Tone of address: what research on formal and informal gives us

The decision every brand makes once

In German, French, Spanish and many other languages, copy has to choose between a formal and an informal way of addressing the reader. English does not. That is why this is the most discussed copy decision in those markets — and the one most often justified with studies that answer something else.

Those studies compare direct address with no address at all. The debate in those markets compares two forms of direct address with each other. On that, nothing has been tested.

What is supported: self-referencing

The well-studied part is self-referencing. Rogers, Kuiper and Kirker (1977) had 32 students judge trait adjectives in four ways; from the condition “Describes you?” participants later recalled 2.84 out of 10 words, from the semantic condition 1.33. Symons and Johnson (1997) pooled 129 effect sizes: a mean of d = 0.50. The d measures how large an effect is — 0.2 counts as small, 0.5 as medium, 0.8 as large. So self-referencing sits right in the middle.

The boundary conditions are unfavourable for ad copy. Three of them hit exactly the situation shop copy is in:

The last three are statistically indistinguishable from zero. Product names are nouns, and ad copy consists largely of desirability claims. The final figure does not transfer directly — an expected memory test is not an expected sales pitch — but it shows the effect hangs on processing being incidental.

On top of that: against reference to an intimate other the effect is only d = 0.20. Self-referencing is not a private road into memory, then, but the top end of a range that nearly closes once the other person is someone close.

Burnkrant and Unnava (1995) built the bridge to advertising — with exactly the contrast at issue here: “you know from your own experience …” against “one knows from experience …”. 61 and 75 students, a calculator from an invented brand, strong arguments throughout.

The finding is narrower than the short version suggests. Direct address helps — but only as long as the copy does nothing else that already prompts thinking:

For ad copy the last row matters most. Rhetorical questions are common in advertisements and on shop pages. Where direct address meets them, the effect reverses: readers’ net positive thoughts fell from 3.71 to 1.50. The authors explain it by tedium — two amplifiers at once push processing past the point where it stays agreeable.

Direct address is therefore not an amplifier for sales success but an amplifier for whatever the copy already says.

Why that does not answer the formality question

The studies cited come from English, which has only one “you”. Moving from an impersonal sentence to direct address raises self-referencing — that is the measured effect.

In languages with two forms the situation differs. The formal pronoun is just as direct as the informal one. Both address the person, both create self-reference. What differs is closeness and formality — a social dimension that English handles by other means.

So anyone saying “informal reads as more personal, that is established” is using a finding about the step from column one to column two in order to justify a step from column two to column three.

What exists instead

The nearest usable work is Packard, Moore and McFerran (2018), and it is broadly based: six studies, among them field data from 1,277 real customer-service emails linked to purchases over the following 90 days, plus lab experiments with more than 1,600 participants.

The result contradicts practice, in two directions at once.

In the field data, agent use of “I” predicts higher revenue (β = 1,245.76, p < .001); “we” (β = −201.75, p = .18) and “you” (β = 108.54, p = .33) do not. The β says how strongly each word choice goes together with later revenue; the p-value beside it says how well that link is secured against chance — from .05 it counts as meaningful. In the lab, satisfaction rose from 4.48 to 5.32 when, in real replies from six firms, only “we” was swapped for “I” — nothing else.

For the question of address, the null finding matters more. The authors tried four times to overturn it: with large samples, with stimuli that had worked before, and with a linguistic amplifier meant to foreground the customer. A Bayesian analysis across all five tests gives: “no effect” is more than a hundred times more likely than “there is one”. That is the stronger statement. Ordinarily a study can only fail to show an effect; this method supports the conclusion that the data actively argue for its absence. For a null finding that is an unusually solid position.

In one case address turns negative — not when the customer is the recipient of the action (“your order is being retrieved”), but when they are made the actor (“if you have your username, you can look into the account”). Satisfaction and purchase intentions then fall from 4.88 to 4.41, because the agent comes across as less caring and less ready to act. In the field data this applied to roughly a quarter of all uses of “you”.

Practice does the opposite throughout. Among 40 of the largest online retailers, 100 percent used “we” and 97.5 percent “you”, but only 45 percent “I”. The service staff surveyed thought that right too: 91.8 percent of managers preferred the “we” version, 83.8 percent the version with an explicit “you”.

Their subject is not formal versus informal but who appears in the sentence: the customer, the company or both. That is a question you can decide independently of the formality question — and one for which field data at least exist.

Where closeness backfires

An address claiming more familiarity than the relationship supports can provoke resistance. That is reactance. As an objection to a blanket informal tone, though, it is weaker than it sounds. The meta-analysis by Rains (2013) pools 20 experiments in which the threat to freedom was produced by commanding phrasing: “you must”, “you have to”, “do it”. Not one of them varied the form of address or the familiarity claimed. Reactance here is an analogy, not a finding.

What reactance research actually says about word choice concerns the imperative, not the form of address. Rosenberg and Siegel (2018) survey five decades of the field. The distinction it works with runs between controlling words — must, ought, should — and autonomy-supportive ones: perhaps, possibly, can, consider. Not between formal and informal.

The distinction has been tested. Dillard and Shen (2005) set two versions of copy against each other, on flossing and on binge drinking, with 196 and 200 students. The pressing version read, among other things, “as any sensible person can see, there is really no choice: you simply have to do it”, the restrained one “you may want to try it today”. Both versions address the reader as “you”. Here too the form of address was not the variable.

Two refinements from it are more usable for copywriters than the word lists. First, threat is not a property of the text but a judgment made by the reader — the same sentence lands differently on different people. Second, the authors separate three dimensions that reactance research usually collapses into one: how plainly the intent is named, how far the copy lets on that the sender believes they can steer the reader, and whether a reason is supplied. It is chiefly the second that produces anger; a reason softens the sense of intrusion.

The absolute values warn against overstating this, though. Even the crude pressing version reached only 2.98 out of 5 on perceived threat to freedom, and anger came to 0.90 out of 4. Significant, but small.

One practical finding from it does carry over. A closing line that explicitly leaves the decision with the reader — “the choice is yours” — damped reactance in one experiment, but only for the strongly pressing version; with the restrained version it did nothing. Copy that is already polite gains nothing from such a sentence.

Typical triggers are the automatic informal address to strangers, the first name in a bulk email, and the markedly casual tone in a situation the customer takes seriously — a payment reminder, say, or an outage notice. For copy in those languages this link is not verified.

Practical cases

What we deliberately do not recommend

• English pronoun studies as evidence for informal address. They measure the step to direct address, not the step to familiarity.
• Forced familiarity. First name plus informal pronoun plus exclamation mark to a first-time contact is the configuration in which closeness turns into resistance.
• Informal address as a substitute for an argument. Self-referencing amplifies what the copy says. If it says little, it amplifies that.
• Mixed address within one flow. Particularly jarring between interface and legal text.

Ethics check: nudge or sludge?

Limits and criticism