AdvancedLesson 26 of 305 min read

Brand, Positioning and Message Testing

Write a five-part positioning line, test three headlines with a chi-square test and a stricter line for multiple comparisons, and keep your brand consistent and honest.

In this lesson you will

  • Write a positioning line
  • Test three messages fairly
  • Use a stricter line for multiple comparisons
  • Keep brand promises true

About 26 minutes

What is the shop about?

Noor asks three friends what her shop stands for and gets three answers: calm evenings, long-lasting candles, handmade gifts. All are true, and a message that tries to say all three says nothing. Positioning is the single place you want to hold in a customer’s mind, compared with the alternatives. Your brand is everything people connect with you: the name, the look, the promise and how you treat them. Strong positioning makes every other marketing task easier, because you know what to say and to whom.

Writing a positioning line

  1. For whom. The segment from lesson 11 that you serve best.
  2. What need. The problem or wish in their words.
  3. Compared with what. What they would otherwise do or buy.
  4. Your difference. One thing you do that others do not, and can prove.
  5. Proof. A fact, a review, a number.

Fill those five in a single sentence, and you have a line to test. Then test it with real people, not only with friends.

Test messages fairly

The program below compares three headlines on the same landing page, shown to 900 visitors each in a made-up test. It computes the signup rate for each, then asks two honest questions. First: is there any real difference among the three at all? It answers with a standard test for counts called chi-square, and for three groups the probability has a neat closed form. Second: is the best pair really different, and does it pass a stricter line, because comparing several pairs gives luck more chances?

import math

# MADE-UP test of three headlines: (visitors, signups)
tests = {"A: calm evenings": (900, 36), "B: 40 hours of glow": (900, 54), "C: handmade gifts": (900, 45)}
tot_n = sum(n for n, _ in tests.values()); tot_s = sum(s for _, s in tests.values())
p = tot_s / tot_n
chi = 0.0
for n, s in tests.values():
    for obs, exp in ((s, n * p), (n - s, n * (1 - p))):
        chi += (obs - exp) ** 2 / exp
pval = math.exp(-chi / 2)           # exact for 2 degrees of freedom
for k, (n, s) in tests.items():
    print(f"{k[:19]}: {100 * s / n:.1f}%")
print("chi-square:", round(chi, 2))
print("p-value, any difference:", round(pval, 3))

def z(a, b):
    (n1, s1), (n2, s2) = a, b
    pp = (s1 + s2) / (n1 + n2)
    se = math.sqrt(pp * (1 - pp) * (1 / n1 + 1 / n2))
    return (s2 / n2 - s1 / n1) / se
zz = z(tests["A: calm evenings"], tests["B: 40 hours of glow"])
p2 = math.erfc(abs(zz) / math.sqrt(2))
print("A vs B z:", round(zz, 2))
print("A vs B p-value:", round(p2, 3))
print("strict line for 3 comparisons:", round(0.05 / 3, 4))
print("A vs B passes strict line:", p2 < 0.05 / 3)
A: calm evenings: 4.0%
B: 40 hours of glow: 6.0%
C: handmade gifts: 5.0%
chi-square: 3.79
p-value, any difference: 0.15
A vs B z: 1.95
A vs B p-value: 0.052
strict line for 3 comparisons: 0.0167
A vs B passes strict line: False

The signup rates are 4.0 per cent for A, 6.0 for B and 5.0 for C. B looks best. But the overall test gives a chi-square value of 3.79 and a probability of 0.15, which means that differences this large would turn up about 15 times in 100 if all three headlines were really equal. That is not strong evidence. Comparing A and B directly gives a z value of 1.95 and a probability of 0.052, only just above the usual 0.05 line. And because we have made three possible comparisons, a stricter line of 0.0167 is sensible, which B does not pass. The honest conclusion is: B may be better, and the test has not shown it. Run it for longer, or test a bolder message.

Brand habits that last

  • Say the same few things. Repetition builds memory.
  • Be consistent. Same voice, same look, same promise in email, shop and social posts.
  • Keep promises. Delivery, quality and replies are brand, too.
  • Collect proof. Reviews and photos from real customers.
  • Avoid false claims. A message that cannot be kept damages trust and may break advertising rules in your country.
Common mistake: fishing for a winner

Testing many headlines and picking the winner by eye. The more you compare, the likelier it is that one looks good by luck. Decide your comparisons and your line before you look.

Statistics has limits

The chi-square test needs enough expected counts in every cell to be reliable. With small numbers use a different method, or collect more data. A test shows whether a difference is likely to be real, not whether it is big enough to matter.

Write your own positioning line with the five parts. Say it aloud to two people. Ask them to repeat it in their own words, and compare.

Practice for this lesson

Check what you learned

Five quick questions. Pick an answer and you will see straight away whether it is right and why.

What is positioning?

  1. Where your shop is
  2. A page layout
  3. A budget
  4. The place you want to hold in customers’ minds

It is how you stand compared with the alternatives.

What did the overall test say about the headlines?

  1. Not strong evidence of a real difference
  2. Proof that B is best
  3. Proof that A is best
  4. It could not run

The probability was about 0.15.

Why use a stricter line for three comparisons?

  1. It is a law
  2. To make winners easier
  3. More comparisons give luck more chances
  4. To save time

Each extra comparison raises the chance of a false win.

Did B pass the strict line?

  1. Yes
  2. It was not tested
  3. It tied with A
  4. No

Its probability was above the stricter line.

What is a sensible next step after the test?

  1. Declare B the winner
  2. Run longer or test a bolder message
  3. Drop all headlines
  4. Change nothing and stop

The test did not settle it.

Summary

  • Positioning is the place you hold compared with alternatives.
  • Say a few things consistently and prove them.
  • Test messages with a plan, and use a stricter line for many comparisons.
  • A test shows likelihood, not size or importance.

Next lesson

Lesson 27 covers privacy, consent and tracking limits.

Questions

What is the difference between brand and positioning?

Brand is everything people connect with you. Positioning is the specific place you want to hold in customers' minds compared with the alternatives.

What is chi-square?

It is a standard test that compares counts you saw with the counts you would expect if all groups were equal. It needs enough expected counts in each cell.

Are the test results real?

No. They are made-up numbers to show the method.

Progress is saved in this browser.