060 279 5587 info@sitect.co.za 139 Davies Street, Doornfontein, Johannesburg, 2001 Gauteng, SA
E-Commerce · Conversion Rate Optimisation

More sales from the same traffic.

Every R1 you pay for ads has the same job: bring a shopper. Whether they buy is up to your site — and most SA stores leave 30–60% of revenue on the table at the conversion step. We run a disciplined CRO programme: audit, hypothesis, A/B test, learn — every month, on your store.

From R7k/mo · 3-month minimum 2–3 experiments shipped per month Lift report every cycle
Experiment #14 · PDP sticky-buy Winner ✓ Day 12 of 14 · 99% conf.
Control · A
Conversion rate
2.14%
Visitors11 240
Orders241
Avg. order valueR612
Revenue / visitorR13.10
Variant · B
Conversion rate
3.06%
Visitors11 188
Orders342
Avg. order valueR624
Revenue / visitorR19.09
Lift+42.9%
Confidence99.2%
Annualised value+R487k

Funnel — last 30 days Step 4 fixed

Visitors
100%
52.4k
Product
74%
38.8k
Add to cart
38%
19.9k
Checkout
18%
9.4k
Confirmed
9%
4.7k
How CRO usually goes wrong

Why most "CRO programmes" don't move the number

CRO became a buzzword and now every agency offers it. Most do it badly. Here are the four patterns that turn it into expensive theatre.

Anti-pattern 01

"Best-practice" copy-paste with no hypothesis.

"Move the CTA above the fold." "Make the button orange." "Add urgency." These are not experiments — they're someone else's results applied to your store with no idea why they worked for them. Sometimes they win. Mostly they don't move the metric, or they win for the wrong reason.

What it costs: 6 months of "tests", no learnings, no compounding insight.
Anti-pattern 02

Tests called before they reach statistical significance.

"It's been running 4 days and the variant is up 18% — let's call it and roll it out." A week later the lift disappears, or worse, reverses. The agency moves on. Half the "wins" in CRO portfolios are noise that was rolled out before the test ended.

What it costs: rolling out losers, eroded trust in the programme, no compounding gains.
Anti-pattern 03

Optimising the wrong part of the funnel.

If 92% of shoppers drop at "Add to cart → Checkout", testing hero-headline copy is a waste of effort. CRO without funnel diagnostics tests where it's easy, not where it matters. The biggest gain is almost always at the worst-converting step.

What it costs: small lifts on big-traffic steps instead of huge lifts on the bottleneck.
Anti-pattern 04

No qualitative input — only Google Analytics.

GA tells you what happened. It never tells you why. Without session replays, on-site surveys, post-purchase questions, and 1:1 user tests, you're guessing at hypotheses. Quant + qual together produce hypotheses that win 40-60% of the time. Quant alone gets 15%.

What it costs: low win-rate, low ROI, low engagement from your team.
Our method

The 5-step loop every CRO win comes from

This loop runs continuously — not once. Each cycle compounds on the last because we keep the learnings and the data.

1

Measure

Funnel diagnostics, quant + qual, find the worst-converting step.

2

Hypothesise

"Because X is happening, if we Y we expect Z lift." Specific. Testable.

3

Design

Variant designed, instrumented, sample size calculated up-front.

4

Test

A/B test runs to statistical significance — no early calls, no early kills.

5

Learn

Document the result + the reason. Roll out winners. Library the loss.

The audit scorecard

Every engagement starts with a store conversion scorecard

Before we run a single test we score your store across 8 conversion-relevant areas. The lowest-scoring areas become the first month's experiment queue.

What we look at — on day one

Most CRO engagements waste month one "ramping up". Ours doesn't. We arrive with a 60-point audit framework, run it inside 5 working days, and present you with a ranked list of opportunities — each tagged with effort, confidence, and projected uplift.

You see what's broken, why it matters, and what we'd test first. You can stop the engagement after the audit if you want — most clients don't.

Example scorecard from a recent SA fashion retailer — anonymised.

Area
Score
Status
Product pageTrust, imagery, copy, social proof
42/100
Checkout flowForm, friction, error states
38/100
Add-to-cart UXSticky bar, mini cart, upsells
61/100
Mobile performanceLCP, CLS, mobile-only friction
54/100
Trust signalsReviews, badges, returns policy
66/100
Search & navigationFindability, filters, empty states
78/100
Lifecycle & retentionCart abandonment, win-back, post-purchase
35/100
Tracking & dataPixel health, attribution, server-side
82/100
Our instrument stack

How we actually measure your store

You can't optimise what you can't see. Every engagement begins by instrumenting your store with the four lenses below — three weeks in, you have more visibility than most agencies achieve in a year.

