Contract or full-time: what 214 engagements tell us about early retention
We followed 214 engagements from start to end or to nine months, whichever came first. Contractors did not leave sooner than employees. They left for different reasons, and the reasons were considerably more fixable.
Engagement flows from start to outcome across 214 placements. Ribbon width represents volume; the crossing pattern shows conversion between contract and permanent status.
What the data shows
- At six months, 91% of full-time and 82% of contract engagements were still active. Excluding contracts that ended on their planned completion date, the contract figure rises to 92% and the gap disappears.
- Of contract engagements that reached six months, 31% had converted to full-time. None of those conversions had ended by the time our window closed.
- The leading cause of early contract termination was unclear scope (38% of endings inside six months), not rate disputes (9%) or performance (14%).
- Full-time endings clustered around the first formal review, wherever a company placed it. Contract endings tracked project boundaries instead.
- Engagements where the company ran a written scope review in the first fortnight were 2.3 times less likely to end inside six months. It was the largest single effect we measured.
The standard objection to contract engineering is that contractors leave. It is stated so confidently that it rarely gets checked, and it shapes a lot of hiring decisions that would look different under scrutiny.
We followed 214 engagements placed through Engagetal from start date to end date or to nine months, whichever came first. About 60% were contract and 40% full-time. Nine months is the longest window we can honestly report: the company started operating in August 2025.
The survival gap is smaller than assumed
At six months, 91% of full-time engagements and 82% of contract engagements were still running. That is a real gap, and it is a fraction of the size most people guess when we ask them before showing the data.
It also mostly disappears under one adjustment. A meaningful share of contract engagements ended because the project finished: the engineer did what they were hired to do and left as planned. Counting that as attrition is a category error. Excluding planned completions, contract survival at six months rises to 92% against 91% for full-time, and the curves in Figure 1 sit almost on top of each other.
Contractors are not less committed. Some contracts are just designed to end.
How engagements survive their first nine months
Share of engagements still active by month, full-time against contract.
They leave for different reasons
Figure 2 is where the two models genuinely diverge, and it is more useful than the survival comparison.
The dominant reason for contract engagements ending early was unclear scope, at 38% of endings inside six months. Rate disputes accounted for 9% and performance for 14%. The thing companies worry about, being held up on rate, barely registers, while the thing they rarely plan for dominates.
Full-time endings look nothing like this. The leading cause was a better offer elsewhere (29%), followed by team or manager change (21%) and performance (17%). Scope clarity accounted for only 11%.
The distinction matters because the two failure modes have different fixes. Losing an employee to a better offer is a compensation and retention problem, largely external. Losing a contractor to unclear scope is a briefing problem, entirely internal, and considerably cheaper to solve.
Why engagements end
Primary reason for engagement ending, split by engagement type.
You are buying information about a person with money instead of buying it with interview time, and the information is much better.
Section 4 on conversion
The review-cycle effect
Full-time endings clustered noticeably around whenever a company held its first formal review, which in our sample ranged from month four to month nine. Contract endings showed no equivalent cluster; they tracked project boundaries instead.
We are describing a pattern in a small sample rather than confirming a well-established one, because our window is not long enough to see a full annual cycle. It is worth noting for anyone modelling headcount: the risk is not evenly distributed across the year, and a retention conversation held after the review is held after the decision was already made.
The week-two scope review
Share of engagements ending inside six months, with and without a written scope review in the first fortnight.
Conversion is the underrated path
Of contract engagements that reached six months, 31% had converted to full-time employment. None of those conversions had ended by the time our observation window closed, though we would not read much into that yet.
The mechanism is not surprising once stated. Both sides have worked together for months before committing. The information asymmetry that makes ordinary hiring risky has largely been resolved by the time the offer is made, and it has been resolved with real evidence rather than interview performance.
Against the mis-hire economics in ER-2026-05, contract-to-hire looks less like a compromise and more like a structurally better process for roles where it is practical. You are buying information about a person with money instead of buying it with interview time, and the information is much better.
| Reason for ending | Contract | Full-time | Median month | Preventable? |
|---|---|---|---|---|
| Unclear scope | 38% | 11% | 4.2 | Largely |
| Planned completion | 24% | 2% | 6.8 | n/a |
| Performance | 14% | 17% | 5.1 | Partly |
| Budget cut | 12% | 9% | 6.4 | Rarely |
| Rate or pay dispute | 9% | 3% | 5.7 | Largely |
| Better offer elsewhere | 8% | 29% | 7.6 | Partly |
| Team or manager change | 6% | 21% | 7.2 | Partly |
Table 1 on reasons for engagement ending, by type. Percentages are of endings within each engagement type and sum to more than 100% because up to two reasons could be recorded.
The single most effective intervention
Figure 3 shows the finding we would act on first. Engagements where the company produced a written scope review within the first two weeks, a document acknowledged by both sides rather than a verbal check-in, were 2.3 times less likely to end inside six months.
The effect held for both contract and full-time, and it was the largest single effect in the study. For contract engagements the early-ending rate fell from 26% to 11%.
Given that unclear scope is also the leading cause of early contract termination, this is close to a direct causal chain: the problem is that nobody wrote down what the work was, and the fix is to write down what the work is. It costs perhaps two hours. We now include a scope-review template in every contract engagement we place, which is a change this research produced.
Methodology and limitations
Sample: 214 engagements placed through Engagetal between September 2025 and May 2026, all followed to a definitive end date or censored at nine months. 128 contract, 86 full-time.
- Nine months is the ceiling. Engagetal was founded in August 2025. We cannot report annual retention, median tenure or multi-year conversion outcomes, because those do not exist yet. Where other studies report medians, we report survival at a fixed point instead.
- Exit reasons. Recorded by the Engagetal account manager after separate conversations with the company and the engineer. Where the two disagreed, the company stated reason is used and the case flagged; this occurred in 11% of endings and those cases are available separately.
- Planned completions. A contract counts as a planned completion only where the end date was specified in the original agreement and the work was delivered. Extensions that then ended are counted as terminations.
- Scope review. A qualifying review is a written document produced within 14 days of start and acknowledged by both parties. Verbal check-ins do not count, which is a deliberately strict definition.
- Limitations. Companies that write scope reviews are likely better organised in other ways, and we cannot separate the document from the disposition that produced it. The 2.3x figure should be read as an upper bound on the causal effect. Exit reasons are also self-serving in predictable directions on both sides.
Want the underlying data?
We share anonymised datasets and full statistical appendices with researchers, journalists and customers on request. Write to us and tell us what you are trying to work out.
Related reports
The 48-hour match
The median engineering role takes 41 working days to fill. Only 6.5 of them involve assessing anyone. We instrumented 287 hires to find out where the rest goes.
Read the report
The remote premium
Rate data from 640 engineers in 31 countries. The geographic discount is narrowing, and within-city spread already exceeds the gap between cities.
Read the report
Where the talent is
Nine thousand assessments across 38 countries in our first three quarters. Regional pass rates sit within seven points of each other, which makes country a poor filter.
Read the report