Insights / gtm-design

How to Test Positioning in Real Sales Conversations

By Daniel Tonkopiy | 2026-11-02

Test positioning by using one written sentence in real sales calls with qualified buyers, then logging what each buyer does with it: which alternative they compare it with, whether they can restate the difference, what they object to first and whether a budget owner agrees a dated next step. This suits founders selling directly before or around launch. Set pass lines before the first call. The main constraint is volume: a small batch gives a judgment, not a statistic.

The result this process should produce

This guide is for founders before or around launch who have a written positioning sentence and real sales calls on the calendar. The wider sequence is in How to Build a Startup Go-to-Market Strategy by Stage, and choosing what to compare with, plus a separate paraphrase session, is in How to Position a Startup Against the Customer Current Alternative. This guide starts where that one stops. The sentence exists, and you need to find out whether it holds up when a qualified buyer hears it inside a real deal.

The result is one written decision at the end of a batch of calls: keep the sentence, rewrite it, or stop rewriting because the problem sits in the offer or the segment. A sentence that sounds better to the team is not a result.

Why calls and not a homepage test? April Dunford's podcast listing for her episode on testing and execution says 'live sales conversations with best-fit prospects are a much better way to test whether positioning truly lands.' In the automated transcript of that episode she explains that a homepage test also tests the messaging and the page design, and that she cannot know what traffic is arriving unless it is controlled. That is a practitioner's view and not a controlled study, and she describes companies with sales teams, not founders selling alone.

Daniel's go-to-market mentoring page says 'The first ten conversations with real buyers usually change the product more than another quarter of building would.' That is his own judgment and not a measured result, but it explains why the sentence belongs in those conversations and not only in a slide.

Prepare the current customer alternative

A positioning sentence makes a claim about what the buyer does today, so write that claim down before the first call. Set up one page per batch:

• The sentence. One sentence for one segment, word for word.
• The alternative it assumes. The workaround, supplier or rival you expect buyers to compare it with.
• The restatement you hope to hear. What a buyer who understood would say, written in plain words.
• The calls. Prospects who fit the segment and can reach the budget owner. Leave out friends, former colleagues and anyone who owes you politeness.
• What stays fixed. The offer, the price and the person pitching. Otherwise you cannot tell what changed the result.
• The pass lines. For each signal below, the count that would count as a pass.

The transcript supports keeping the seller fixed. Dunford describes pitching to 'pre-qualified prospects' with one or two of the strongest reps, and she counts 'the reps ability to pitch' as part of what the test measures. A founder is usually the only seller, which keeps delivery constant, but it also means the person running the test wants the sentence to work.

That is the reason to write the pass lines first and leave them alone. Nosek and colleagues, in The preregistration revolution in PNAS, describe defining 'the research questions and analysis plan before observing the research outcomes' so that predictions stay distinct from explanations made afterward. That paper is about scientific research, and a call log is not a study, but the habit carries over: a line chosen after the calls can simply describe what happened.

As a planning number, use a batch of six to eight qualified calls per sentence. This is an editorial starting point. None of the sources read for this guide sets a number of sales calls.

Work through the meaningful product difference

Run the sentence inside the normal call, in the same order each time. It takes a few minutes, and the rest of the call stays a sales call.

1. Ask about today first. Before saying anything about the product, ask how the account handles the problem now. Write down what the buyer names, in the buyer's words.
2. Say the sentence once, then stop. Use the written words with no demo, example or rescue.
3. Ask what it would replace. Offer a choice: would this replace what the account does today, or sit alongside it? Note what the buyer compares it with before you suggest anything.
4. Ask for the restatement. Ask how the buyer would explain it to the person who approves the spend. Write the answer as spoken and do not correct it.
5. Let the first objection arrive. Do not head it off. The first objection is a signal, so write it down before you answer it.
6. Ask for the next step. Make the normal request at the normal price, with a date and the budget owner in it.

The difference shows up in steps three and four. A buyer who names the alternative you expected and cannot say what changes has heard the comparison and missed the difference. That is a different failure from a buyer who compares you with the wrong thing, and it needs a different fix.

