Finding and keeping customers: how the challenges overlap

EPOP is NORC at the University of Chicago’s Entrepreneurship in the Population survey. Its 2026 questionnaire asked current owners and freelancers which customer-reach challenges they were currently facing in their entrepreneurial activity. They could select all that applied.^[1][2]

Two options were “Finding customers” and “Keeping existing customers.” The first did not specify only new customers, so we retain the survey’s wording rather than silently turning it into a measure of first-time customer acquisition. The question also offered digital or online presence, identifying market opportunities, and none of the listed challenges.

A person did not have to choose finding over keeping. That matters because separate percentages cannot tell us whether the same respondents selected both. We kept everyone with a valid answer to the pair in one common base, then counted the four possible combinations.

Finding without keeping was one distinct group; keeping without finding was another. A third group selected both. Treating all three as one undifferentiated customer-growth brief would hide what the respondents actually selected.

That does not make the finding-only group a set of businesses that should stop thinking about existing customers. Nonselection records what someone did not select as a current challenge. It does not prove that the person has no existing customers, sees no value in them or will never need to address the issue. The survey did not ask respondents to divide spending between the two tasks.

For a planning conversation, the result supports a simpler move: record each answer separately. If someone names finding customers, do not fill in the keeping-customer answer for them. If they name both, do not assume that one generic message will explain both needs.

The same data can answer a narrower question. Among the 970 respondents who selected finding customers and had a valid pair of answers, 333 also selected keeping existing customers. After weighting, the conditional share was 35.1%.

The denominator has changed. The 12.1% in the first table describes both selections among the entire valid-pair group. The 35.1% describes keeping among those who selected finding. Both are useful, but they answer different questions and should not be substituted for one another.

For the video producer, this is a reason to leave space for a second answer. A conversation that starts with “we need to find customers” may also uncover a current difficulty keeping existing customers. The estimate describes how often those selections coincided in this defined survey population. It does not predict whether the next person in a sales conversation will select both.

The confidence interval also gives a more useful picture than an exact-looking percentage alone. Under the survey’s design-based calculation, the conditional estimate has an interval of roughly 31%–39%. The business takeaway does not depend on pretending that 35.1% is a fixed rate for every type of company or creator.

The largest group, at 56.5%, selected neither finding nor keeping existing customers. Their choices could include digital presence, market opportunities or none of the listed challenges. The table keeps those possibilities together, so “neither” refers only to the two options we compared.

For a business commissioning content, that leaves room for a different assignment. If neither is a current challenge, ask what the actual need is before filling a calendar with acquisition or retention messages. If the answer is uncertain, keep that uncertainty visible in the planning notes. The worksheet below gives it a place.

The fictional producer might find that a previous client understands the service but needs a confirmed date for another project. The useful next step is a scheduling conversation. A prospective client, meanwhile, may need an explanation of what to prepare before commissioning work.

Those situations lead to different assignments. Describe the question in ordinary language, decide what an explanation can address and keep any other required action visible. Choosing the social format comes after that decision.

The producer identifies both finding customers and keeping existing customers as current challenges. Here is a completed brief that holds the two needs separately. The producer, the needs and the content are invented; this is an editorial application, not a marketing intervention tested by NORC.

Purpose of the first piece. Give someone considering a first project a clear introduction to the service and the information needed for an initial conversation. The producer will open the service post like this:

“A useful business video answers one clear question. What does the product do? What happens during the service? Why is this process hard to explain? If you are planning a project, send me the question your viewers need answered, where the video will appear and an example of the style you have in mind. We can discuss the scope from there.”

Purpose of the second piece. Help an existing client describe a possible next project. This reusable project note begins with the shared context rather than repeating the whole service introduction:

“Ready to plan another video? Start with what has changed since our last project. Send me the new audience, the channels you have in mind and the date you are working toward. Add any footage, graphics or product details you would like us to consider reusing. We will review the scope and schedule together before the project is agreed.”

