How they really cook the books: Give thanks for the unsung skills of recipe testers

At 11am in a kitchen in London, six dishes are in various stages of preparation and cooking, the air thick with the smells of browning meat and baking pastry. But the food is not for lunch: this is neither a restaurant nor a party in the making. This cooking is being tested for publication in a cookbook on southern France by the renowned cookery writer Caroline Conran. Conran has enlisted two helpers to test her recipes. One is Beth Coventry, chef-patronne of The Wells gastropub in Hampstead. The other is me. Three days a week, we gather in Caroline's kitchen, aiming to perfect six dishes a day from Caroline's drafts. Those splattered, scribbled sheets vie for space on the kitchen table with chopping boards, measuring cups, electronic scales and piles of ingredients.

Caroline has cooked all these dishes at least once and we follow her drafts exactly. While cooking through the recipe, we consult the boss about anything we think might be done differently – less garlic, more stock, shorter browning – and make a note on the draft. When the dish is done, we stick forks or spoons in and taste together.

Sometimes it works perfectly the first time. Sometimes it needs tweaking to simplify, clarify or expand. Occasionally we decide we need to cook it again. A third attempt is never needed. We make notes as needed on the draft and the recipe is ready. It is a time-consuming process: Caroline's 250 recipes will take us around 20 working weeks. Failed attempts cost more time and more money. And every cookery writer has to steer a course between practical limitations and the need to make sure everything is exactly right. The cookery writer Sue Lawrence estimates that it costs £3,000 to develop 100 recipes and advances are not large in cookery-book publishing outside the small circle of big-selling media stars.

Magazines with a cookery department are well placed for exhaustive testing. Good Housekeeping, for instance, has three in its department and takes pride in triple-testing every recipe (full disclosure: I write for Good Housekeeping). Meike Beck, GH's Cookery Editor, describes the process as follows: "Once we are happy with a recipe we've developed, it is then tested again by another member of the team and then at least once more when we cook and style the food for photoshoots. We always say we triple test our recipes, but in truth it's often many more times than that.

"Testing is quite different to home cooking. You have to follow every step precisely, time everything and measure all the ingredients accurately and you need plenty of theoretical knowledge to help troubleshoot along the way." Such rigorous testing can be done, of course, by a lone cookery writer. The cookery writer Jill Norman, formerly Elizabeth David's editor, says that "from the time we worked together I saw her test recipes several times until she was happy with the outcome. I baked my way through much of the bread book and we compared notes", but the principal testing was done by David on her own.

Solitary working has its disadvantages. For one thing, it's a lot slower. More seriously, without extra eyes to catch slip-ups, the danger of letting through an error looms larger. Miss the full-stop key on your keyboard and 1.5 teaspoons of salt turn into 15 teaspoons. The Indian cookery writer Anjum Anand scaled down a recipe and initially assumed it didn't need testing, but then tested it after all and discovered she hadn't scaled down one crucial ingredient. Someone working less carefully would have let the recipe go out untested and therefore uncorrected. Scaling-down errors are even more crucial and more common in recipes originating in restaurant kitchens. Here the whole recipe can be changed fundamentally when cooked in home-sized quantities and complete retesting is essential.

In theory, an editor should catch mistakes. But it's difficult to do this without cooking the recipe, which is why Olive and BBC Good Food Magazine retest every recipe they consider for publication.

Lulu Grimes, food director for both magazines, says: "Taking book extracts is always an education. Not all recipes are 100 per cent accurate and we then have to either ask to print a different version of the recipe or drop it for another."

Sometimes mistakes creep in because of "typos or heavy-handed editing", she says. Cutting recipes to fit the space is particularly dangerous, as important details can be left out.

But it's often just lack of time and money that causes problems. Sometimes a book gets the green light at the last minute and is needed in six weeks.

Lulu Grimes says: "I think publishers probably don't understand what writing tested recipes involves." That suspicion is borne out by cookery writer Kathryn Hawkins. She tells of being asked by a publisher to test recipes from a TV show, only to discover that they were "a complete shambles in style. I could tell instantly that none of them were going to work and would need massive amounts of rewriting and testing". Recipes from TV chefs may be especially open to doubts about testing. Lulu Grimes says that chefs "who produce a book a year are on a constant treadmill. The ones that have good support staff or a ghost writer are fine, but the others are often all over the place and, as well as plagiarising, don't test". Sometimes recipes are cobbled together from what the cook has done on TV. If an expensive back-up team turns the improvised dish into a polished recipe, there's no problem. But some chefs expect overworked editors to do the recipe entirely on their own. One editor experienced in this kind of work calls certain prominent chefs "a disgrace when it comes to testing their TV recipes".

Jennifer John, managing director of a specialist food PR and marketing agency and a qualified home economist, broadly agrees.

"I have seen some terrible recipes from food writers and well-known chefs – cooking times and temperatures are wrong, or instructions potentially misleading. A successful recipe isn't just about a great idea but about communicating it accurately and clearly."

Has there been a drop in the quality of recipe testing? Judith Hannam of Kyle Books, a leading cookery publisher, doesn't think so.

"There is more competition, and in general the standard is as high if not higher as it was 10 years ago." Jill Norman thinks that if there are more substandard recipes around, it's simply because more recipes are being published – online as well as in print.

The challenges and pitfalls of testing should serve as a warning to wannabe cooks who want to publish their recipes. You don't need a team of testers like Caroline Conran's, but you do have to work meticulously. Don't think that cooking a dish once will necessarily make it a good recipe. Test it again. Watch out for your typing.

And don't be tempted to falsify your results. The Canadian writer Margaret Visser, author of the classic Much Depends on Dinner, quipped that when a cookery writer says: "I like to serve this dish with new potatoes and steamed broccoli," what they really mean is: "The only time I cooked this dish, I served it with new potatoes and steamed broccoli." You might be able to get away with it. But you might not.