Pricing a Micro SaaS: 4 Frameworks and When to Use Each
“What should I charge?” is the question solo founders agonise over for weeks and usually answer wrong. Most micro SaaS undercharge by 2–3× for the first year. These four frameworks give you a structured way to land on a price you can defend.
1. Cost-plus pricing
How: Calculate your per-customer cost (hosting, AI API, support time, third-party services), add your desired margin, that's the price.
When it works: Almost never. Cost-plus ignores what customers will actually pay; usually produces prices far below what the market would bear. Useful only as a floor — you need to charge at least cost-plus-something to not lose money.
Example: AI API costs £2/customer/month, hosting £0.50, support 30 min/month at £35/hr = £17.50. Cost-plus 50% margin = £30/month. Almost certainly too cheap if your product is doing real work.
2. Value-based pricing
How: Estimate the value your product creates for the customer (time saved, revenue increased, costs avoided). Price at 10–30% of that value.
When it works: Best framework when you can quantify value clearly. Particularly powerful for automation products where the value math is obvious.
Example: Your product saves the customer 5 hours/month at £35/hour loaded cost = £175/month of value. Price at 20% capture = £35/month. Or for high-value workflows, capture 10% of a £2,000 saving = £200/month.
3. Competitor-anchored pricing
How: Find 3–5 direct competitors. Price within their range, positioned by how your offering compares (cheaper if you offer less, more expensive if you offer more).
When it works: When the category is established and customers already have price expectations. Risky in new categories where you might be anchoring against confused competitors.
Example: Competitors charge £29–£99/month. Your offering matches mid-market features → £49–£59/month. The competitors did the willingness-to-pay research for you, mostly.
4. Willingness-to-pay testing
How: Set up a fake pricing page. Show different prices to different visitors. Measure click-through to signup at each price point. The price that maximises (CTR × price) is your sweet spot.
When it works: When you have meaningful traffic (1,000+ pricing-page visits before you can read the signal). Less useful at zero traffic, very useful from month three onwards.
Example: Test £29, £49, £79 across three audience segments. £49 produces best CTR-price product. Iterate to £39 vs £49 vs £59. Tune within ±£10 ranges.
The framework we'd actually use
For most new micro SaaS in 2026, combine the four like this:
- Cost-plus sets your floor — never go below this.
- Value-based sets your ceiling — never go above what you can defensibly justify.
- Competitor-anchored sets your starting range within that band.
- Willingness-to-pay testing tunes within the range once you have traffic.
The pricing mistakes solo founders make
- Free tier too generous — most users never upgrade. Make the free tier's ceiling visible early.
- Annual discount too aggressive — 20% off annual is plenty; 50% is leaving money on the table.
- Too many tiers — 3 is right. 5+ creates decision paralysis.
- Listing every feature in every tier — customers compare what's different, not what's the same.
- Avoiding the highest tier — every micro SaaS should have a £200+/month tier even if barely anyone buys it. It anchors the perceived value of the cheaper tiers.
When to raise prices
You're ready to raise prices when 60%+ of new prospects accept the price without negotiation. You're too cheap when 95% accept; you're too expensive when fewer than 30% accept. Most micro SaaS sit in the 90%+ range for years and never realise.
Raising prices by 30% typically loses 10–15% of conversions but more than makes up for it in revenue. Grandfather existing customers at their old price for goodwill; charge new customers the new price. Net effect: more revenue, same headcount, no customer outrage.
See the micro SaaS pillar for the broader pragmatic engineering and product approach.
Got a workflow you want to talk through?
30 minutes, no pitch. We'll tell you honestly what we'd build — or whether automation isn't right yet.