Research and practical reviews

Impressions, clicks, and inquiries: how to read our open experiment

Igor Dushutin · BoardTempo field notes

A woman compares figures on a chart with notes in a notebook
AI-generated illustration, not a documentary photograph.

“A Pin got plenty of impressions” sounds good. But what did people do next? Did they save the idea, visit the site, read the article, ask for a niche review, or simply scroll past? With BoardTempo, I want to show the individual steps, not one impressive number.

What Pinterest measures

Pinterest reports those events separately. Opening an image inside Pinterest is not a visit to our website. Site visits, submitted inquiries, and any eventual purchases are later steps and need separate measurement.

Look at ratios, not just counts

To compare Pins, calculate the outbound click rate: outbound clicks ÷ impressions × 100%. On the site, use a separate measure: relevant inquiries from a defined group of visits ÷ those visits × 100%.

Compare like with like: the same time period, source, and counting method. Dividing a month’s total inquiries by one day of Pinterest visits won’t tell you the conversion rate. Decide what counts as an inquiry; a click on a button and a submitted form should not be recorded as two inquiries from one person.

  1. Impressions
  2. Outbound clicks
  3. Site visits
  4. Inquiries
Check each step separately.

Why the numbers can differ

Pinterest’s outbound click count and your site analytics need not match exactly. They measure different events. Page loading, analytics consent settings, measurement limitations, and the selected time period can all contribute to a difference. Check what the two reports measure before treating a gap as a fault.

Similarly, a sale on the day a Pin ran is not proof that the Pin drove it. Leave an inquiry unattributed when you can’t establish its source.

How I want to document the experiment

For each topic, connect the article, Pins, impressions and saves, outbound clicks, inquiries, and results. Look at Russian and English separately at first; the searches, audiences, and responses to images may differ.

Record what you tested, what you changed, what you expected, and what happened. Then the table becomes a useful record of the project, not just a scoreboard.

  1. What was measured
  2. Over what period
  3. What changed
  4. What remains unknown
Check each step separately.

What we can and cannot say so far

An email delivery test or a walkthrough of the form confirms a step in the process. Neither proves demand or growth in sales. A plan entry isn’t an article, a Pin idea isn’t a published Pin, and a published article isn’t a new customer.

We’re preparing a way to measure the full pilot. At the point described here, we don’t have a verified chain of results to present as a successful promotion case study. I want the open experiment to show the honest questions: What interested people? Where was the promise unclear? Which visits led to a useful conversation? Those answers will help us improve BoardTempo.

Sources