Fewer people were shown your listings, and almost exactly as many clicked. That combination is the whole story of the month, and it is worth taking slowly because the two halves pull in opposite directions.
Impressions are the number of times one of your pages appeared in a Google result. They fell 11.3%, from 80,567 in August to 71,464. Clicks are the number of times somebody actually came through. They went from 539 to 531, a fall of 1.5%, which on these volumes is flat. Divide one by the other and you get the click-through rate: 0.74% in September against 0.67% in August. So the exposure that disappeared was the exposure that was not earning visits anyway.
The most likely reason sits on the results page itself. On the day of our live check, 32 of your 37 tracked searches had a Google AI answer sitting above the ordinary results. When Google answers a question on its own page, it has less need to show a list of suppliers underneath, so your pages get put in front of fewer of those searches. The people who scroll past the answer and click are the ones who want a supplier rather than a definition, which is the shape of what we measured: less exposure, the same number of visits, a better rate between the two. We cannot prove that link from Search Console data alone, so treat it as the best available reading rather than a finding.
Where the clicks came from shifted. Your homepage, which takes most of the brand-name searches, fell from 156 clicks to 103, and its share of all organic clicks went from 28.9% to 19.4%. Over the same month the other nine pages in your top ten added 35 clicks between them: the 2026 surcharge ban explainer went from 25 to 40, the "buy an eftpos machine" page from 5 to 16, the ICC declined explainer from 32 to 38, and the paywave explainer from 22 to 27. Those are non-branded searches, people who did not know your name when they started. Brand-side traffic eased and category discovery grew, which is the better half of that trade.
Against last September the picture is mixed: impressions are up 3.8% (71,464 against 68,876), so across the year Google has shown you more often, while clicks are down 15.7% (531 against 630). That twelve-month gap between rising exposure and falling visits is the clearest number on this page, and it is the same shift the AI answers describe, measured over a long enough window to be reliable.
Checked live on Google New Zealand on 12 October 2026 across your 37 tracked terms. This is a position check run on one day, so treat it as a snapshot of the current state rather than a score.
| Where they sit | Count | Of 37 tracked | Notes |
|---|---|---|---|
| Number 1 | 4 | "eftpos machine nz", "self serve kiosk", "pos for hospitality" and "pos for hair and beauty". The last two are industry pages built in July. | |
| Top 3 | 11 | Adds "pos software", "point of sale systems", "pos systems nz", "hospitality point of sale software", "eftpos terminals", "eftpos terminals nz" and "hire eftpos machine" | |
| Page one | 19 | Page one means a position in the top 10 | |
| Top 50 | 27 | One more sits at 72 ("portable eftpos machine nz") | |
| Ranking anywhere in the results | 28 | The remaining 9 did not surface on the day of the check | |
| Google AI answer present | 32 | Whether an AI answer appeared at all. It does not tell us who it quoted. |
What is holding: two of July's four industry pages are still at number 1, "pos for hospitality" and "pos for hair and beauty". Eftpos Now is the top result on Google for all four of the number 1 terms. Across the category terms you sit second to a different competitor on each of "pos software", "point of sale systems", "pos systems nz" and "hospitality point of sale software", which means four separate near-misses rather than one rival consistently ahead of you.
What slipped: "pos for events" was at number 1 in the August check and came back at 20 this time, and the broader "pos system for events" sits at 13. The retail side is split: "retail pos systems for small business" is at 4, while "pos systems for retail" and "pos for retail store" did not surface. "point of sale nz" is at 18, "merchant services nz" at 33 and "pos nz" at 40.
Still unclaimed: nine terms did not appear on the day, including "restaurant tablet ordering system", "pos system for small business", "pos software for restaurant", "mobile eftpos", "eftpos mobile", "merchant fees nz" and "eftpos specialists". Those are the open territory on this list.
