Shopify Sidekick can build customer segments from purchase behaviour, audit your existing tags for duplicates and dead entries, explain whether RFM scoring is worth the effort for a store your size, and wire up a native repeat-purchase reward with Shopify Flow. What it can’t do is run a real points program — Shopify has no points ledger, and no prompt changes that. Here’s how to get the parts that do work in the right order.
Step 1: Audit your tags before you build anything on them
Segments built on messy tags inherit the mess. If VIP, vip, and VIP-Customer all exist, every segment referencing “VIP” is quietly wrong, and you won’t notice until a campaign under-delivers.
Ask Sidekick for an audit — explicitly an audit, with no changes applied:
- Every distinct tag currently applied to any customer, with the number of customers holding each
- Likely duplicates and typos grouped into suggested merge sets — plural versus singular, case differences, near-identical wording
- Dead tags applied to fewer than
[MIN_CUSTOMER_COUNT]customers, usually leftovers from a one-off campaign or a Flow nobody turned off - The whole thing as a table: tag name, customer count, suggested action (keep / merge into X / review / likely dead)
Then stop and read it. The consolidation itself belongs in a separate, follow-up prompt where you name the exact tags to merge. Bulk changes to live customer records are a decision, not a cleanup pass — and a merge that looked obvious in a table sometimes turns out to be two genuinely different groups once you check who’s in them.
Step 2: Decide how sophisticated your segmentation needs to be
Before you build segments, spend one prompt on the strategy question: simple recency, or full RFM scoring?
Ask Sidekick to explain, in plain language, what RFM adds that recency misses — and specifically to give you a concrete example of two customers who land in the same recency bucket but should be treated completely differently. That example usually settles the question faster than any framework. A customer who spent $3,000 across eight orders and went quiet is not the same as one who bought a $40 item once and never returned, yet “hasn’t ordered in 90 days” treats them identically.
Also ask for:
- The setup and maintenance tradeoff between the two approaches
- Whether RFM is worth building as a recalculated customer metafield, versus running several parallel simple segments — recency, frequency, and AOV — that overlap
- A rule of thumb for when a store outgrows simple segments
Keep this prompt conceptual. Don’t have it pull store data or build anything yet; the point is to choose an approach before you commit effort to it. Most stores are better served by three overlapping simple segments than by an RFM system nobody maintains.
Step 3: Build the segments that map to real campaigns
Now build. The segments worth having are the ones attached to something you’d actually send:
- VIP customers — spent over
[AMOUNT]total, or placed more than[X]orders - At-risk customers — previously active, nothing in the last
[90]days - New customers — first purchase in the last
[30]days - High-value prospects — added to cart, never purchased
- Repeat buyers — 2 or more purchases in the last
[6 months] - Geographic — customers in
[CITY/STATE/COUNTRY]
Pick your own thresholds. Sidekick has no idea what “high spend” means in your category, and a default it invents will be wrong in one direction or the other.
For each segment, ask it to attach four things: the best campaign type (email, SMS, or ads), a specific offer or message angle that fits that group’s situation, the revenue impact if you convert [X]% of them, and how often you can contact them without causing fatigue. That last one matters more than it sounds — VIP segments get over-mailed precisely because they respond, right up until they don’t.
Step 4: Run a native repeat-purchase reward
This is where expectations need setting. A real points program — a balance customers accumulate, watch, and redeem against reward tiers — needs a third-party app such as Smile or LoyaltyLion, because Shopify has no native points ledger. Sidekick can’t build one and shouldn’t pretend to.
What it can build is the incentive without the ledger. Ask for a Shopify Flow that:
- Triggers when a customer’s order count reaches
[ORDER_COUNT_TRIGGER]— every third order, for example - Generates a discount code worth
[DISCOUNT_VALUE]off their next purchase - Tags the customer with
[LOYAL_CUSTOMER_TAG] - Sends a Shopify Messaging email telling them about the reward
Then have it create a customer segment of everyone carrying [LOYAL_CUSTOMER_TAG], so you can watch your repeat-customer base as a number that moves rather than a feeling.
For plenty of stores this is genuinely enough. The behaviour you want — come back and buy again — gets rewarded automatically, without an app subscription or a points balance to explain. Where it falls short is anything requiring tiers, a visible balance, or point redemption at checkout. If your customers need to see progress toward a reward, that’s the moment an app becomes worth paying for.
The loop that keeps working
Audit tags, choose your segmentation depth, build segments tied to campaigns, reward repeat purchases natively. Then re-run the tag audit periodically, because every new Flow and every new staff member writes tags — and the mess always comes back.