Avoid asking whether it makes sense. Pew Research Center's guide to writing survey questions says some respondents tend to agree with agree-disagree statements, that this is more pronounced when an interviewer is present and that 'A better practice is to offer respondents a choice between alternative statements.' Step three borrows that idea. Digital.gov's guide to paraphrase testing says to ask participants to explain a section in their own words and warns against questions answerable with a yes or no, and step four borrows from that. Both pages are about surveys and documents, not sales, so these steps are adaptations.

Dunford's transcript describes saying something another way when a customer does not get it. That is a good way to learn, and this guide adds one stricter rule: score the reaction to the written sentence first, then try other wording, and carry any wording that worked into the next batch as a new sentence.

Have someone else take notes, or write them straight after the call. This guide does not cover call recording rules, so check them before recording anything.

Resolve proof the buyer understands the offer

Score four signals per call and keep them apart, because they answer different questions.

The objection signal follows what Dunford says she watches for in the transcript: where prospects are 'getting confused' and what their questions reveal. A price objection is usually better news than a confused question, because the buyer has understood enough to weigh the offer.

Understanding and demand are separate. A buyer can restate the sentence perfectly and still not buy. If that happens across a batch, read When Customer Interviews Say Yes but Nobody Buys before changing any words.

The sales call positioning grid

The grid in the earlier positioning guide compares alternatives so you can choose one. This one compares versions of a sentence against lines written in advance. Fill in one column per sentence and do not add calls to a column after the sentence changes.

Then read the column:

• Every line met. Keep the sentence and roll it out.
• Comparison and restatement both miss. Buyers do not share your frame. Rewrite against the alternative they named on the opening question.
• Comparison met, restatement missed. Buyers know what you resemble and not what changes. Rewrite the difference and keep the comparison.
• Comparison and restatement met, next step missed. The sentence is doing its job. Leave it alone and check the offer, the price, the timing and who was on the call.
• One kind of account passes and another misses. Split the segment and give each part its own sentence and batch.

A worked test with explicit assumptions

The following is a hypothetical example with invented details. It illustrates the method and does not describe a real company, customer or measured result.

A founder sells a quoting tool to owner-run tree care companies at an intended $1,800 a year, or $150 a month. It prices routine jobs from a homeowner's photos. Sentence A reads: 'Field service software built for tree care companies.' It assumes that scheduling software is the alternative. The founder writes the pass lines for 6 calls per sentence, runs the calls and finds that 5 of 6 buyers already have scheduling software and see no reason to switch. Yet 4 of the 6 answer the opening question by describing the free site visit, where the owner drives out to price every job.

So the founder writes sentence B: 'For owner-run tree companies that drive out to price every job, this prices routine jobs from the homeowner's photos, so the visits go to the jobs worth winning.' The founder then runs it with 6 other buyers at the same price. During the fourth call, held at a job, the owner reaches up and taps the trunk of a large tree with a small tool while the homeowner watches, then tells the founder that no photo shows whether a trunk is hollow.

Sentence A met the comparison line and still failed: buyers knew what it resembled and saw no reason to change. Sentence B met all four lines. A typical restatement was 'So the crew stops driving out for quotes that go nowhere.' The first objections to sentence B were 4 about accuracy, 1 about price and 1 about timing, so the objection on the tree visit was a question of proof and not confusion. The founder keeps sentence B and narrows the promise: routine jobs only, with hollow trunks and large removals still getting a visit. The 2 dated next steps are paid 30-day pilots at $150.

The founder also checks the size of the difference. An owner making 5 free visits a week for 40 weeks makes 200 visits. At an assumed 1 hour each and $45 an hour, that is $9,000 a year. If photos replace 40 percent of visits, or 80 visits, the saving is $3,600, which is twice the $1,800 price. The pilots exist to measure that. The assumptions stay in the file with the decision: that the saved hours would go to paid work, that 40 percent of visits can be replaced, that $45 an hour is a fair cost, that 12 calls say something about the segment, and that a founder scoring these calls alone scored them fairly.