August's tab reported 31 of 37 on page one from a rankings check run a different way, so the two counts are better read as separate snapshots than as a month-on-month trend. One of the 37 tracked terms, "Salespoint", is a competitor's brand name rather than one of yours; it sits at 20, so it adds nothing to the page-one count, but it is part of the 37 that every count here is measured against. The Search Console average position figure is a diagnostic we use internally and is not a measure of where you rank, so it is deliberately not quoted here. live Google NZ check
| Measure | September 2026 | August 2026 |
|---|---|---|
| Contact form submissions | 69 | 74 |
| Phone number clicks | 79 | 77 |
| Email clicks | 41 | 34 |
| Total tracked contact actions | 189 | 185 |
| Quote request button clicks | 619 | 488 |
| "Contact us" and "Get in touch" button clicks | 171 | 141 |
| All tracked actions on the site | 986 | 822 |
| Website sessions (all channels) | 2,742 | 2,290 |
| Organic search sessions | 897 | 864 |
| Visits from AI assistants | 14 | 12 |
The top line held: 189 tracked contact actions against 185, with phone and email both up and the contact form a little softer. Below that, intent moved up sharply. Quote request button clicks rose 26.8%, from 488 to 619, and the two general contact buttons rose from 141 to 171. Every tracked action on the site together came to 986, up 20.0% on August's 822. More people on the site were reaching for the next step than in August.
Total sessions rose 19.7% to 2,742, and most of that lift came from paid activity rather than search: paid channels went from 544 sessions to 999 across the two months, mostly new paid social. Organic search grew more modestly, +3.8% to 897, which means organic's share of all visits moved from 37.7% to 32.7% even though the number of organic visits went up. The share fell because paid grew, so the absolute number is the one to read here.
Visits from AI assistants reached 14, from ChatGPT (12) and Claude (2), against 12 in August. Small, and the direction has been upward all year.
Visits from AI assistants count the sessions whose referrer was an AI assistant, 14 in September and 12 in August. GA4 also carries its own AI Assistant channel grouping, which reads 16 and 13 for those months, and the narrower referral-source count is the one used here so the two cannot be confused. Contact actions follow the same definition as every previous report, the contact form plus phone and email clicks, so the trend line stays consistent. Quote and contact button clicks are counted separately because a button click is a step toward an enquiry rather than an enquiry itself. Three further form events (two consumables, one apply now) sit outside the series. Earlier tabs quoted visitors as users by first channel from your SEO sheet; the sheet has no September column, so September is given here as GA4 sessions, with August's sessions alongside for a like-for-like comparison. Session counts are higher than the visitor counts in earlier tabs because one person can start more than one session. Any enquiry that arrives by direct phone call or email lands on top of all of these. GA4
September closes a four-year engagement, so this last section is the handover rather than a plan. Here is the position as we measured it on 12 October, with no promises attached to any of it.
| What you have | The state of it |
|---|---|
| A page-one footprint across the category | 19 of 37 tracked terms on page one, 11 in the top 3, and 4 at number 1 where Eftpos Now is the top result on Google. That footprint is spread across the homepage, the POS systems page, the industry pages and the eftpos explainer articles, so it does not rest on any single page. |
| An authority base that keeps working | 820 referring domains, 1,063 live backlinks and a Domain Rating of 12. This is the asset that let July's industry pages rank within weeks rather than months, and it does not need monthly topping up to keep doing that. |
| A library of explainer content that is growing | The surcharge ban, ICC declined, paywave and personal terminal articles all grew their clicks in September while the homepage fell. These are non-branded entry points, and right now they are the part of the site adding new audience. SEMrush has the site visible for 846 searches in New Zealand. |
| A technical base built for AI answers | Schema is deployed across the site, your hand-written llms.txt file is genuinely one of the better ones we see, and AI crawlers are welcomed rather than blocked. Whoever works on the site next should leave all three alone unless they are improving them. |
| Open territory, named | Nine tracked terms did not surface on the day of the check, and "point of sale nz" (18), "merchant services nz" (33) and "pos nz" (40) are all within reach from where they sit. The retail pages are the least settled of July's four builds. Any of these is a content job rather than a technical one. |
| The thing that changes the measurement | 32 of your 37 tracked searches carried a Google AI answer on the day we looked. AI answer presence moves about day to day, so that count is a reading rather than a fixed score. It is the reason impressions can fall while clicks hold, and it means click volume alone will keep understating how visible you are. Anyone reading these numbers next year should expect that gap to widen. |
Three notes on reading the data after today. Search Console revises recent months upward for a few weeks after they close, so a month looks a little better the later you pull it. Your SEO sheet is a manual record and its last filled column is July 2026, so the sheet and this page will not line up for August or September. And Search Console's average position is a diagnostic, useful for spotting movement, misleading as a statement of where you rank, which is why every ranking figure on this page comes from a live check on Google instead.
Thank you for the four years. The access, the reporting history and the sheet are all yours, this page stays where it is, and if a future team wants the background on how any of it was built, it is written down here.