Reaching the ERC interview means the proposal has already survived a demanding scientific selection. The interview is not a second written proposal delivered aloud. It is a live test of whether the PI owns the scientific vision, understands the assumptions and can defend the programme when challenged.
The exact format is set by the panel. ERC guidance notes that interviews typically last around 30 minutes and include a presentation and questions. Applicants receive panel-specific instructions before Step 2.
Official sources: ERC Starting Grant, ERC evaluation webinar.
The presentation should establish the intellectual core quickly
The interview presentation has limited time. I would not use it to repeat every work package.
The panel already has the proposal.
The presentation should help them see the scientific architecture:
- the problem;
- the unknown;
- the central hypothesis or proposition;
- the decisive scientific tests;
- the possible outcomes;
- the field-level payoff.
The strongest presentation often feels simpler than the proposal because the structure is clearer.
Start with one concept, not a long introduction
The opening should give the panel something they can hold onto.
That can be one sentence and one figure.
For example:
Current models explain A and B, but fail in regime C. I propose that the accepted limit is not fundamental and can be tested through X.
The exact wording will depend on the field. The principle is the same: establish the scientific tension early.
A figure is valuable when it makes that tension visible. It is less useful when it merely reproduces the work plan.
Prepare for questions about the assumptions
A good panel will probe the weakest or most consequential assumptions in the proposal.
Prepare explicit answers to questions such as:
- What observation would falsify your central hypothesis?
- Why has this not been solved already?
- Which assumption is most fragile?
- What would you conclude if the expected effect is absent?
- Which competing explanation is most credible?
- Why is your method able to discriminate between the possibilities?
A strong answer does not need to pretend certainty. It should show scientific control.
Failure can be a strong answer when the science remains productive
One of the most important interview skills is being able to discuss failure without becoming defensive.
If the hypothesis fails, explain what the programme learns.
A useful answer structure is:
If A happens: the hypothesis is supported and we test generality.
If B happens: the effect is conditional, so we map the boundary conditions.
If C happens: the hypothesis is rejected, which elevates an alternative mechanism and changes the direction of the next objective.
This is the same scientific optionality that should already be visible in Part II.
The panel should see that uncertainty is part of the design, not something discovered only after an experiment fails.
Know the difference between scientific and execution risk
Expect questions about both.
A scientific risk is that the preferred explanation is wrong. That can still produce valuable knowledge.
An execution risk is that the programme cannot generate interpretable evidence because an instrument, fabrication route, dataset or recruitment plan fails.
For execution risks, show realistic alternatives.
For scientific risks, show meaningful outcome branches.
Confusing the two can make the project sound either overconfident or operationally fragile.
Be ready to defend the team architecture
The interview is also a test of whether the programme has been designed as a research group rather than a list of salaries.
You should be able to explain:
- why each role exists;
- which scientific objective it supports;
- where senior expertise is needed;
- what the PI must personally lead;
- how PhD and postdoctoral roles complement each other;
- how supervision load is managed;
- where collaboration is genuinely necessary.
My preference is to think in terms of a group capable of sustaining the programme, not simply hiring the minimum number of people required to execute tasks.
Defend the budget scientifically
A budget answer should connect money to the research question.
Instead of:
We need this equipment because it is advanced.
Prefer:
This instrument gives access to the regime in which the two competing models produce different predictions. Without that measurement, the central question cannot be discriminated experimentally.
The same logic applies to personnel, facilities, travel and computing.
The interview is easier when the budget follows the science visibly.
Prepare for the independence question
A Starting Grant is a major step in scientific independence.
If the project is close to your PhD or postdoctoral work, be ready to explain what is now intellectually yours.
That can be:
- a different central question;
- a new mechanism;
- a new experimental platform;
- a new combination of disciplines;
- a different scale of programme;
- a shift from contribution to leadership.
Remaining in the same institution or field does not automatically weaken independence. The panel needs to see the scientific direction as yours.
Do not over-answer
Long answers can create new problems.
A strong response normally has three layers:
- answer the question directly;
- give the scientific reason;
- add one piece of evidence or one consequence.
Then stop.
If the panel wants more, they will ask.
This matters because interviews reward precision. A technically brilliant answer can lose force if it takes two minutes to reach the point.
Rehearse adversarially, not ceremonially
A friendly rehearsal often produces false confidence.
The useful rehearsal panel is mixed:
- one expert close to the field;
- one strong scientist from the broader panel area;
- one person willing to challenge the assumptions;
- one person who watches only clarity and timing.
Ask them to interrupt.
Ask them to disagree.
Ask them to force short answers.
The goal is not to memorise responses. The goal is to expose where your scientific reasoning is still dependent on prepared wording.
Build a question bank from the proposal itself
Every strong claim creates a possible interview question.
For each page, mark:
- strongest claim;
- least supported assumption;
- largest cost;
- highest-risk objective;
- dependency;
- most surprising preliminary result;
- point where your approach differs most from the state of the art.
Turn each into a question.
You should also prepare the obvious strategic questions:
- Why ERC?
- Why now?
- Why you?
- Why this host?
- Why this team?
- Why five years?
- What is the single result that would most change the field?
Use backup slides as a scientific map
If panel rules and the interview setup allow them, backup slides should not be a second presentation.
They should support questions that are difficult to answer verbally:
- control experiments;
- sensitivity analysis;
- alternative mechanisms;
- team structure;
- detailed timeline;
- budget logic;
- preliminary data;
- technical architecture.
Each backup slide should answer one predictable question.
The interview is a test of ownership
The proposal may have benefited from colleagues, grants offices, consultants, collaborators and internal review. At the interview, the PI has to own the science.
The panel should feel that the PI can make decisions when the project stops following the original plan.
That is why I would prepare less for perfect delivery and more for scientific judgement under pressure.
Final rehearsal test
Before the interview, I would expect the PI to answer these without slides:
- What is the central question in one sentence?
- What result would most strongly falsify your hypothesis?
- What would still be valuable if that happened?
- What is the single largest execution risk?
- Why is your approach different from the state of the art?
- Why are you the right person to lead this programme now?
- What can this project make possible for the field in ten years?
If those answers are crisp, the presentation will usually become easier as well.