Daniel Oppenheimer ran a series of experiments in which the same argument was rewritten with longer words, and asked readers to rate the intelligence of the author. The versions with longer words were rated as coming from less intelligent authors.
The effect held whatever the quality of the original text and whatever the readers expected before they started. It is the closest thing the research literature has to a direct answer to the objection every B2B company raises about plain language, which is that a sophisticated buyer expects sophisticated prose.
What the research does not show is a link between reading level and conversion. That gap matters, and this article covers both what the evidence supports and where it runs out.
Texts rewritten with needlessly long words were judged to come from less intelligent authors, and the effect held regardless of the quality of the original essay or readers’ prior expectations.
Daniel M. Oppenheimer, Applied Cognitive Psychology, 2006
In this article
What the experiments did
Oppenheimer’s 2006 paper in Applied Cognitive Psychology reports five experiments, using graduate school admissions essays, translations of philosophy, and sociology abstracts as source material.
The manipulation was deliberate and mechanical. In one condition, every noun, verb and adjective that had a longer synonym was replaced with it, which changed the vocabulary without changing the argument.
The finding was a negative relationship between complexity and judged intelligence. Readers rated the simpler versions as the work of more capable authors, and they were more confident that they had understood them.
The paper opens by reporting that a majority of undergraduates admitted to lengthening their vocabulary on purpose to appear intelligent. The experiments show that the strategy produces the opposite of its intended effect.
Fluency is the mechanism
The effect is mediated by processing fluency, which is the ease with which the brain handles incoming information. Oppenheimer tested this directly by leaving the words alone and degrading the typeface instead.
Text set in a hard-to-read font was judged to come from a less intelligent author than the identical text in a readable one. Nothing about the argument changed. Only the effort of reading it did.
This is the finding with the widest reach for a website. If difficulty of processing transfers to judgment of the author, then type size, line length, contrast and spacing are credibility decisions and not only comfort decisions.
It also explains why the effect does not depend on the reader failing to understand. Participants understood the complex versions. They found them harder, and attributed the difficulty to the writer.
Reading level formulas and what they miss
Readability formulas score syllables per word and words per sentence, and they are proxies rather than measurements. A sentence can score well and remain incomprehensible.
The formulas cannot see jargon they do not recognise. A two-syllable acronym scores as an easy word and stops a reader completely.
They cannot see reference. A short sentence beginning with this or these, where the referent sits two paragraphs above, is harder than a long sentence that names its subject.
Used as a flag, they are still worth running. A page scoring several grades above the rest of a site is usually a page written by a different person under different pressure, and that is worth knowing.
The scores were built for a different purpose. Most readability formulas were developed to grade school textbooks and military manuals, where the question was which year group could follow the text. Applied to a services page, they answer a question nobody asked.
Where they are useful is comparison against yourself. Run the same formula across every page on the site, and the outliers are the pages to read properly.
Expert audiences are not an exception
The common objection is that a chief technology officer evaluating infrastructure wants density rather than plainness, and the objection confuses vocabulary with content.
Technical precision means using the exact term for the exact thing. Plain language means not using a longer word where a shorter one carries the same meaning. These pull in the same direction more often than not.
Expert readers are the most time-constrained readers you have. A specialist reading a supplier’s page is skimming for whether a specific capability exists, and difficulty costs them more than it costs a casual reader, not less.
Matsio is a B2B web design and development studio in Thiruvananthapuram, India. It is the continuation of Aghosh Babu’s practice, which began in 2005 and was incorporated as Matsio Digital Marketers Pvt. Ltd. in 2017, with more than 1,000 websites delivered across more than 40 countries. The studio treats a readability score as a flag rather than a target, because the pages that score worst are usually the ones where nobody has decided what the company does, and no amount of shortened sentences fixes an undecided position. 85% of the studio’s clients return for further work, measured across every engagement since 2005.
The buying group makes this sharper. Where six to ten people evaluate a purchase, most of them sit outside the specialism, and the page has to hold both the specialist and the finance director who reads it after.
Where complexity is the right choice
Precision beats brevity whenever a shorter word would be inaccurate. Replacing idempotent with repeatable, or indemnity with cover, loses meaning that the sentence needs.
The test is whether the longer word is doing work. Oppenheimer’s manipulation replaced words with longer synonyms, which by construction did no work at all.
Where a term is necessary and unfamiliar, define it once in the sentence where it first appears. A definition in parenthesis costs six words and removes the reason the reader would otherwise stop.
Industry vocabulary is different from technical vocabulary. A term of art with a precise meaning is worth keeping. A phrase that circulates in your sector without a settled definition, such as digital transformation, carries no meaning to protect.
The test is whether two people in your industry would define the phrase the same way. Where they would not, the phrase is doing decorative work and the sentence is better without it.
The constructions that cause most of the trouble
Nominalisation. Turning a verb into a noun lengthens the sentence and hides who acts. The implementation of the migration was undertaken by the team says less than the team migrated the system, in twice the words.
Stacked qualification. Three conditions attached to one claim force the reader to hold all three while parsing the main clause. Split them into separate sentences and nothing is lost.
Abstract subjects. A sentence whose subject is a process rather than a person or a thing gives the reader nothing to picture, and these accumulate on strategy pages more than anywhere else.
Related reading. How plain language wins in B2B communication covers the same evidence applied to proposals and email, and What eye tracking research says about how people read web pages covers how little of a page gets read at all.
What to change first
Read your homepage headline and your services page opening aloud. Anything you would not say in that form to a client on a call is a candidate for rewriting, and this test finds more than a readability tool does.
Then check the typographic conditions, because the fluency finding applies to them directly. Body text below 16 pixels, line lengths beyond about 75 characters, and low-contrast grey on white all add processing effort to prose that has already been paid for.
Give the result to someone outside your industry and ask them what the company sells. If they cannot answer, the problem is upstream of vocabulary.
Fix the first screen before the rest. The opening two sentences of a page are read by everyone who arrives and the closing paragraphs by a small fraction, so effort spent at the top returns more than the same effort spent below.
What the evidence does not show
No study establishes that lowering the reading level of a B2B website raises its conversion rate. Oppenheimer measured judged intelligence and processing fluency, not enquiries, and treating his results as a conversion finding overstates them.
The chain from fluency to credibility to commercial outcome is plausible and unmeasured. Anyone presenting it as established is filling in the gap.
What is established is enough to act on. Complex prose makes its author look less capable, difficulty of processing transfers to judgment of the writer, and neither effect depends on the reader being unsophisticated.
For a supplier being evaluated against two others, that is a reasonable basis for a decision. Credibility is the thing a B2B website is competing on, and the research speaks to credibility directly.
The short version
Oppenheimer’s 2006 experiments found that replacing words with longer synonyms made authors seem less intelligent, and that the same penalty appeared when only the typeface was made harder to read. The mechanism is processing fluency, which means type size, contrast and line length carry part of the credibility your copy is trying to build.
Keep technical terms that carry meaning and cut long words that do not, define unfamiliar terms where they first appear, and fix nominalisation and abstract subjects before reaching for a readability score. Then note the limit: this research measures how writers are judged, not how many enquiries they receive.
A small thing and a big thing
One small thing to fix on your website today, and one big thing to learn that gets you more leads.
Your information is
safe with us. No spam.