How Should a Technician Handle Working a Diagnostic with Intermittent Failures?

When a technician encounters an intermittent failure during diagnostics, the best approach is to document every occurrence with exact conditions (temperature, driving pattern, DTC codes), reproduce the fault systematically under controlled conditions, and use data logging rather than guesswork to isolate the root cause. Intermittent faults account for roughly 30% of comeback repairs at dealerships that don't have a structured protocol—but dealers with repeatable diagnostic processes cut that figure to under 8%.

Why Intermittent Failures Are the Technician's Toughest Challenge

Intermittent failures live in a special hell. The customer drives in saying the check engine light came on, then turned off. You plug in the scanner and find nothing. You drive it for 30 minutes. Nothing. The vehicle sits in the lot overnight. Nothing. Then the customer picks it up, and three days later the light comes back.

This pattern wastes time and erodes customer trust faster than almost anything else in service. A typical scenario: a 2016 Subaru Outback with an intermittent P0101 (Mass Air Flow sensor circuit range/performance). The customer reports it happens when humidity is high and the vehicle has been idling in traffic. But when you test it in the parking lot on a dry afternoon, the sensor reads clean. You clear the code. Customer calls back two weeks later—same issue, different diagnosis by a less experienced tech, wrong part replaced, and now the customer's relationship with the dealership is damaged.

The root cause of poor intermittent diagnostics isn't usually the technician's skill,it's the lack of a repeatable framework. Most techs rely on intuition, pattern recognition from years of experience, or worse, trial-and-error parts replacement. Even experienced technicians struggle without a system.

How to Document Intermittent Failures Without Guessing

The first rule: stop relying on memory and the customer's vague description. You need written documentation of every clue.

When the customer describes the issue, ask these specific questions:

Exact conditions: Highway vs. city driving? Cold start or hot soak? Wet weather or dry? Idling, accelerating, or both?
Frequency: Does it happen every drive, once a week, or randomly? This matters more than you might think.
Duration: Does the symptom last seconds, minutes, or does it persist until the vehicle is shut off?
What triggers resolution: Does turning the key off and back on clear it? Does it go away after the engine warms up?
DTC codes: Even if the light isn't on now, what codes appeared the last time the customer saw it?

Document this in the Repair Order notes,not just a scrawl, but structured fields. This information becomes the blueprint for your reproduction strategy. If you skip this step, you're working blind.

Now here's where a lot of techs make a mistake: they assume intermittent means "random and unreproducible." Often it doesn't. Many so-called intermittent faults follow a hidden pattern. A loose ground connection might only cause a fault when the engine is under load and vibration peaks. A failing fuel pump relay might only drop voltage when the ambient temperature exceeds 85 degrees Fahrenheit. A corroded connector might only lose signal when moisture from humidity gets into it.

Your job is to find that hidden pattern.

Using Data Logging to Prove What's Actually Failing

This is where the diagnostic jumps from guesswork to engineering. Most dealership technicians have access to data-logging capability in their scan tool,either native in the DMS software or through a standalone diagnostic device. If you don't, this is a gap worth closing with your service manager.

Data logging captures real-time sensor values, PIDs (Parameter IDs), and fault flags over an extended period,sometimes hours,while the vehicle is driven under the exact conditions the customer reported. Unlike a static code read, a data log shows you when the fault occurs relative to other variables.

Example: A customer reports an intermittent misfire on a 2014 Ford F-150. You clear the code, set up data logging on:

Fuel trim (long and short)
Ignition timing advance/retard
Cylinder balance (if available)
Engine load and RPM
Intake air temperature
Fuel pressure
Spark plug voltage (on some platforms)

You drive it under the conditions the customer mentioned,highway acceleration to 65 mph, then cruise. The data log runs the entire time. If a misfire occurs, you don't just see "Misfire detected",you see exactly what the fuel trim, timing, and air intake were doing at that moment. That's the difference between replacing a $40 spark plug and replacing a $1,200 fuel pump.

