Make the next step
obvious.

01 / The decision

A useful choice should meet the person where they are.

Someone opening a nutrition and training app may want to log a meal, record a workout, or get a plan. Asking them to explore every tab before doing that adds work. Our design goal was to put a useful next action directly into that first moment.

In MyBodyLab 2.7.4, the fresh Home screen made tracking the clear invitation. Two large cards said “Track Food” and “Track Workout.” Profile setup was a smaller choice below them.

The current entry introduces a different priority: a guided plan. Real MyBodyLab screen artwork explains the destination. A large white “Build my plan” button offers the next step, while food and workout remain available as quieter links.

This combines product thinking with AttentionProof. We choose what should matter in context, then inspect whether the visual treatment supports that choice. The heatmap gives us evidence to discuss before the next release.

02 / The prediction

What the two heatmaps actually show.

We ran both English captures through the same AttentionProof scoring pipeline. The values below are each region’s share of the prediction across its full image. They are not percentages of users, viewing time, or taps.

The illustration and main prompt receive more predicted attention than the plan button: 31.8% and 27.6%. The strongest point falls inside the pictured phone. That gives us a concrete review question: does the artwork help someone understand the plan, or draw attention away from starting it?

A larger region can collect more predicted attention simply because it occupies more space. The secondary links are more salient per unit of area than the white button. Size, contrast, words, imagery and task relevance need to be considered together.

Compare the priorities inside each frame. The captures contain different surrounding UI, so their percentages do not establish a before/after improvement.

04 / Recorded choices

Food first. Workout next. What happened after matters.

An earlier study reconstructed one real 2.7.5 signup from named app events. The Home choice components were unchanged from 2.7.4. Its Russian-language fixture gave Food 17.6%, Workout 15.2% and profile setup 5.6% of predicted attention. This is a separate capture from the English comparison above.

1. First Home choiceFood

   The person selected food three seconds after the recorded Home entry.

2. After cancelling the cameraWorkout

   Workout was selected one second after cancellation. Neither route produced a saved entry in this trace.

3. Later in the same sessionOther areas were explored

   The person reached Plan and Day. The record does not support a claim that they never left Home.

These events show which named controls were selected. We did not record gaze or touch coordinates, and the fixture was not a recording of that person’s screen. The agreement is descriptive evidence from one person, not a validated click predictor.

Inspect the earlier study’s exact overlay ↗

05 / The wider usage context

Staying on Home can be useful. Check what gets done there.

Our working idea is that people may stop exploring once a visible choice solves their immediate need. A well-placed action can reduce the need to navigate. The logs can show activity; they cannot tell us that this was someone’s reason for staying.

The 2.7.5 observations span September 3–12 across 14 installations recording Home. The 2.7.6 observations span September 11–12 across four. These are repeated sessions, not counts of distinct people.

Of the 226 earlier-version sessions without another recorded main tab, 31 included a logging or first-value event. This supports Home’s role as an action hub. Missing lifecycle and navigation events mean we cannot treat every other session as either satisfied or abandoned.

The current version’s four observed App Store installations had all been observed on older versions. No Welcome events were recorded for them. Missing Welcome records do not establish whether onboarding was completed. They do not form a new-onboarding cohort, so there is no measured new-onboarding conversion result here.

06 / Try it on your creative

The same review loop works for ads, UI and visuals.

1. Name the job. For an ad, it might be noticing the product and understanding the offer. For an app, it might be finding the next useful action.
2. Capture the actual image. Keep language, device, crop and state consistent when testing a specific design change.
3. Inspect the prediction. Run the image through AttentionProof. Compare the original and overlay, including anything competing with the intended focus.
4. Change one priority deliberately. Try a clearer label, stronger contrast, quieter artwork or a more useful arrangement. Then recheck.
5. Measure what happens next. Use recorded actions and task completion for UI; use campaign outcomes for ads. A visually prominent action still needs to lead somewhere useful.

For iOS, Android, Mac and web apps, the input can be a screenshot. Through MCP, Claude or Codex can send that image, inspect AttentionProof’s result and help revise the design. This case documents iOS; the workflow can be applied elsewhere without assuming the same outcome.

Apple’s button guidance recommends using prominence to distinguish a preferred action. Nielsen Norman Group’s visual hierarchy guide explains how scale and contrast communicate priority. Neither replaces observing whether someone completes the task.