How Nordic Shopify agency Grafikr turns four post-purchase questions into a RICE-scored CRO backlog
Survey Type:
Post-purchase survey on the Shopify Order Status page
Survey Questions:
Who are you buying for? Why did you buy today? How did you first hear about
us? Was anything close to stopping you from buying today?
Grafikr is a Nordic Shopify Platinum Partner, founded in 2016 to give fashion and lifestyle DTC brands full ownership of their customer journey on Shopify Plus. They aren't a general agency — ecommerce is all they do, and they've helped build some of the most recognizable DTC shopping experiences in the region.
We had the chance to chat with Erik Holflod Jeppesen, who co-founded Grafikr and works with its clients as an E-commerce Advisor, about how the team uses Zigpoll across client engagements to:
- Replace cookie-based attribution that was failing on the majority of sessions with a first-party answer on every order
- Push survey answers into Klaviyo as customer properties that drive segmented post-purchase flows
- Feed a RICE-scored CRO test backlog that now sits at the center of how they run every client engagement
"It's the only data source that tells you why, not just what, and it costs almost nothing to run compared to what a single good insight is worth in ad spend saved or a checkout fix shipped."
Why Grafikr adds Zigpoll to client stores
The questions cookie-based attribution stopped being able to answer
Grafikr's clients were asking a question their analytics stack could no longer answer honestly. iOS privacy changes and cookie blocking had broken cookie-based attribution across most sessions, and no amount of dashboard work in GA4 or Meta was going to fix it:
"'Where do our customers actually come from' had no reliable answer, since cookie-based attribution in GA4 and Meta was failing on the majority of sessions due to iOS privacy and cookie blocking.
We also had no good way to answer 'why do people abandon at checkout' beyond guesswork, no way to separate gift buyers from self-purchasers, and no way to size how big word-of-mouth actually was as an acquisition channel."
Those four gaps have something in common: none of them are measurement problems. They're all questions where the customer is the only one who holds the answer.
On the Order Status page with no dev lift
Grafikr runs Zigpoll across a portfolio of client stores, so anything that requires a development ticket per client is a non-starter:
"Three things: it drops straight onto the Shopify Order Status (thank-you) page with no dev lift, the pricing is fair, and it answers zero-party-data questions none of the other tools in our stack can touch."
The four-question survey that goes in at kickoff
The post-purchase survey on the Order Status page is always the first thing in, usually one of the first things installed during a new client's kickoff phase. The format is standardized across every client:
- Who are you buying for? — self, gift, or business
- Why did you buy today? — single choice, plus an "other" option
- How did you first hear about us? — channel attribution
- Was anything close to stopping you from buying today? — optional open text
Each answer is tagged onto the customer in Klaviyo as a custom property, which means the survey isn't only a reporting exercise — it's segmentation that drives the post-purchase flows that run afterward.
"Setup is typically 1–2 hours; we then roll the same survey out market by market as the client expands into new Shopify Markets storefronts."
Takeaway — standardizing the same four questions across every client is what makes the data compound. Grafikr isn't starting a new research project per brand; they're running one instrument they already know how to read.
Turning open-text friction into a test backlog
The optional open-text question is the one most teams treat as qualitative color. Grafikr treats it as a queue. Answers get categorized on a monthly cycle, and there's a hard threshold for what earns a place on the roadmap:
"Open-text friction answers get categorized monthly, and anything mentioned by 10% or more of respondents goes straight onto the CRO test backlog."
Takeaway — a stated threshold is what turns customer verbatims into a process. Without one, open-text answers become anecdotes that get quoted in decks; with one, they become tests that get shipped.
The client win: 3,000 survey responses and 2,600 reviews in one roadmap
For one client, Grafikr combined almost 3,000 Zigpoll post-purchase responses with more than 2,600 Stamped reviews into a single RICE-scored CRO prioritization list. The survey surfaced two things the client had never quantified:
"Sizing complaints were the single biggest driver of negative reviews, and word-of-mouth was their number one acquisition channel, but completely unmonetized."
That second finding is the kind of thing no ad platform will ever report. A channel that generates more first-touch demand than anything else, with no budget line, no referral program, and no attribution credit attached to it.
Alongside it came the checkout-friction quick wins the open-text field had been collecting — adding PayPal, and a welcome discount. Together, the analysis reshaped the client's entire CRO roadmap.
Takeaway — survey responses and review data answer different halves of the same question. Reviews tell you what went wrong after the purchase; post-purchase surveys tell you what almost stopped it from happening. Scored together, they rank far better than either does alone.
How Zigpoll changed the way Grafikr sells and delivers
Zigpoll has moved from a tool Grafikr reaches for into a fixed part of their delivery model. It's a standard line item on every new client's kickoff checklist, and the data it produces is the raw material for the test backlog they build for every CRO client:
"The survey data plus review data is what feeds the RICE-scored test backlog we build for every CRO client."
Roughly ten client brands are running Zigpoll today — several of them multi-market, so meaningfully more storefronts than that once each market instance is counted separately.
What Grafikr would tell another agency
Erik's advice to agencies weighing it up is to stop treating it as optional:
"Install it before you think you need it. It's the only data source that tells you why, not just what, and it costs almost nothing to run compared to what a single good insight is worth in ad spend saved or a checkout fix shipped.
If you run CRO or lifecycle work for clients, treat it as table stakes, not an add-on."
It's a fair way to frame the cost. Four questions on the thank-you page is the cheapest line item in an agency retainer, and it's the only one that answers the question every other tool in the stack leaves open.