One honest caveat: data logging assumes the vehicle will exhibit the fault during your test drive. If it doesn't, you're still stuck. But even a negative data log is valuable,it tells you the fault isn't related to normal driving patterns and might be tied to something more environmental (temperature storage, altitude change, specific load condition) or electrical (a marginal connection that only fails under exact vibration resonance).

Reproducing the Fault Under Controlled Conditions

Once you've documented the pattern, the next step is systematic reproduction. This might sound straightforward, but most shops skip it or do it haphazardly.

If the customer says the fault happens when the vehicle idles in hot weather, don't just sit in the lot for 10 minutes. Plan a reproduction test:

Run the engine at idle for 30 minutes with the A/C off (building heat).
Monitor coolant temperature, intake air temperature, and relevant sensor data.
If no fault appears, run it again the next day in direct sunlight.
If it still doesn't fail, consider whether the original condition was actually different (maybe the customer had the A/C on, or the vehicle was in traffic, not stationary).

A pattern we see across top-performing dealerships is that they'll spend 2–4 hours on systematic reproduction before moving to parts replacement. This sounds expensive,and it is, on the surface. But the comeback rate is so much lower that the math works out. You replace the right part once instead of replacing three wrong parts and having the customer come back twice.

Some faults require specific environmental conditions you can't easily create. A humidity-related electrical fault, for example, might need a climate chamber or overnight storage in a specific temperature range. Most independent shops and many dealerships don't have that. In those cases, document what you've tested, inform the customer of the limitation, and either offer a conditional repair guarantee (we'll fix it, and if the same code returns within 30 days, we'll re-diagnose at no charge) or request the customer return the vehicle the next time the fault occurs so you can perform live diagnostics.

When to Replace Parts vs. Keep Diagnosing

Here's the operational tension: diagnostic time costs money, but so do comebacks. How do you know when to stop diagnosing and commit to a repair?

A strong rule of thumb is this: if you've completed all of the following, you're ready to make a repair decision:

Documented the exact customer-reported conditions in writing.
Reviewed all stored and pending DTCs, freeze-frame data, and event codes.
Performed at least one controlled reproduction attempt under the stated conditions.
If reproduction failed, run data logging during normal driving and analyzed for anomalies.
Confirmed that the suspected component can actually cause the reported symptom (not just a coincidence).
Checked for technical service bulletins (TSBs) or known issues for that vehicle/year/fault combination.

If you've done all six and still don't have a clear culprit, then you're in the gray zone. At that point, you have options:

Replace the most likely component (based on TSBs, failure frequency for that model, and diagnostic data) with a warranty on that specific repair.
Request a follow-up appointment with the customer when the fault recurs, so you can perform live diagnostics.
Escalate to a factory hotline or regional technical specialist (some manufacturers offer this).

The dealers who get this right tend to be transparent about the limitation. They tell the customer: "We've tested the vehicle thoroughly and haven't reproduced the fault. Based on the data we collected, we believe it's most likely the fuel pressure regulator. We can replace it with a 30-day warranty,if the same code comes back, we'll re-diagnose at no charge and find the real cause." Most customers accept this, especially if you show them the data log printout.

Building a Diagnostic Workflow for Intermittent Faults

The best dealerships we see have a repeatable process baked into their RO system. This workflow lives in the service advisor's notes, the technician's job card, and sometimes in the DMS itself.

The workflow looks like this:

Intake documentation: Service advisor asks the six questions (conditions, frequency, duration, resolution, codes) and records them in a structured field or free-text notes that the technician must acknowledge.
Initial diagnosis: Technician scans for codes, reviews freeze-frame, and notes any relevant TSBs.
Reproduction attempt: Technician schedules a 1–2 hour test drive under the documented conditions. This is logged as a separate line item on the RO so the customer understands why the diagnostic time is higher than usual.
Data logging (if needed): If reproduction fails, technician sets up data logging and performs another drive cycle. Results are printed and attached to the RO.
Decision point: Technician and service advisor review findings and decide on repair, follow-up appointment, or escalation.
Warranty condition (if applicable): If a repair is made based on educated guessing rather than definitive proof, the RO clearly states: "Repaired under intermittent-fault warranty. If the same DTC returns within 30 days, re-diagnosis will be performed at no charge."

