ABSTRACT One of the simplest ways to identify an exoplanetary transit is to phase fold a photometric time series upon a trial period – leading to a coherent stack when using the correct value. Such phase-folded transits have become a standard data visualization in modern transit discovery papers. There is no analogous folding mechanism for exomoons, which would have to represent some kind of double fold: once for the planet and then another for the moon. Folding with the planet term only, a moon imparts a small decrease in the surrounding out-of-transit averaged intensity, but its incoherent nature makes it far less convincing than the crisp stacks familiar to exoplanet hunters. Here, a new approach is introduced that can be used to achieve the transit origami needed to double fold an exomoon, in the case where a planet exhibits transit timing variations (TTVs). This double fold has just one unknown parameter, the satellite-to-planet mass ratio, and thus a simple one-dimensional grid search can be used to rapidly identify power associated with candidate exomoons. The technique is demonstrated on simulated light curves, exploring the breakdown limits of close-in and/or inclined satellites. As an example, the method is deployed on Kepler-973b, a warm mini-Neptune exhibiting an 8-min TTV, where the possibility that the TTVs are caused by a single exomoon is broadly excluded, with upper limits probing down to a Ganymede-sized moon.