How Do You Know Whether to Change Your Target Customer or Your Offer? Patch's Indoor Plant Shift
Short answer: Test the customer and the offer as separate variables before changing both. Expose each relevant cell to enough suitable buyers to secure 20 to 30 normal-priced orders under comparable price and acquisition conditions. Change the customer when the unchanged offer performs better with the new group; change the offer when the variation improves results with the current group; call it a combined shift when both must move.
Weak sales do not tell you which assumption failed. You may be reaching people who do not feel the problem, or offering the right people a version they do not want. Changing the audience, product and message together can improve sales while leaving you unable to explain why.
Patch's history is a useful warning against forcing a neat label onto a messy decision. The company began with outdoor potted plants for London balcony owners, then found greater unmet demand for indoor plants. That changed the primary customer context and the assortment, while retaining the broader job of helping inexperienced urban customers choose, receive and care for plants. It was not a pure customer-segment change with an unchanged product. Patch describes the move from balcony plants to indoor plants here, while founder Freddie Blackett explains the original balcony problem here.
Use the Segment-or-Offer Diagnosis Grid
The Segment-or-Offer Diagnosis Grid separates two decisions that founders often collapse into one.
| Test cell | Customer | Offer | What the result tells you | |---|---|---|---| | Baseline | Current | Current | How the business performs now | | Segment test | Candidate | Current | Whether the original offer works better for someone else | | Offer test | Current | Modified | Whether the original audience wants a different version | | Combined test | Candidate | Modified | Whether the strongest opportunity requires both changes |
My position is that you should not describe a combined shift as proof that the original product was right and only the audience was wrong. That story may sound reassuring, but it teaches the wrong lesson. Changing both can be commercially sensible. You still need to know whether future growth depends mainly on reaching a different buyer, supplying a different offer, or preserving the interaction between the two.
Define the variables before testing
Write the current proposition in four lines: who buys, what situation triggers the purchase, what is delivered, and what the customer pays. Then write the proposed change beside it. If more than one line changes, you are not running a clean segment test.
Treat the offer as more than the physical item. Range, size, packaging, service level, delivery timing, support and guarantee can all change what the customer is buying. Treat the segment as more than age or location. A useful segment shares a problem, buying situation, route to purchase and ability to pay.
Patch retained a broad outcome, making plant ownership easier for inexperienced city residents, but indoor plants changed the use setting and stock emphasis. Educational content and aftercare also helped remove the confidence barrier. Calling all of that the same product would hide material operating choices.
Use [customer-segment selection](/market-research/first-profitable-customer-segment/) to define buyers by behaviour and economics, not by a loose demographic label.
Read the signal before choosing the test
Different evidence points to different first moves.
| Observed behaviour | Likely first test | Reason | |---|---|---| | A different group buys the current offer without special changes | Segment test | The offer may already work for a better buyer | | Current buyers repeatedly request the same version or outcome | Offer test | The audience may be right while the offer is incomplete | | A new group buys only after a material variation | Combined test after isolation | Both customer and offer may matter | | Nobody pays under either condition | Return to the problem and demand evidence | Neither change has earned priority |
Unsolicited behaviour is stronger than a brainstorm. Record who asked, the exact situation, what they already use, whether they paid, the acquisition source and the service work created. Ten thousand views from a new audience are weaker than 20 normal-priced purchases.
Do not let one unusually enthusiastic customer define a segment. Look for repeated buying reasons among people who do not know you personally.
Run the segment and offer tests under comparable conditions
Start with the segment cell because it usually costs less than changing stock or operations. Present the current offer, at the normal price, to one candidate group. Change only the language required to describe the same value accurately. Keep fulfilment and terms stable.
Then test one bounded offer variation with the current segment. The change might be an indoor rather than outdoor plant set, a smaller service scope or a different pack size. Hold the acquisition source and price logic as steady as practical.
For each cell, record:
- suitable prospects or visits
- normal-priced orders
- revenue and variable cost
- acquisition cash and founder time
- refunds, replacements and support
- repeat or referral behaviour in the natural cycle
Use the same observation window. A two-week candidate test compared with a six-month baseline will mix seasonality and learning with the variable you intended to examine. Twenty to 30 orders per cell is a useful working minimum for a low-value purchase, not a statistical guarantee.
Worked example: RoomRoot Plants
RoomRoot sells plant-and-pot sets for £62. Its current audience is flat owners with balconies, and its current offer is an outdoor set. It suspects that private renters in city flats are a better audience, but that group also asks for low-light indoor plants. RoomRoot runs all four grid cells with 200 suitable visits and £240 of acquisition spend in each.
The outdoor set costs £27 for the plant, pot and packing, £8 to deliver and £4 for support and replacements, leaving £62 - £27 - £8 - £4 = £23 before acquisition. The indoor set costs £26 plus the same £8 delivery. Support and replacements are £4 for balcony owners and £6 for inexperienced private renters, leaving £24 and £22 respectively.
| Test cell | Orders | Conversion | Contribution before acquisition per order | Net cohort contribution | |---|---:|---:|---:|---:| | Balcony owners, outdoor set | 20 | 10% | £23 | 20 x £23 - £240 = £220 | | Private renters, outdoor set | 6 | 3% | £23 | 6 x £23 - £240 = -£102 | | Balcony owners, indoor set | 18 | 9% | £24 | 18 x £24 - £240 = £192 | | Private renters, indoor set | 32 | 16% | £22 | 32 x £22 - £240 = £464 |
The candidate segment performs badly with the unchanged offer. The indoor variation almost matches the baseline among existing buyers, while the combined cell produces the strongest conversion and more than twice the baseline cohort contribution. The evidence does not support saying, "we had the right product for the wrong customer". It supports a combined move in which the indoor offer is necessary and the renter segment amplifies it.
