How to Start Logging Again After a Break (and What Gaps Cost)

2026-09-12

The short answer

Log the bad day roughly — one sentence, a guess, thirty seconds — because a wrong entry is worth far more to the arithmetic than no entry. And if you've already lost a fortnight, don't try to reconstruct it: log today, and treat the missing stretch as a gap in the data rather than a debt to be repaid.

The reason isn't discipline. It's that missing days are never a random sample of your days. You skip the wedding, the airport, the Friday that got away from you — and what's left in the file is a flattering, orderly version of how you eat.

What does a gap actually cost?

Suppose you log 22 days out of 30 and your logged average comes out at about 2,050 kcal. The eight you missed were the pub, the birthday, two travel days and a week where work went sideways. Those days weren't 2,050. They were 2,600, 3,100, 2,800.

Your real average was somewhere near 2,250. The record says 2,050.

Now look at what that does downstream. If your weight held steady across the month, the honest conclusion is that you maintain on roughly 2,250. The record instead concludes you maintain on 2,050 — and any target built from it inherits the error. You set what you believe is a 300 kcal deficit and you are actually running about 100. Three months later the scale has barely moved and nothing in the log explains why. The log is the thing that's wrong, and it's wrong in a specific, predictable direction: gaps always bias downward, because nobody skips the day they ate a salad.

This matters more for any system that infers your energy needs from your own data rather than a formula. Day So Far works that way — it compares what you logged against what the scale did over about a fortnight — and a fortnight with four holes in it is a fortnight of confidently understated intake. The estimate doesn't know the days are missing. It only knows what it was given.

Is a rough guess really better than nothing?

Do the sum. Say the real total for a chaotic day was 3,400 kcal. You write down a hurried 3,000. You're out by 400, which across a 30-day month distorts your average by about 13 kcal per day.

Skip it entirely and the day doesn't vanish — it silently takes the value of your average logged day, maybe 2,100. Now you're out by 1,300, which distorts the monthly average by about 43 kcal per day.

A guess that's badly wrong is roughly three times better than a blank. That's the whole argument. You don't need accuracy on the difficult days; you need presence.

How do you log a day you can barely remember?

Work in courses and anchors, not ingredients. You are not trying to reconstruct the meal, you're trying to land in the right thousand.

Useful anchors, all approximate and all the ordinary published figures:

A pint of standard lager: about 180–200 kcal
A 175 ml glass of wine: about 160 kcal
A restaurant main course: about 800–1,000 kcal, more if it's fried or creamy
A portion of chips: about 400–500 kcal
Two slices of takeaway pizza: about 500–600 kcal
A croissant: about 230 kcal
A standard chocolate bar: about 230 kcal
A 25 g bag of crisps: about 130 kcal

Use whatever evidence you have. Card statements tell you where you ate. Photos have timestamps. Group chats often contain the menu. If two numbers seem equally plausible, take the higher one — food you can't remember eating is nearly always food that got missed, not food that got double-counted.

Then write it as one line rather than six entries. "Pub roast, three pints, shared chips, sticky toffee pudding" is a legitimate log. An app that reads plain sentences will turn that into a number in a few seconds, which matters mainly because the alternative — searching a database for six items you half-remember — is exactly the friction that makes people give up on the day.

A half-logged day is the worst outcome of all. Breakfast recorded, everything after 2pm missing, and the file now contains a 600 kcal day that never happened. If you're going to abandon a day mid-way, either finish it approximately or leave it entirely blank.

What should you do after weeks or months away?

Start with today's food. Don't backfill from memory — three weeks reconstructed from vague recall introduces false precision, which is more corrosive than an honest hole, because you'll trust it.

The important thing is that your old target probably doesn't apply any more, and re-adopting it is the most common way a restart fails. Your weight has changed. Bodies need less energy at a lower weight and more at a higher one. Your activity has likely changed too — a different job, a different season, an injury that's healed or one that hasn't.

So re-measure before you re-decide. Two things, in this order:

Weight, as a trend. Several readings across a week, taken at a consistent time, and read as an average. A single morning number carries a kilo or two of noise from salt, water and timing.

Your actual current intake. Log for about two weeks without deliberately changing anything. This feels pointless and it isn't — it's the only way to find out what your normal now is, as opposed to what it was in March. Anything that learns your energy needs from intake versus scale movement needs roughly that long to produce a number worth acting on, and needs the run to be reasonably complete. Patchy coverage doesn't just delay the estimate, it skews it.

Only then think about direction. A target chosen before the baseline exists is a guess dressed as a plan.

The record is an instrument, not a report card

The all-or-nothing failure mode runs on a category error: treating the log as a record of behaviour that a bad day disqualifies you from keeping. It isn't. It's a measuring device, and a measuring device you only consult when things are going well tells you almost nothing.

There's no penalty for a high day inside the data. There's no credit for a low one. There's only coverage — how much of your actual eating the file has seen — and coverage is the single variable that determines whether any conclusion drawn from it is worth having.

Which also means the next day needs no correction. Eating less to offset yesterday doesn't repair the record; it adds a second unrepresentative day and makes the underlying pattern harder, not easier, to see. Log the heavy day. Log the ordinary one after it. The average will take care of itself, and it will be a real average, which is the only kind that's any use.