Common mistakes and limits

• Testing on friendly calls. A former colleague will restate almost anything to be helpful.
• Scoring the rescue. If you explained the sentence and the buyer then understood, the sentence failed and the explanation passed. Keep the explanation for the next batch.
• Editing mid-batch. Each change starts a new column.
• Asking whether it makes sense. A polite yes tells you little.
• Counting a step from someone without budget. A colleague's enthusiasm is not the approver's date.
• Rewriting when the offer is the problem. Clear sentences do not fix a price or a segment that does not fit.
• Scoring alone. Ask someone else to score the notes against the lines without knowing which sentence you prefer.

The method has limits. A batch of calls gives a judgment and not a statistic, and a result mixes the sentence with the delivery and the sample of buyers. It suits products sold in conversation, while self-serve and consumer products may need purchase and usage data instead. The signals, pass lines and reading rules are an editorial framework, not a measured result.

Acceptance criteria and the next action

Accept a sentence only when all of these are true:

• The pass lines carry a date earlier than the first call.
• Every call in the batch was with a qualified buyer.
• The written sentence was used unchanged in every call in the batch.
• Every line was met within that one batch, with no pooling across versions.
• The notes keep buyers' words, including the failures.
• A second reader scored the notes and reached the same reading.
• Each recurring objection has a written answer.

The next action is to put the accepted sentence where buyers meet it first: outreach, then the opening of the pitch, then the homepage. Keep logging the four signals in ordinary calls so that a drift shows up early. Retest when the segment, the product or the main rival changes. If a second batch also fails, stop rewriting and check the offer and the segment before writing a third sentence.

• The wider sequence by stage: How to Build a Startup Go-to-Market Strategy by Stage
• Choosing what to compare with: How to Position a Startup Against the Customer Current Alternative
• Learning about the buyer before any pitch: How to Run a Customer Interview Without Pitching

Daniel's go-to-market page says the first session is free and takes 30 minutes. It says the aim of that session is to 'identify the bottleneck and choose the next sales test'. Book a startup mentoring session to review your startup positioning decision with your actual evidence and constraints, and bring the filled grid with the buyers' own words.

Frequently asked questions

How do you test positioning in real sales conversations?

Write one sentence for one segment and set pass lines before the first call. In each qualified call, ask how the account handles the problem today, say the sentence once and stop, ask what it would replace and how the buyer would explain it to the approver, note the first objection, then ask for a dated next step. Read the counts for the batch and decide whether to keep or rewrite.

What evidence should a founder collect for startup positioning?

For each call: what the buyer does today in the buyer's words, what the buyer compared the sentence with, the restatement as spoken, the first objection as spoken, and any dated next step from the budget owner. Keep the pass lines with their date, and keep the failures in the log along with the passes.

What are common mistakes in startup positioning?

Testing on friendly calls, rescuing the sentence and scoring the rescue, changing the wording in the middle of a batch, asking whether it makes sense, counting a step from someone without budget, moving the pass lines afterward and rewriting the sentence when the offer or segment is the real problem.

How many sales calls does it take to test positioning?

None of the sources read for this guide sets a number. Plan on six to eight qualified calls per sentence as an editorial starting point, set the pass lines in advance and treat the result as a judgment about what to do next, not as a statistic.

Sources

1. Post Positioning - Testing and Execution (Positioning with April Dunford), Apple Podcasts Checked 2026-11-02
2. Post Positioning - Testing and Execution, episode page and automated transcript, Podscan Checked 2026-11-02
3. The preregistration revolution (Nosek, Ebersole, DeHaven and Mellor, PNAS) Checked 2026-11-02
4. Writing Survey Questions, Pew Research Center Checked 2026-11-02
5. Paraphrase testing (Plain language guide series), Digital.gov Checked 2026-11-02
6. Go-to-Market Mentor and GTM Strategy Advisor Checked 2026-11-02

Startup positioning | Sales call positioning grid | GTM design | Sales conversations | Founder playbook