Quant analytics

GA4 + server-side tracking (Meta CAPI, GA4 Measurement Protocol), funnel events, custom dimensions, deduped attribution.

GA4PlausibleBigQuery

Session replay

Real shopper sessions on video — watch the moments people get stuck, rage-click, or bail. The single highest-ROI tool in CRO.

Microsoft ClarityHotjarFullStory

Voice of customer

On-site polls, exit-intent surveys, post-purchase questions ("what almost stopped you?"), 1:1 user testing.

Hotjar pollsUserTestingTypeform

A/B testing engine

Server-side or client-side experiments depending on your stack and traffic. Sample sizes calculated up-front. No early calls.

GrowthBookConvertVWOOptimizely

Heatmaps

Click maps, scroll maps, attention maps. Identify dead zones, ignored CTAs, and unexpected interactions.

ClarityHotjar

Performance monitoring

Real-user Core Web Vitals tracking by page-type, device, and country. Slow pages are conversion killers.

SpeedCurvePageSpeed InsightsCrUX

Tag manager

Centralised, governed tagging — server-side GTM where possible, consent-aware, with versioning + rollback.

GTM SSStapeCloudflare Worker

Shopping & lifecycle

Abandoned-cart, browse-abandon, win-back, post-purchase email + SMS + WhatsApp flows wired to your CRO experiments.

KlaviyoOmnisendCustomer.io
Our SA-tested playbook

Experiments that consistently win for SA stores

Every store is different — but certain patterns lift conversion across most SA e-commerce categories. These are the experiments we'd queue up first, with the typical lift band we observe.

Sticky add-to-cart on PDP

+8–15%

Mobile-first persistent bar on product pages — keeps "Add to cart" reachable as shoppers scroll the gallery and read reviews.

Express-checkout buttons

+12–22%

Shop Pay / Apple Pay / Google Pay above the cart line — bypasses 80% of the form for returning shoppers. Single biggest mobile-only win.

Reviews above the fold

+6–14%

Star rating + count near the price, not buried below variant pickers. Trust must precede the decision, not follow it.

Delivery date estimate

+5–12%

"Order in the next 2h 14min and get it Wed 12 Mar". Removes the #1 SA shopper anxiety — "when will it actually arrive?"

Returns policy on PDP

+4–9%

"Free returns within 14 days" pill near the buy button. Removes the second-biggest SA shopper anxiety — "can I send it back?"

Pay later / instalments

+6–18%

Payflex, Mobicred, Float — surfaced on PDP and at checkout. Particularly strong on AOV over R1 500.

WhatsApp support on PDP

+3–8%

"Question about this product? WhatsApp us." Visible button → real human. SA shoppers prefer WhatsApp over chat-bot widgets.

Address-autocomplete

+4–11%

Google Places autocomplete at the address step — cuts the most frustrating typing on mobile and avoids "address not found" errors.

Exit-intent abandon flow

+7–14%

For cart-abandoners with email captured — a 3-message sequence over 24h, mostly recovers AOV-loss-equivalents within the first 6 hours.

What you actually get

Every month, every retainer

CRO retainers can be vague — ours are not. Here's exactly what shows up in your inbox every month.

Funnel diagnostics report

Drop-off by step, by device, by traffic source, by product category. Highlights the top 3 leak points and the projected revenue at stake.

Experiment queue

Ranked list of hypotheses for the next 6 weeks — each with effort, confidence, projected uplift, and the page/segment it targets.

Live experiments (2–3 / mo)

Variants designed, copywritten, built, instrumented, and shipped behind a feature flag. We handle the dev work; you don't queue engineering.

Lift report (per cycle)

Every experiment ends with a written report — outcome, statistical confidence, lift, learnings, recommendation to roll out / iterate / shelve.

Insights library

Every win, every loss, the reason it moved the metric — stored in a shared Notion workspace so the learnings compound over years.

Weekly stand-up

30-min weekly call with your team — experiment status, blockers, next-cycle prioritisation. Slack channel in between for fast questions.

A typical month

What week-by-week looks like

Each calendar month a CRO retainer ships 2–3 experiments and produces a lift report. Here's how a representative month is structured.

