Same traffic. More of it turning into enquiries.
Conversion rate optimisation increases the share of visitors who take the action you want. Haryes runs a structured cycle (analytics review, hypothesis, test, measure) on the pages that already carry traffic, rather than redesigning on opinion.
Most sites are redesigned on opinion and then called an improvement.
Someone senior does not like the hero image, so it changes. Conversion rate moves, or does not, and nobody can say why because six things changed in the same week.
Optimisation is the opposite of that.
- No baseline, so there is nothing to compare a change against.
- Several changes shipped together, which makes every result uninterpretable.
- Decisions made from the opinion of the most senior person in the room.
- Tests stopped the moment the numbers look good, which is when they are least reliable.
How the cycle runs
Monthly, on the pages that already carry traffic. Optimising a page nobody visits is arithmetic you cannot win.
Find where people leave
Analytics, funnel drop-off and session recordings on the journeys that actually produce enquiries.
Evidence rather than instinct.
Form a hypothesis
A specific statement: this element causes this drop-off for this reason, and changing it will do this.
Something that can be wrong.
Change one thing
A single variable, built and shipped cleanly, with the baseline recorded first.
An interpretable result.
Measure honestly
Run it long enough on your traffic volumes to mean something, which on most Kenyan sites is weeks rather than days.
A result worth acting on.
Keep or revert
Losing tests are kept as findings. Knowing what does not work is most of the value over a year.
A record that compounds.
Where the gains usually actually are
Rarely the button colour. Almost always one of these.
Clarity of the offer
Visitors who cannot tell what you do in five seconds leave, and no amount of layout work fixes that.
Say it plainly.
Friction in the form
Every unnecessary field costs completions. Most forms ask for things nobody reads afterwards.
Cut what you do not use.
Missing answers
Price, timeline and what happens next. Pages that withhold these lose people who were ready.
Answer the objection.
Speed
On mobile data, a slow page is a conversion problem before it is a performance one.
Measure in the field.
What you get each month
Runs as a monthly engagement on sites with enough traffic to measure.
An analysis
Where people leave, and what it costs.
- Funnel drop-off by step
- Session recordings on key journeys
- Device and connection breakdown
A tested change
One variable at a time.
- Hypothesis stated before the build
- Baseline recorded
- Run to a meaningful sample
Implementation
Built properly, not bolted on.
- Changes in the real templates
- No third-party script slowing the page
- Reverted cleanly if it loses
A findings log
The asset that compounds.
- What was tested and why
- What won, what lost
- What we would try next
Questions clients ask before they commit
How much traffic do we need?
Enough for a result to mean something. On a site with a handful of enquiries a month, structured testing is not the right spend, and we will say so and point you at traffic instead.
Do you run A/B tests?
Where volume supports it. Below that we use before-and-after measurement over defined periods and are honest about its limits, rather than pretending a week of data proves anything.
Is this the same as a redesign?
No. A redesign changes everything at once and teaches you nothing. Optimisation changes one thing and tells you whether it worked.
What if a test loses?
We revert and record it. Roughly half of well-formed hypotheses lose, and a year of knowing what does not work on your audience is worth real money.
Traffic problem, or conversion problem?
People arrive and do not act.
That is a conversion problem and it is the cheaper one to fix.
Improve my conversion rateToo few visitors to measure anything.
Fix the traffic first. We will tell you honestly which you have.
See SEO servicesDoubling conversion rate is usually cheaper than doubling traffic, if you have the traffic to work with.
