DCMA-14 Metric 2 — Leads: Why They're Almost Always Wrong

A lead is a negative lag on a logic link, which lets a successor start before its predecessor finishes. In the DCMA 14-point assessment, schedule leads carry a goal of zero because they distort total float and can create resource conflicts. The pamphlet adds that, per the IMS data item description, "negative time is not demonstrable."

This article covers what leads do, whether the rule is universal, and how to model the same overlap without one.

What is a lead in a construction schedule?

A lag is a time offset on a relationship. A positive lag delays the successor after the logic is satisfied. A lead is a negative lag: a finish-to-start link with a lag of -5 days says the successor can start five days before the predecessor finishes.

The DCMA pamphlet defines the metric as logic links with leads divided by total logic links, with a goal of zero. It counts links, not tasks. A single schedule with 600 links and 12 leads has a lead percentage of 2%, and still misses the goal.

A simple example

Illustrative example, with assumed durations: framing takes 20 days, and drywall can begin once five days of framing remain.

The lead makes A2020 start on day 15 of framing. Nothing in the schedule says why five days remain, and if framing grows from 20 to 24 days, the lead still triggers at 5 days before the end, not when the physical condition is met.

Why does DCMA want zero leads?

The pamphlet gives three reasons for the metric: leads distort total float, they may cause resource conflicts, and, in its words, "per the IMS DID, negative time is not demonstrable." The last point means a reviewer can't verify what a negative offset represents.

Oracle's P6 help for the Negative Lags check adds a scheduling-mechanics reason: negative lags can affect the critical path, distort total float and cause unrealistic scheduling when durations change. The recommendation is a positive lag on a start-to-start relationship instead of a negative lag on a finish-to-start relationship. The Oracle page gives no numeric threshold for the check.

Is the zero-lead rule a universal scheduling principle?

No, and the sources disagree about how firm it is. Mosaic Projects, in its DCMA 14-point white paper, says the ban "is not based upon any universal scheduling principle." That is an opinion, but it matches the pamphlet's own rationale, which rests on the IMS data item description.

Ten Six Consulting, in an undated article, says DCMA has zero tolerance for negative lag in audited defense schedules and recommends a shorter finish-to-start with no lag. SmartPM cautions that rejecting negative lag may create owner liability. That is a secondary, legal-flavored claim, so treat it as a reason to have counsel review the contract rather than as a rule.

What this means in practice depends on who is reading your schedule:

How do I model overlap without a lead?

Choose the replacement that matches how the work physically happens. There are three standard options.

Start-to-start with a positive lag. If the successor really starts a fixed time after the predecessor starts, say so. In the framing example, drywall starts 15 days after framing starts: A2010 to A2020, SS +15. It pairs well with a finish-to-finish link if the successor cannot finish before the predecessor does. See relationship types for when SS and FF are defensible.

Split the predecessor. Break framing into A2010a (first 15 days) and A2010b (last 5 days). A2020 follows A2010a with a plain finish-to-start link. This is often the clearest option because each piece is an activity with its own dates and progress, and it can be statused.

Finish-to-start with no lag. If the overlap is small relative to the activity duration, shorten the predecessor's duration or accept the sequential logic. Ten Six Consulting recommends a shorter finish-to-start with no lag for audited schedules.

The same example, three ways

Which one you choose should come from field reality. If drywall starts when a certain floor area is framed, a split or an area-based breakdown matches that better than a day count.

How do I find leads in P6?

Look for relationships with negative lag values. Oracle's Schedule Check Report includes a negative lags section, and P6 shows lag in hours in the check output, so a -40 hour lag on an 8-hour calendar is a 5-day lead. Beyond the report, review the lag values on each relationship, sorting for negatives.

Then compute the metric yourself: count the links with negative lag and divide by all logic links. Record the tool, the date and whether you included links on completed activities, because tools differ in what they include in the denominator.

For each lead, record why it was used. The three answers you will most often see are a genuine overlap, a placeholder for work that should be its own activity, and an old fix for a date someone wanted to hit. The first can be remodeled with SS plus lag or a split. The second needs a new activity. The third is plain distortion and should be removed along with the date target behind it.

How do leads distort float in practice?

Take the framing example again with a 5-day lead. Because A2020 can start five days before A2010 finishes, the forward pass puts A2020's early start inside A2010's duration, and the backward pass applies the same offset in reverse. Float on both activities therefore reflects an offset that no one can observe on site, not just the work itself. DCMA's concern is that this makes total float a less reliable guide to what is critical.

The problem grows when durations change. A lead of -5 days on a 20-day predecessor represents 25% of the activity. If the predecessor is re-estimated to 8 days, the same lead now represents most of it, and the successor may be scheduled to start before the predecessor even could have produced enough completed work. Oracle's warning about unrealistic scheduling when durations change describes this situation.

Resource conflicts follow the same pattern. A lead pulls a successor earlier, so two crews may be planned in the same space at once with nothing in the logic to show it. A reviewer reading the resource histogram sees the overlap but not the reason.

What should I do before removing every lead?

Do not delete leads in bulk, because each removal changes dates. A short sequence avoids surprises:

1. Save a copy and note the project finish date and the critical path.
2. List every lead with its predecessor, successor, lag value and calendar.
3. Classify each one: real overlap, placeholder for work, or schedule compression.
4. Remodel the real overlaps using SS plus lag or a split, and give placeholders their own activities.
5. Reschedule and compare the finish date. If the finish moves, the lead was carrying time, and the project team should know that before the next submittal.

What do reviewers look for beyond the count?

A zero lead count does not prove the logic is sound. A scheduler can replace every lead with an SS link and a large positive lag, which moves the problem into the lag metric without changing the dates. The lags article covers that trade, and the logic metric matters because overlaps often hide missing ties.

Reviewers also compare updates. A lead that appears after a baseline is approved usually means someone is compressing a delayed schedule on paper. If you must carry one, note the reason on the activity, and expect to defend it.