Digital Agency for Tunbridge Wells Businesses

Changing the Website on Opinion

Most website changes are decided in a meeting. Somebody dislikes the hero. Somebody read that pages should be shorter. A competitor did something and it looked good.

The change gets made, everyone moves on, and nobody ever establishes whether it helped. Six months later a different opinion prevails and it changes again.

This is not incompetence; it is the absence of an alternative. Without evidence, the loudest or most senior view wins by default, and there is no mechanism for anyone to be shown wrong.

Starting With a Question Rather Than a Change

Useful improvement work starts by asking what is actually happening rather than proposing what to do.

Which pages do people arrive on? Where do they go next, and where do they stop? Are there pages attracting attention that produce nothing? Is there a step where a noticeable proportion of people disappear?

That reframing matters. "The homepage needs redesigning" is a proposal. "Most people arriving on the homepage leave without visiting a service page" is an observation — and it can be investigated, explained and addressed in several ways, of which redesign is only one.

What the Data Can and Cannot Tell You

Analytics is good at showing what happened and poor at explaining why.

It will tell you that people leave a particular page, that a journey has a drop-off, that one page is visited far more than another. It will not tell you whether that is confusion, boredom, an expectation mismatch or somebody being interrupted.

That distinction keeps expectations sensible. The numbers identify where to look. Working out why usually requires something else — observing behaviour, asking customers, or simply reading the page carefully with the drop-off in mind.

Behaviour Tools, Where They Are Available

Session recordings and heatmaps, where a business has them, sit between the two.

They show how far people scroll, what they click, where they hesitate, what they attempt to interact with that is not actually interactive. Occasionally they reveal something no amount of aggregate data would surface — people repeatedly clicking an image expecting it to be a link, or stopping just above something important.

Worth using with some care. A handful of recordings is anecdote rather than evidence, and it is easy to over-interpret one person's confusion. They are most useful for generating hypotheses that can then be checked more broadly.

Forming a Hypothesis Worth Testing

A hypothesis is a change plus a reason plus an expected effect. "Make it clearer" is not one.

Something like: people arriving on this page are leaving without reaching the enquiry step; the likely cause is that the next action is below several screens of explanation; moving it higher should increase the proportion who reach it.

The value is that it can be wrong. If the change is made and nothing shifts, you have learned the explanation was incorrect — which is genuinely useful and impossible when the reasoning was never stated.

Changing One Thing at a Time

The temptation, having decided a page needs work, is to improve everything about it at once.

Do that and you will not know what mattered. If the page performs better, was it the heading, the structure, the imagery or the new call to action? If it performs worse, which change caused it?

Sequential, targeted changes are slower and considerably more informative. Where traffic is high enough for proper testing, running variants against each other is better still — though many business sites do not have the volume for that to produce meaningful results, and it is more honest to say so than to run a test that cannot conclude anything.

Comparing Fairly

Before-and-after comparisons mislead easily.

Compare a month against the previous month and seasonality, working days and campaign activity all confound the result. Compare against the same period last year and most of that falls away. Give it long enough to accumulate meaningful numbers, and remember that other things changed too — a campaign, a competitor, the weather.

Where the result is genuinely unclear, saying so is the right answer. A good deal of reported improvement is noise interpreted generously, and treating an inconclusive result as a success removes the possibility of learning anything.

Asking People as Well as Measuring Them

Quantitative data tells you what; people tell you why.

Useful sources cost very little: what enquirers say when they get in touch, what questions come up repeatedly, what customers say confused them, and occasionally an on-site question asking whether visitors found what they came for.

Where the numbers and the comments disagree, that is informative rather than inconvenient — it usually means the metric is measuring something adjacent to what actually matters.

Working With a Kent-Based Digital Agency

Maidstone Digital is based in Maidstone and works with businesses in Tunbridge Wells, elsewhere in Kent and further afield. We do not have premises in Tunbridge Wells. Most work is carried out remotely and reviewed through shared data and regular calls, though we are happy to meet where useful.

Improvement work usually touches design and development together. If your interest is specifically organic search — particularly diagnosing why visibility has changed — our SEO services for Tunbridge Wells businesses page covers that.

Tunbridge Wells Digital Agency FAQs

How much traffic do we need before this is worth doing?

Less than you need for formal split testing, which requires substantial volume to reach a reliable result. Reviewing journeys, identifying drop-offs and making targeted improvements is worthwhile on modest traffic — you simply rely more on judgement and qualitative feedback, and give changes longer before assessing them.

Will improving the website increase our conversions?

It may, and we would not promise it. Some changes make a clear difference, some make none, and occasionally something performs worse than what it replaced. What a structured approach gives you is knowing which of those happened, rather than assuming the change helped because it was made.

What if the results are inconclusive?

That is a common and legitimate outcome, particularly on lower-traffic sites. The honest response is to say the evidence does not support a conclusion rather than interpreting noise favourably. Sometimes the sensible next step is to leave the change in place if it is defensible on other grounds, and look elsewhere.

Is this the same as redesigning the website?

No, and it is often the alternative. A redesign changes everything at once, which makes it impossible to tell what helped. Iterative improvement changes specific things for stated reasons and checks the effect. It is slower and it produces knowledge, which a redesign generally does not.