Case study
From position 19,9 to 8,4 in actual traffic gained
International smartphone case retailer · Portugal · A 163% increase in clicks and why only part of it counts
I’m going to make a confession that would give a sales rep a heart attack: of that +163% increase in clicks on the headline, only part of it is down to us. And we put that in writing in the closing report before anyone even asked. Let me explain 🙂
In the May–July quarter, clicks in Search Console rose from 29.211 to 76.793 year-on-year. Breaking this down, branded traffic grew by 399% and non-branded traffic by 31%. The first figure is explained by a recovery in CTR for searches of the brand name itself — from 13,5% to 55,5% — which isn’t attributable to our work.
The latter, however, is. And it is accompanied by a figure that leaves no room for debate: the average desktop position for non-branded searches fell from 19,9 to 8,4. That’s from the second page to the middle of the first.
Let’s begin:
+163%
year-on-year organic clicks
19,9 → 8,4
non-branded desktop position
34,5% → 47,6%
share of voice (all-time high)
46% → 81%
demand captured in the Top 3
1. The starting point
Portugal entered January 2026 with a 34,5 per cent share of voice on desktop and 57 out of 97 keywords in the Top 3. It wasn’t a disaster; it was a plateau. The brand was capturing 46 per cent of the search volume within its niche across the top three positions.
Beneath that plateau lay some very specific work to be done:
- 16 out of 21 meta descriptions were empty in the accessories section; the H1 tags literally replicated the title; the alt text for the images was the same generic text duplicated across all of them.
- Empty product descriptions or descriptions written directly in Spanish on the Portuguese site.
- Four products with a mismatch between the slug and the URL in PrestaShop.
- No structured data across the entire site.
- And a conversion rate of 0,63 per cent with a clear bottleneck: out of every hundred product pages viewed, fewer than ten resulted in an ‘Add to basket’.
Oh, and a licence keyword — the B licence — languishing in 52nd position.
2. What we did
Four areas of focus, six months:
- On-page SEO audit of 36 pages — 21 for charms and accessories, 15 for Licence A — with complete metadata, image optimisation and identification of the four slug discrepancies marked as top priority.
- A comprehensive UX audit cross-referencing GA4 and Clarity over twenty-eight days, with seven prioritised tickets for the technical team, including the necessary Apache and Nginx code. Three were critical: a 302 redirect on /index.php affecting 3.242 sessions with a 79,8% bounce rate, JavaScript errors on the product page, and the customiser with broken interactivity.
- Ongoing management of Google Ads with monthly reports, tROAS optimisation per campaign, and analysis of 500 Search Console queries to identify gaps.
- Reconciliation between PrestaShop orders and GA4 transactions: 35 orders totalling €589,31 were found that the analytics system had not recorded.
3. The results
| Metric | Before | After | Source |
|---|---|---|---|
| Organic clicks (quarter) | 29.211 | 76.793 | GSC · measured |
| — of which, non-branded | - | +31% | GSC · measured |
| Non-branded position · desktop | 19,9 | 8,4 | GSC · measured |
| Share of voice · desktop | 34,5% | 47,6% | Semrush · est. |
| Share of voice · mobile | 32,0% | 45,3% | Semrush · est. |
| Keywords in the Top 3 (out of 97) | 57 | 64 | Semrush PT |
| Traffic captured in the Top 3 | 46% (5.820 searches/month) | 81% (10.300 visits/month) | Semrush · est. |
| Licence B | Rank 52 | Rank 1 | Semrush PT |
| Licence A / a classic animation licence | 3,3 | 1,7 | Semrush PT |
| Organic sessions (90 days) | - | 130.611 (53,9% of the total) | GA4 · measured |
Year-on-year comparison for the May–July quarter using Search Console metrics; average of the snapshots from 14, 21 and 28 January 2026 used as the baseline for ranking data. The 47,6% share of voice is an all-time high for the Portuguese market.
On a month-on-month basis, year-on-year click growth was +51% in May, +105% in June and +152% in July. An accelerating trend usually indicates that the work is starting to pay off, rather than that there was a one-off spike.
On Google Ads, June closed with an ROAS of 167%, compared to 109% in May. What matters is not the percentage itself, but the fact that this was achieved with 44% less investment, whilst conversions fell by just 7,3% and the cost per acquisition dropped by 40%. That is efficiency, not volume. It’s worth noting that the cut in investment wasn’t an optimisation decision on our part, but a business decision by the client that had been in place since 20 May; what is attributable to our work is that efficiency increased rather than decreased.
4. What other reports won’t tell you
Two things, and both are uncomfortable truths. It’s clear that no aggregated data tells the whole story. That’s why we’ve broken it down before anyone even asked.
You’ve already read the first point: a large part of the aggregate growth is down to the brand, and we didn’t build that brand ourselves. When a report presents you with a +163% figure without breaking it down, ask for details. In this case, the honest breakdown is +31% for new business and +399% for recovered business.
Secondly: at the end of the period, metadata for 72 URLs was still pending upload. In other words, all of the above was achieved with a partial implementation, and there remains scope for growth—which has already been identified and prioritised in the handover document.
To sum up: what does an SEO project actually measure?
Essentially, the search ranking where your brand isn’t yet the top result. Everything else — visits, impressions, even aggregate clicks — can fluctuate for reasons entirely unrelated to the work carried out.
I’d recommend three things for tomorrow. First, I’d split the dashboard into two sections: branded and non-branded, each with its own average position. Secondly, I’d focus on captured demand — moving from 46% to 81% of market volume in the Top 3 says more than any sessions chart. And thirdly, I’d close the loop with analytics: if your shop records 35 fewer orders than it has made, all subsequent profitability calculations will be wrong.
Remember, at the end of the day, it’s all about knowing how to distinguish what you’ve gained from what you already had 😉