This is the kind of workflow Dealer1 Solutions was built to handle,structured diagnostics with clear handoff points, documented decisions, and warranty tracking. Even without specialized software, you can run this on paper and in your DMS notes. The key is consistency.

Common Pitfalls That Trap Technicians

Several patterns emerge when intermittent diagnostics go wrong:

Over-reliance on the scanner: A scanner shows you codes, but it doesn't show you why the code occurred. A P0401 (EGR flow insufficient) could be a clogged EGR valve, a faulty EGR position sensor, a vacuum leak, or a wiring fault. The code alone doesn't tell you which. Data logging does.
Skipping the TSB search: Many intermittent faults are known issues with published fixes. A 2015 Honda Civic with an intermittent P0128 (coolant thermostat) is probably the radiator fan relay issue Honda issued a TSB for. If you don't search, you'll spend two hours diagnosing what Honda already solved in 30 minutes.
Assuming "intermittent" means "electrical": Intermittent faults can be mechanical (a valve sticking when hot), fuel-related (a fuel pump relay that drops voltage under load), or sensor-related (a mass air flow sensor fouled by a dirty filter). Don't jump to electrical connectors just because the fault is intermittent.
Charging diagnostic time without explaining it: Customers balk at a $300 diagnostic for a vehicle that "only had a code that's now gone." If you don't explain that you spent three hours data logging and two reproduction attempts to *prove* the fault was intermittent and identify the cause, they think you're padding the bill. Show them the data. Show them the work.

Frequently asked questions

How long should I spend diagnosing an intermittent fault before deciding to replace a part?

Most dealerships budget 2–4 hours of diagnostic time for an intermittent fault, depending on the system and how reproducible the issue is. If you've completed all six steps in the diagnostic checklist,documented conditions, reviewed DTCs, attempted controlled reproduction, run data logging, confirmed the component relationship, and checked TSBs,you've done the job. Beyond that, you're in the zone of diminishing returns.

What should I do if I can't reproduce the fault during my test drive?

Document that you attempted reproduction under the customer-reported conditions and it didn't occur. Set up data logging during normal driving and analyze for any anomalies or pending codes. If nothing appears, discuss with the service advisor whether to request a follow-up appointment when the fault recurs, offer a repair with a 30-day warranty, or escalate to a factory technical hotline. A negative test drive is still valuable data.

Are there tools or software that make intermittent diagnostics easier?

Advanced scan tools with data-logging capability are essential,most modern DMS platforms include this. Some independent diagnostic tools offer extended logging and graphing that make pattern identification easier. The key is using whatever tool you have systematically, not expecting the tool to do the thinking for you.

How do I explain a higher diagnostic charge to a customer when the fault is intermittent?

Be transparent about the work: "We spent two hours attempting to reproduce the fault under the conditions you described, then ran a two-hour data log during normal driving. Here's the printout showing what we found." Show them the data. Most customers understand that intermittent faults take longer to diagnose than obvious ones. If you skip this explanation, they assume you're padding hours.

What's the difference between a pending code and a stored code on an intermittent fault?

A pending code (also called a one-trip fault) has occurred once but hasn't met the criteria to illuminate the check engine light. A stored code (two-trip fault) has occurred at least twice and triggered the light. On intermittent faults, pending codes are extremely valuable,they tell you the fault just happened and you can pull freeze-frame data showing exactly what conditions were present.

Should I replace the most likely component if I can't prove it's the cause?

Only if you document it clearly on the RO as an "intermittent-fault repair" with a warranty condition. State: "Based on diagnostic data and TSBs, we believe the cause is [component]. We are replacing it with a 30-day warranty. If the same DTC returns, re-diagnosis will be performed at no charge." This protects both you and the customer.