Week 1

Diagnose

  • Pull funnel + replay data
  • Review prior-cycle results
  • Hypothesis workshop
  • Prioritised queue v.next
Week 2

Design & build

  • Variant designs (Figma)
  • Sample-size calc
  • Implementation + flag
  • QA on staging
Week 3

Ship & monitor

  • 5% ramp-up day 1
  • 50/50 split from day 2
  • Daily health checks
  • Guardrail metrics
Week 4

Report & learn

  • Statistical significance
  • Lift report written
  • Winner rolled out 100%
  • Insights library updated
What "shipped right" earns

The kind of compounding we engineer for

Indicative figures across 6 active SA CRO retainers, measured at the 6-month mark. CRO compounds — month-on-month lifts stack on each other when the loss-rate is low.

+38%
Conversion rate
Average · 6-month programme
52%
Experiment win-rate
Industry avg: 12–18%
+R420k
Median annualised lift
By month 6 on a mid-sized store
6.4×
ROI on retainer
12-month rolling, across cohort
Pricing — ZAR, ex VAT

Three ways to work with us

We offer a one-off audit for teams that want to know what's broken, and two retainer tiers for teams that want it fixed and re-tuned every month. 3-month minimum on retainers; month-to-month afterwards.

One-off Audit

Full conversion scorecard + 90-day roadmap
R7 000 · once-off
5 working days · keep the roadmap
  • 60-point conversion scorecard
  • Funnel diagnostics report
  • 12 ranked experiment hypotheses
  • 90-day prioritised roadmap
  • 60-min handover & Q&A call
  • Audit fee credits if you sign a retainer
Order an audit

Growth Team

Embedded squad · 5–6 experiments / month
R22 000 /mo · ex VAT
6-month minimum · embedded pod
  • Strategist + designer + engineer + analyst
  • 5–6 experiments per month
  • Lifecycle + email + WhatsApp flows
  • Dedicated BigQuery + dashboards
  • Quarterly C-level review
  • Priority Slack support
Talk to us about Growth Team
FAQ

The questions we get asked most

Honest answers about traffic minimums, win-rates, what we won't do, and how to know if CRO is right for your stage.

How much traffic do I need before CRO makes sense?
Rough rule: 50 000 monthly sessions across your funnel-relevant pages, or ~500 orders/month. Below that, A/B tests take too long to reach significance and you're better off investing in qualitative research (user testing, surveys) and obvious wins (express checkout, address autocomplete, sticky add-to-cart) rather than statistical testing. We'll tell you honestly when you're below the threshold.
How long until we see results?
First measurable lift typically lands in week 6–8 — long enough to ship the first experiment, run it to significance, and roll out the winner. Compounding gains show up by month 3. The first month is mostly diagnostics, instrumentation, and queue prioritisation — necessary but not yet revenue-moving.
What if an experiment loses?
We expect about half of all experiments to lose or be inconclusive. That's normal — and useful. Each loss kills a hypothesis you would have otherwise believed. We document the loss + the likely reason, the variant doesn't ship, and the insight goes into your library so we don't re-run the same hypothesis next quarter.
Do we need to be on a particular platform?
We work on Shopify, WooCommerce, custom Laravel/Node/PHP stores, BigCommerce, and headless setups. We choose the testing engine that fits your stack — GrowthBook server-side for headless, Convert / VWO / Optimizely for traditional. We don't force a platform change to fit our toolkit.
How is this different from CRO software like VWO/Optimizely on its own?
Tools don't form hypotheses, design variants, or write lift reports. A CRO programme is people running a disciplined loop on top of the tools. Most stores that bought CRO software and tried to run it themselves abandoned the programme within 6 months — not because the tool was bad, but because the work was discipline-heavy.
What if our team wants to run our own experiments alongside?
Encouraged. We share the insights library, the hypothesis-prioritisation framework, the experiment-design template, and the lift-report template with your team. We co-run experiments and stop you running concurrent tests that interfere with each other. Many of our best engagements end with the client team running their own CRO programme — that's a healthy outcome.
What's the difference between "CRO" and "growth marketing"?
CRO focuses on the post-arrival journey — turning visitors into buyers. Growth marketing also covers acquisition, lifecycle, retention, paid spend. Our Growth Team tier includes the wider scope; the Growth CRO retainer focuses on on-site conversion. Most teams hire CRO first and add lifecycle later — it's the higher-leverage starting point.
Are you POPIA-compliant around the data you collect?
Yes — every instrumentation we ship respects consent state, anonymises PII on session replays (we never record card forms, addresses, or password fields), uses data-processing agreements with every analytics vendor, and adds a data-flow document to your information officer's records.

Send us your store URL — we'll send back a free 20-minute funnel teardown.

No commitment. We'll record a screen-share walkthrough of your funnel, point out the 3 biggest leak points, and tell you what we'd test first. About 40% of teardowns turn into retainers. The other 60% are useful anyway.