Giving feedback on pupils' completed work is a curious business. I couldn’t even begin to calculate the hours I have spent over the years writing comments in pupils’ books to tell them how to improve the answer to a question that they are unlikely to ever answer again. The problem is, as Dylan Wiliam explains in his chapter of What does this look like in the classroom?, that we seek to improve the work and not the student.

If we do try to leave comments that apply also to future pieces of work then these will be, by necessity, very generic. Writing, as I have done countless times, “you need to explain your points more” or “include examples as evidence for your point” is only useful feedback if you genuinely believe the pupil did not do these things as a deliberate choice and rather because they didn’t know how. Wouldn’t it be more useful to give them this advice before they started?

What would it look like if we gave feedback to pupils before they had done the work? “Feed-forward” rather than feedback? We could then ensure they had the information and advice needed to complete the work well the first time and to apply the same ideas to future pieces of work. Luckily, this type of “feedforward” already has a much more familiar name: modelling.

Why model?

I would like to suggest that we start to think of modelling as a form of pre-emptive feedback. Years of experience has taught me that I am likely to need to tell pupils to explain their points better, so rather than writing this on their work when the work has been done, I now model how to do this before they begin. Rather than giving feedback that they should have used examples, I now model how to do this before the error is made. By making my expectations clear at the start I can ensure that my classes practice getting their work right and avoid making these errors permanent.

Modelling in this way also helps to dramatically reduce workload as there are far fewer errors that need to be corrected, and much less need for individual written comments in books explaining where they went wrong in the hope that they will put it right in the future.

Finally, modelling like this helps our pupils to experience success. They can complete a piece of work to a very high standard, and a piece of work that doesn’t then end up with corrections scrawled all over it. It is very hard to build any kind of “growth mindset” if they don’t believe they are capable of meeting your expectations. Modelling and careful scaffolding helps them to see just what they are capable of.

Better modelling

At its most simple, modelling is showing pupils what it is you want them to do. But, as with all simple ideas, doing it well can be a bit more complex. Here are five ideas for getting modelling right.

1. Have a clear criteria for success

Before we can model what an excellent piece of work looks like, we need to decide on the criteria for ourselves. What are the characteristics for your particular subject? How might these expectations change over different key stages? How might they vary depending on the type of question? Share these criteria with the students and produce annotated examples of work to typify them.

2. Plan for errors

Start by considering the errors that are likely in a piece of work. These might be fairly generic issues or ones specific to a particular topic. Make a list of these issues that you will want to address before students start work.

3. Model live

Write a model answer yourself in front of the class. As you are doing this, explain your thought process. Show how you are overcoming the issues and common errors on your list: “Notice I am explaining this point by…” or “I am going to consider the economic case for sustainability here so will say…”

4. Model the errors

Ultimately, we want students to be able to self-regulate their learning. An important aspect of this is the ability to spot and correct errors in their own work. Take your list of common errors for a piece of work and use them to create a terrible model answer that includes as many of these mistakes as possible. Ask the class to identify the issues and suggest how it could be improved.

5. Remove the scaffolding

A potential issue with modelling is that pupils might be tempted to simply copy the model you have produced – after all, how much better could it be? One way to get around this is to write a model answer to a similar question but with a different focus. In geography this might mean modelling an answer to the question “How well prepared was Italy for the 2009 earthquake?” before asking them to answer the same question for an earthquake in Nepal.

Feedback is an important part of the learning process. We need to know where we went wrong and how to improve in the future. However, what is even more useful is being given clear instructions in advance so that we get it right first time and can apply the lessons again and again.

That's a model I can get on board with.

Mark Enser is head of geography at Heathfield Community College. His first book, Making Every Geography Lesson Count, is out soon. He tweets @EnserMark