Ask a web team how long a page redesign will take and you will get a number. Ask the same team how long the last five redesigns actually took and, with luck, you will get a different number. The distance between the two is where schedules and budgets are lost. This list sets out eight website tasks where hours guessed and hours actually spent diverge most often, with a practical way to measure each and adjust the next estimate. The method is the same throughout: record the guess, record the actual, track the ratio.
1. Content migration
Moving pages from an old platform to a new one looks mechanical, which is why it is underestimated. The hours go on formatting repair, broken embeds, redirects and the discovery that a quarter of the old pages need rewriting. Track hours per page rather than per project. If your first ten pages cost forty minutes each against a guess of fifteen, your ratio is 2.7, and the remaining two hundred pages need to be planned accordingly.
2. Template and theme customisation
The base theme is easy; the exceptions are not. Custom archive layouts, odd header variants and the one landing page that needs a different grid all appear after the estimate has been agreed. Keep a tally of exceptions found per template. Teams that log this often find that the number of exceptions, not the size of the design, predicts the overrun.
3. Plugin and extension integration
Connecting a form tool, a search add-on or an analytics layer is typically guessed at an hour. Real elapsed time includes reading documentation, resolving conflicts with existing extensions and testing on staging. Record separately, for every extension, the time spent on first installation and the time spent debugging afterwards. The second figure is usually the larger, and it is the one that never makes it into the quote.
4. Service and landing page copy
Writing is the task most people believe they can predict, and the one where review rounds inflate the clock. A piece of copy that takes two hours to draft may take another three across feedback. If you want a refresher on what a persuasive page needs before you guess, this article about service pages that turn browsing into client enquiries covers the elements worth listing. Log draft time and revision time as two lines.
5. Trust pages such as About and contact
These pages seem small and so get tiny estimates, yet they require input from people outside the web team, such as approvals, photos and wording sign-off. Waiting is the hidden cost. For a view of why those pages matter, see the case for a strong About page as a trust signal. In your log, mark waiting hours separately so you can see how much of the overrun was not work at all.
6. Search and metadata clean-up
Titles, descriptions, headings and image text across a medium site take far longer than the first count suggests, because each page needs reading and a decision. Sample twenty pages, time them, and multiply. This is a classic place to apply a scaling check: if the sample varies widely in time per page, widen your range rather than trusting the mean.
7. Testing across devices and browsers
Quality checks are often estimated as a flat afternoon. In practice the time scales with the number of templates and the number of defects found, and each defect triggers a fix and a re-test. Record the number of defects per template as well as hours. A defect rate that is high in the first round tells you to expect a second and third round; budget for them from the start of the next project.
8. Launch and post-launch fixes
Go-live day carries a long tail: redirect surprises, caching quirks, a form that sends nothing and a menu that breaks on one phone. Teams that treat launch as a single event under-plan the weeks after it. Log the hours spent in the first fortnight after launch against the build hours; many teams discover a stable percentage, which then becomes a standard line on every new project plan.
How to run the tracking without drowning in admin
A spreadsheet with five columns is enough: task, guess in hours, actual in hours, ratio, and a note field for the cause of any miss larger than 1.5. Fill the guess before work begins. Group tasks into the eight types above, or your own, and compute a median ratio for each, since a single weird project can distort an average. Medians are sturdier for small samples.
After about ten finished items in a category, start applying the median as a correction factor. Before that, the data is a hint, not a rule. Report the factor with its sample size, for example “migration 2.1 over eleven projects”, so everyone knows how firm it is.
An example log for one project
To make the method concrete, imagine a mid-sized brochure site rebuilt on a new CMS. The team’s sheet after launch might read: migration, guessed 40 hours, actual 96, ratio 2.4; template work, guessed 30, actual 41, ratio 1.4; plugin integration, guessed 12, actual 25, ratio 2.1; copy, guessed 24, actual 38, ratio 1.6; metadata, guessed 10, actual 19, ratio 1.9; testing, guessed 16, actual 22, ratio 1.4; launch, guessed 8, actual 18, ratio 2.3.
Total guessed hours were 140 and total actual were 259, an overall ratio of 1.85. Seen alone, that figure looks like a verdict on the team. Broken down, it shows two clusters: three categories near 1.4 to 1.6 where the team’s skills are solid, and three above 2 where uncertainty lives. The next quote can apply 1.5 to the first group and 2.2 to the second, and the sum will land far nearer the truth than a single blanket uplift.
What the numbers usually reveal
- Work that depends on outside input carries the largest ratios.
- Repetitive, templated work is the most predictable, with ratios close to 1.
- The ratio for a task type tends to fall over the first year as the team learns, then flatten.
- Large misses cluster around first-time tools, not first-time people.
A short note on outside reading
Teams who want some reading outside delivery work sometimes visit Cotocus Foundation’s own site, where a nonprofit organisation presents its work through pages for programs, research, community and a blog. It is not an estimating resource, merely a place to see how an education-minded organisation lays out its sections, which can spark ideas for your own site structure.
Pitfalls in the measurement itself
Two traps catch careful teams. The first is mixing scope changes into the ratio: if a client adds ten pages halfway through, the extra hours belong to a revised guess, not to estimation error. Re-baseline the guess in writing when scope moves, and keep the original beside it. The second is logging hours from memory on Friday afternoon. Time notes made a week later are consistently low, which quietly pulls ratios toward 1 and flatters the team. Capture hours daily, even roughly.
Putting the list to work
Pick one category from the eight, the one that has hurt your last project the most, and start logging there. Do not try to measure everything at once. After three projects you will have a ratio you can defend in a planning meeting, and the conversation changes from “can you do it faster” to “here is what the last three took”. That shift, from opinion to record, is the real return on a few extra minutes of logging per task.