By "potential Pareto improvement", what Fallis means is a "Kaldor-Hicks improvement" or more simply an overall increase in wealth.
The argument by Kaldor (1939) and Hicks (1939) was that if there is an overall increase in wealth, then we could potentially compensate the losers, so that there is potentially a Pareto improvement. Therefore, a "potential Pareto or Kaldor-Hicks improvement" (i.e. an increase in wealth) is "just like" an "actual Pareto improvement".†
So, a free-trade agreement is a "potential Pareto" or "Kaldor-Hicks" improvement so long as it boosts our country's GDP. However, it is probably not an "actual Pareto improvement" because there will likely be at least one person in our country who suffers a loss due to the free-trade agreement.
And so, as you can tell and as Fallis says, most policy recommendations by economists are "potential Pareto" or "Kaldor-Hicks" improvements, but not "actual Pareto improvements".
†Note that this argument by Kaldor and Hicks is controversial, but is also one that most economists have gleefully latched on to, precisely because it is nearly impossible to ever have an actual Pareto improvement. Further reading: Feldman, 1998, "Kaldor-Hicks compensation".