Conversion strategy
Find the decision break after the click
Traffic is useful only when the page helps the right visitor decide. Diagnose weak conversion by checking intent, clarity, proof, friction, speed, lead quality, and the next step.
Define the action before judging the rate
A conversion is a completed action you have chosen to measure. It might be a purchase, a booking or an enquiry. These actions have different values, and a button click is not the same as a completed order.
Choose one landing page and one intended outcome. Record the period, audience and counting rule before looking for a design fix.
- Research visits: help the reader compare or understand. An immediate sale may be the wrong expectation.
- Ready buyers: make fit, availability, price information and the next step easy to find.
- Returning customers: provide a clear route back to their task.
- Support visits: answer the policy, replacement or refund question. Do not judge successful support by new sales.
The same rate can hide different problems
These hypothetical figures describe two service pages over the same period. Each receives 1,000 recorded sessions and 20 accepted enquiries, with no session counted more than once. Both have a 2% session enquiry rate: 20 ÷ 1,000 × 100.
- Page A: 100 sessions reach the enquiry form; 20 submit successfully. Start by checking what happens during form completion, while also examining why the other visits did not reach it.
- Page B: 25 sessions reach the form; 20 submit successfully. Start by checking the match between the arriving audience, the offer and the route to the form.
The form-completion rates are 20% and 80%. Neither number identifies the cause. Page A may have an error, difficult requirements or visitors who change their minds. Page B may serve mostly researchers, hide its action or fail to explain the offer.
Confirm that tracking uses the same definitions and captures successful receipt, not just a click. Duplicate events, spam and missed submissions can make the apparent problem misleading. These figures are an investigation exercise, not conversion benchmarks.
Find evidence for the point where people stop
Before the useful answer
Compare the query or campaign promise with the landing page. Is a visitor seeking a replacement part sent to a general product range? Is a local enquiry arriving outside the service area? Separate those visits before judging the offer.
While judging the offer
Ask whether a new reader can identify who the service is for, what is included and what they should do next. Look for unanswered questions in sales calls, email, chat and support tickets.
Before taking action
Check the information needed to continue: real examples, relevant reviews, specifications, availability, price conditions or the stages of the work. Use guarantees only where the business actually offers them, with their terms visible.
During the action
Try the form, booking or checkout on the devices visitors use. Look for errors, unexpected charges, unnecessary fields, moving controls and missing confirmation. A route can fail even when the offer is clear.
After the enquiry
Check whether the request reaches the business, gets an appropriate reply and fits the service. A page change cannot repair every delay or qualification problem after submission.
Match proof to the question being asked
Broad claims such as “trusted quality” leave the visitor to infer what they mean. Put the relevant evidence beside the decision it supports.
For a hypothetical upholstery service, the question “Can you repair this kind of damage?” needs an example of that work and its limits. A long company history alone does not answer it.
- Will it fit? Show dimensions, compatibility or service eligibility.
- What will I receive? Explain scope, deliverables and exclusions in plain language.
- Can you do this? Show relevant completed work and explain what was done.
- What happens after I ask? State the actual assessment, booking or response process.
- What will it cost? Give the available price information and explain what must be established before a quote.
Test the whole route on a phone
Give someone unfamiliar with the page a realistic task: “Find out whether this workshop can assess your damaged sofa and request a reply.” Observe where they hesitate without explaining the page for them.
- Start at the page that actually receives visits. Check loading, readable text, usable images and controls that remain in place.
- Follow the intended action. Note unexpected steps and information requested before its purpose is clear.
- Use a clearly identified test submission with the business’s agreement. Try a missing required field and check that the error explains how to correct it.
- Complete the action and verify confirmation, receipt and the promised next step. Keep test records out of performance totals.
- Record the exact obstacle, device and step so the person responsible can reproduce it.
Forms should give clear feedback on both errors and successful completion. The W3C form notification guidance explains how to make that feedback understandable.
Fix a reproducible broken route directly. Use observation to investigate uncertainty about wording or sequence; a few observed users do not establish a population-wide conversion rate.
Measure useful progress, not just activity
Reach the next step
Compare the relevant landing-page visits with visits that reach the intended action. Keep the period and counting method consistent. A scroll or click can locate a point to inspect; it does not explain what the visitor thought.
Complete the action
Count confirmed submissions, bookings or purchases. If you use a funnel report, define its steps and whether visitors must enter at the first step. Those settings affect the result; see Google Analytics funnel exploration.
Fit the offer
Review whether enquiries match the service, area and customer need. Keep spam, duplicates and tests separate. More submissions can create work without creating more suitable demand.
Become a customer
Follow qualified enquiries through the sales cycle. Keep unresolved cases visible instead of counting them as failures too early. Record reasons for lost work where the customer supplies them.
Choose the first fix from the evidence
Use a short record: affected page, audience, observed obstacle, proposed change, owner and outcome to check. A useful entry says “The photo upload fails on this phone; repair it and verify successful receipt”, rather than “Improve conversion”.
If visitors are mostly seeking something you do not supply, repair targeting and landing-page fit. If relevant visitors cannot complete the action, fix the route. If the route works but buyers cannot judge the offer, strengthen the missing explanation or proof.
Record the change date and compare suitable periods. Allow for seasonality, traffic mix and the sales cycle. A before-and-after difference suggests something to investigate; it does not isolate the effect of the edit. Avoid declaring a winner from a handful of visits.
For changes to titles, openings and content structure, follow the on-page improvement guide. Keep the test tied to the buyer’s task.