing more learning. Statsig’s product blog once described this pattern as “the company’s documentation-trained chatbot responded to a customer request with an unexpected dish suggestion.” The chatbot was trained on a corpus of activity. It learned to suggest activity. It did not learn to suggest evidence.

The three properties of work that scratches the anxiety itch

• Immediate visible output: The diff is visible inside one work session. A copy change. A new tab open. A draft saved. Compare with running a 30-day experiment, where the output arrives in 30 days. Anxiety prefers the diff.
• No external commitment: The action does not require you to put anything in front of a stranger. Editing a homepage that nobody is reading has zero risk. Cold-emailing 20 strangers has 20 units of risk. Anxiety prefers the homepage.
• Reversibility: The action can be undone in one click. Anxiety treats irreversibility as a threat. A published tweet is more frightening than a drafted one. The drafted one preserves all options. The published one closes them.

// 03The 7-day evidence test

The single fastest way to separate pipeline work from anxiety work is a forced question. Before any marketing action, ask: what evidence would change my mind about this in 7 days? If the answer is “none, but I will feel better,” the action is anxiety. If the answer is “this metric will move from X to Y, and if it does not, I will stop doing it,” the action is an experiment.

The 7-day window is load-bearing. A longer window lets drift hide. A shorter window biases toward channels with cheap feedback loops, which is fine for a first audit. The point is that an action must be falsifiable inside the window to count as marketing.

// 04The 14-day audit

The audit is one experiment. The metric is the percentage of marketing actions you take over 14 days that fail the 7-day evidence test. The kill threshold is 15%. The success threshold is 50%. The audit costs you no time except the act of logging.

How to run it, step by step

• Day 0: instrument: Open a single doc. Title it "marketing log." Three columns: timestamp, action, why I did it. Nothing else. The friction must be low or you will stop logging by day 4.
• Days 1 to 14: log every action: Every time you touch anything marketing-adjacent, write a one-line entry. "10:42, rewrote hero h1, because the current one feels weak." "14:03, opened Plausible, because the morning post should have data by now." Do not edit the entries. The honest log is the audit.
• Day 7: midpoint review: For each logged action so far, write yes or no next to it: "would the result of this action have changed my mind about anything within 7 days?" Yes counts as evidence work. No counts as anxiety work. Compute the percentage. Do not change behavior yet.
• Day 14: full audit and cut list: Repeat the yes/no scan on the full log. Compute the percentage. If the no rate is above 50%, you have a kill list. Pick the five action categories that show up most often on the no side. Those are the cuts. Replace them with one or two real experiments from the cluster index.

The reason this audit works where general “be more disciplined” advice fails is that the audit produces evidence the founder cannot dismiss. The brain accepts arguments from a log that it rejects from a lecture. PostHog learned this in its own internal experiment program when it shipped its self-driving product retrospective. The team did not move until the dashboard showed them what they had been doing. Then they moved fast.

// 05What to keep, what to cut

Keep, even if it feels slow

• One-to-one outreach with named targets: A cold email or DM to a named human is always evidence work. Reply or no reply, you learn something within a week. The fear that comes with sending it is exactly the friction that makes it real.
• A landing page tied to a specific traffic source: A page that has 20 paid clicks or 30 organic visitors a week can produce a conversion-rate signal in 7 to 14 days. Editing that page is evidence work. Editing one with zero traffic is drift.
• A 30-day channel experiment with a kill threshold: The five playbooks in the Xi distribution series each have a pre-committed kill threshold. The clock runs. The verdict is automatic. See the kill-threshold definition for the long version.

Cut, even if it feels productive

• Optimization on dead pages: Any page with fewer than 50 weekly visitors does not earn the time. Ship traffic first, then optimize. The order matters and is non-negotiable.
• Tool research without a stalled decision: Comparing 6 newsletter platforms when you have no list is anxiety. Comparing 2 when you have 400 subscribers and a deliverability issue is evidence.
• Vanity polling: Twitter polls that ask people who are not your customer to choose between two options you control entirely. Replace with a 5-customer interview using the Mom Test rules.

// 06What changes after the audit

The first week post-audit feels worse, not better. The micro-tasks that were soaking up anxiety are gone, and the anxiety is still there. This is the point at which most founders re-introduce one or two of the cut categories under a new label. Resist this for 14 days. The anxiety attenuates only when the brain learns that the open question is genuinely being addressed by the real experiment, not the micro-task substitute.

Two outcomes are common by week 3. The first is that founders who ran the audit honestly end up shipping one to two more real experiments per month than before, with no increase in time worked. The shape of the calendar has changed, not the size. The second is that founders who could not give up the micro-tasks discover they were never short on time. They were short on willingness to put something irreversible in front of a stranger.

Both outcomes are useful. The second one is the more important diagnosis. If the audit reveals that the bottleneck is willingness, not time, the right next read is the imposter and identity essays in this cluster, not another tactical playbook. The right next action is one published artifact this week, even a bad one. The fastest way to dissolve irreversibility-fear is to make 30 things reversible by shipping the 31st.

// 07Five things to carry forward

• 01: Drift is anxiety regulation, not marketing. Naming it correctly is the first move that lets you cut it without guilt.
• 02: The 7-day evidence test is binary. Either the action is falsifiable inside a week, or it is not. There is no in-between.
• 03: The 14-day log produces evidence the brain accepts. General advice produces compliance for one day. The log produces a verdict.
• 04: Cut on the day-14 audit, not before. The honesty of the log requires that you do not adjust behavior during it.
• 05: Replace cuts with real experiments, not nothing. The cluster index has 7 of them with kill thresholds pre-filled. Pick one. The replacement matters more than the deletion.

For the structural fix to the imposter loop that often drives drift in the first place, the next essay is imposter syndrome for solo founders. For the decision-paralysis variant, where drift compounds because no choice is ever forced, read founder decision paralysis.