The source material. Both pieces use the producer’s accurate service description and, if an example is included, genuine work they have permission to show. The first explains the service from the beginning. The second focuses on what a next project requires. Each has a clear reader and a next step, even if the source material overlaps.

What gets checked. The producer reviews the service details, permission to use any example and the accuracy of any client quotation. The scope and schedule remain subjects for a real conversation. The examples above avoid inventing a price, confirmed availability or a client result.

How the brief will be reviewed. After using the pieces, the producer can revisit the questions that prompted them: do prospective clients still need the service explained, and do existing clients still need help describing a next project? Their own observations can inform the next revision.

A company working with an influencer or independent creator can make the same kind of assignment: name the intended reader and the question to explain before choosing a social format. When two needs are present, two reader tasks can sit in one plan. The business can then decide which pieces to create from its own circumstances.

Use the following seven questions before deciding what to publish. Keep two sets of answers, headed “Finding customers” and “Keeping existing customers.” You can copy the questions directly into your planning notes.

1. Is this a current challenge for us? Record Yes, No or Unsure separately for each need. Keep the answer explicit so an empty space cannot be mistaken for agreement.

2. What makes us give that answer? Write the observation or question behind it. This is where the business describes its own situation rather than borrowing the national survey percentage as an explanation.

3. Who needs an answer from us? Describe the intended reader for each current need. An existing client considering another project may need different context from someone trying to understand the service for the first time.

4. What should that person understand? Name one concrete question the explanation should answer. For the fictional producer, that might be what to bring to an initial discussion or what has changed for the next project.

5. What can this piece explain? Describe the explanation and the real example or source it needs. Keep that assignment specific enough for a writer or creator to act on.

6. What else needs attention? Note any issue the content cannot resolve on its own. In the producer’s example, confirming a date requires a scheduling conversation.

7. What is the appropriate next step? State the action the explanation prepares the reader to take. Make the invitation match the information the piece has provided.

A Yes for both needs calls for examining both, rather than automatically writing twice as many posts. The answers might lead to separate explanations, a revision to an existing page or an operational change. Keep the completed questions alongside the content plan so the reason for each assignment remains visible when circumstances change.

We used the 2026 public-use release of NORC’s EPOP survey, containing 9,999 records. The survey covers noninstitutionalized U.S. adults aged 18 and over and used the probability-based AmeriSpeak panel. Fieldwork ran from February 20 to March 11, 2026.^[1][4]

We selected the 2,794 respondents assigned to the current-business-owner or current-freelancer pathways. The owner pathway has priority when roles overlap, so a person is not counted in both categories. The question refers to that person’s entrepreneurial activity; it does not identify every separate business they may own. Among the 2,755 valid-pair respondents, the weighted composition was 46.8% current owners and 53.2% current freelancers. That describes the combined analysis group, not an outcome comparison between the roles.^[2][3]

A valid pair required both challenge items to be explicitly coded selected or not selected. The other 39 eligible respondents had missing-coded answers: 1.4% of eligible respondents without weighting and 1.9% with weighting. Missing, unknown and logical-skip codes were not treated as nonselection. Checks found no contradictory none-plus-substantive selections or unexplained structural skips in the current-owner/freelancer group.

We applied the released cross-sectional weight, WTPROB, and the released sampling-stratum and primary-sampling-unit variables. The full survey design was constructed before selecting the analysis groups. The tables report design-based 95% logit confidence intervals, rather than simple-binomial intervals. The four-state estimates use 526 design degrees of freedom; the conditional estimate uses 193 for its finding-selected domain. We used the 2026 release alone, without pooling survey waves.^[1][3]

A separate implementation reproduced the central weighted proportions and Taylor covariance. The analysis materials retain field definitions, aggregate counts and estimates, and reproduction code.^[6] To reproduce the analysis, request your own 2026 data copy through NORC’s public-use-file request form^[5]. Our materials contain no respondent records. NORC’s codebook already describes the individual items. We compared the two selections within each person’s answers to show how the reported challenges overlap and separate.