Optimisation is sold as one service and covers two unrelated disciplines. Performance work makes the site fast and technically sound. Conversion work makes more visitors do what you want. A supplier strong at one is frequently weak at the other, and buying the wrong one is a quarter spent improving something that was not the constraint.
Performance work is engineering
Image handling, script loading, server response, caching and layout stability. It is measurable, largely objective, and it matters for both users and search. It is also finite: a site can be made fast and then stays fast until someone adds something heavy, which makes it a project rather than a permanent retainer.
Conversion work is research and testing
Understanding why visitors leave, forming hypotheses, changing the page and measuring the effect. It needs enough traffic for results to mean anything, which is the constraint most sites hit: below a certain volume, tests take months to reach confidence and the honest answer is to make judgement-led improvements instead.
Which one you need
If the site is slow, unstable on a phone or technically broken, performance first, because conversion testing on a broken page measures the breakage. If it is fast and visitors still leave, the problem is the offer, the clarity or the friction, which is conversion work. A supplier who tells you which without being asked is worth listening to.
Judging conversion work honestly
Ask how a test is called: what sample, what duration, what threshold. A supplier declaring winners from small samples after a few days is producing noise, and the reported gains will not survive. Ask what proportion of their tests fail; an honest answer is most of them, and anyone claiming otherwise is either testing trivia or not measuring.
Questions people ask about website optimisation services
How much traffic is needed to test?
Enough that a meaningful difference can be detected in a reasonable period, which for most small sites means testing large changes rather than small ones, or not testing at all.
Does site speed affect rankings?
Page experience is part of how pages are assessed and it matters far more for users, particularly on phones. Treat it as a user problem that also helps visibility.
Is conversion optimisation a retainer?
It suits one where there is enough traffic to test continuously. Below that, a periodic review with judgement-led changes is more honest than a testing programme that never reaches confidence.