The Oklahoma City Thunder are a very good offensive team — so good, actually, that they average 107.5 points per 100 possessions, the best offensive efficiency in the league. Despite their overall success, the Thunder struggle in one specific area: scoring late in close games. As the game gets tight and the clock winds down, Oklahoma City becomes less efficient on the offensive end. Their inability to score late in games is what doomed Oklahoma City last postseason, and with the playoffs approaching, Thunder fans should be worried about their team getting consistent offense late in tough, grind-it-out games against quality opponents.

Using the new NBA.com/Stats tool that lets us look at crunch-time situations, you can tell that the Thunder haven’t gotten much better (though they have made an effort to be more creative) late in close games this season. Below is a chart of the Thunder’s field goal percentage, 3-point percentage, and 3-point rate (the percentage of total shots that are 3-pointers):

What’s troubling about both the numbers and the tape is that the Thunder are not only shooting worse, but their shot selection gets worse as well. From what I have seen, this poor shot selection and reliance on the 3-point field goal is their biggest late-game problem. Despite being an average 3-point-shooting team — their 35.8 percent from behind the arc is good for 11th in the NBA — they seem to rely heavily on the 3 as the game gets tight. Ignoring the last two categories, where the Thunder need to take 3-point shots when down by three with less than 30 seconds left, you can see how a reliance on the 3-ball hurts both their shooting percentages and efficiency. They go from shooting 3s 25.8 percent of the time overall, to shooting them 37.1 percent of the time when down five with five minutes left, and 42.3 percent of the time when down five with three minutes to go. Looking through some of Oklahoma City’s late-game possessions, it becomes clear that these 3-point attempts are a result of poor decisions, poor play design, and poor work from the players off the basketball. Here are two examples from a game a few months ago against Portland:

On this particular possession, there are eight seconds left on the shot clock, and Oklahoma City is inbounding the basketball from the baseline. They initially look to pass the ball in below the free throw line, but when no one is open, the ball goes to the safety valve, Russell Westbrook. As Westbrook catches the ball, he has two teammates, Kendrick Perkins and Kevin Durant, coming toward him to set ball screens. Instead of using either screen and attacking the rim, something Westbrook can easily do with the time left, he settles for a 3-point shot off the dribble with six seconds left on the shot clock.

In this clip, the Thunder want to post Kevin Durant up against Gerald Wallace. Durant starts fighting for position on the block, but Wallace is able to push him back outside the 3-point line. As a result, everything breaks down. Perkins tries to salvage the possession by setting a screen for James Harden, who pulls up for a 3 with 16 seconds left on the shot clock.

Three-point shots late in games are fine, especially if they are coming off sets designed to get an open look for a good 3-point shooter. In the two examples above, and most of the time for the Thunder, the late-game 3-point attempts are not designed. In the first clip, Westbrook makes the catch about 45 feet away, walks up to the 3-point line, and lets it fly. In the second, Harden forces up a tough 3-point attempt early in the clock simply because there isn’t anything else going on.

Despite all this, the Thunder are starting to show some improvement, including the screen-the-screener play I linked to earlier. Although the game against Miami wasn’t close late, Oklahoma City showed it can produce buckets late in games without having to rely on isolations. This could give teams nightmares in the playoffs:

This play starts with Westbrook bringing the basketball up the court and quickly pitching it ahead to Harden. After making the pass, Westbrook cuts to the opposite side of the court.

Once Westbrook clears out, Harden gets the ball to Perkins, who was trailing the play. As Perkins makes the catch, Westbrook runs by him and gets a dribble handoff.

Perkins then heads straight toward the block to set a pin-down screen for Kevin Durant. Durant does a good job of setting up his man for the screen, something he has really improved upon this season, and then comes off it.

Perkins’s man, Chris Bosh, then extends himself out in order to stop Durant from curling off the screen and getting an open shot. This allows Perkins to slip the screen and dive to the front of the rim.

Westbrook does a good job of keeping his head up and identifying the slipped screen, and he hits a wide-open Perkins in the paint for a strong dunk. Here is the play in real time:

Once a team is beaten by a slip like that, they are going to be hesitant in guarding the original option so hard. The result is that the original option opens up, which is what happened against Miami.

Here, we have the same play. Although it isn’t run as crisply, it frees up Durant to make the catch off the pin-down screen. After some more very good ball movement, the ball goes to Durant for the wide-open 3. This is a 3-point shot late in the game that does come off a set. More importantly, it comes off ball movement.

This isn’t a brilliant play call that will work for every single team. It’s actually a very simple action. But when you have three scorers like Harden, Westbrook, and Durant, intricate plays aren’t necessary to get guys open. What is necessary is not relying on isolations, and I think Scott Brooks is starting to understand that. Even a few simple late-game sets should be enough to scare opposing teams. The action I would love to see more of from the Thunder, especially late, is using Kevin Durant as the screener in pick-and-rolls with either Russell Westbrook or James Harden, something Durant has done on just 3.6 percent of his possessions this season. Again, it’s simple, but it is pretty damn effective. This season, Durant is posting a PPP of 1.213 as a roll man, good enough to put him in the 87th percentile (on 54.1 percent shooting). When perhaps the best scorer in the NBA is the screener on a play, it puts the defense in a tough position:

In most screens that the Thunder set, Kendrick Perkins is the screener. This allows teams to hedge off him and force the ball out of Harden’s and Westbrook’s hands. With Durant as the screener, teams are forced to pick their poison. Hedging out on Westbrook or Harden means Durant is going to be open popping out or rolling to the rim. Sticking with Durant means Westbrook or Harden can get to the rim with a head of steam. The defense is put in a tough situation either way, and it’s these types of simple sets that can give the Thunder the late-game advantage they’ve been lacking.