RoomRoot should lead with the indoor range for renters, retain the profitable outdoor set during transition and measure repeat purchases and replacements before expanding stock. These figures are illustrative, not plant-retail benchmarks. Replace the price, costs and thresholds with your own recorded figures.
Make the decision from economics, not conversion alone
A cell can win more orders and still be worse. Calculate contribution after product, fulfilment, acquisition and segment-specific service. Include founder time where one group needs repeated explanation or custom handling.
The candidate audience may also be easier to reach but too small, or large but expensive to serve. Check reachable depth, purchase frequency and operational fit after the test. A segment that converts at 16% during one local promotion is not yet a durable market.
Choose the smallest honest change supported by the evidence. If the segment cell wins, move acquisition and positioning first. If the offer cell wins, modify the offer before abandoning buyers you already understand. If only the combined cell works, plan inventory, messaging and operations as one controlled transition rather than pretending one variable stayed fixed.
Complete the diagnosis in 30 days
In the next two days, classify the last 30 enquiries by buyer, use situation, requested offer and outcome. By day five, define one candidate segment and one offer variation, with price and contribution thresholds written in advance. Run the segment cell first, then the offer cell, using comparable traffic and a complete buying cycle. Add the combined cell only when either isolated test produces evidence worth pursuing. On day 30, choose the winning cell, name exactly what changed and set the next 30-order test. Do not rewrite the whole brand from a handful of interviews.
Related guides
Frequently asked questions
Can I change the customer and the offer at the same time?
You can, but the result will tell you that the bundle worked, not which change caused it. That may be acceptable when cash is short and one obvious combination can be sold immediately. Record it honestly as a combined test. Before investing heavily, run at least one isolation cell: offer the old version to the new group or the new version to the current group. The comparison shows what must remain stable as you scale. If safety, regulation or technical feasibility makes one cell inappropriate, skip it and state the limitation rather than manufacturing a clean experiment.
How many sales do I need before changing direction?
For a low-value consumer offer, 20 to 30 normal-priced orders in each relevant cell is a practical starting point. Extend the test when one week, channel or buyer accounts for much of the result. A high-value business service may produce useful evidence from fewer sales, but you then need deeper evidence on sales time, implementation, contribution and renewal intent. The number is not a universal confidence threshold. Decide it before promotion, exclude friends and heavy discounts from the main comparison, and keep measuring after the move because early novelty can exaggerate demand.
What if the new segment asks for one extra feature?
Treat the feature as an offer change, even if it looks small. First record how many suitable buyers request it, which problem it solves and whether they will pay more or buy sooner. If possible, sell the unchanged offer to the candidate group before adding it. Then test the feature with a defined price and delivery cost. A request that changes compliance, fulfilment, support or the core promise is not a minor adjustment. Do not attribute improved sales entirely to the new customer when the added feature may be doing the commercial work.
Should I stop serving my original customers?
No, not while they remain profitable, serviceable and consistent with the revised promise. A target segment determines where you concentrate acquisition and product decisions; it does not ban every other buyer. Keep the original route during a controlled transition, then compare contribution, support and repeat behaviour. Reduce it if stock, messaging or scarce capacity conflicts with the new route. Give customers clear notice if availability or service will change. An abrupt withdrawal based on one promising cohort destroys cash and useful comparison data that could have shown whether the shift survives beyond novelty.
Can customer interviews tell me which variable is wrong?
Interviews can identify the likely variable, but they cannot settle the decision without behaviour. Ask about the last time the problem occurred, what the person used, what it cost and why they did not buy your current offer. Then request a normal order, paid pilot or another proportionate commitment. If current buyers describe the same missing outcome, test the offer. If a different group already improvises with your current product, test the segment. Avoid asking whether people prefer two hypothetical ideas. Polite preferences create less evidence than one purchase under clear terms.
What if the new segment is larger but less profitable?
Do not move merely because the audience is larger. Calculate contribution per order, acquisition cost, service time, repeat behaviour and the capacity required at the expected volume. A lower first-order contribution can still work when acquisition is reliably cheaper or repeat buying is stronger, but those benefits need observed evidence. Model a base case and a downside case before changing stock or staff. The right segment is reachable, valuable and supportable. A large group that consumes cash on every sale makes the failure bigger rather than making the opportunity better.
Do I need a new business name after a combined shift?
Usually not during the test. Change the entry message, examples and stock emphasis first. A name needs attention when it explicitly excludes the new buyer, promises the old offer or creates legal and trust problems. Renaming at the same time introduces another variable and can erase recognition you still need. Test a descriptive page or bounded campaign under the existing identity, then measure confusion in actual enquiries. Trade mark, domain and company-name implications vary by country, so check current registers and obtain qualified local advice before adopting a new legal or trading name.
Comments (0)