Every proposal for search or conversion work runs into the same wall. The client asks what they will get back, and the honest answer contains a range and a delay. So most agencies do one of two things: they quote a number they cannot defend, or they avoid the question and lose the deal to someone who didn’t.
There is a third option, and it converts better than both: show the model, not the number.
Why the usual ROI slide fails
It uses revenue instead of margin. “This project returns 800%” is calculated on incremental revenue. If the client’s gross margin is 22%, the return on the money they actually keep is a quarter of that. A CFO spots this in seconds, and everything else in the proposal loses credibility with it.
It presents one scenario. A single optimistic outcome is a claim. A range with a stated conservative case is an analysis. Clients do not buy claims from suppliers they have not worked with yet.
It hides the assumptions. If the client cannot see which input produced the result, they cannot argue with the input — so they argue with the conclusion, which is the argument you can’t win.
It ignores time. A number with no payback period is unusable, because the client is comparing your project against other places the same money could go this quarter.
The five inputs a defensible case needs
- Current baseline. Monthly qualified traffic and the measured conversion rate. Measured — not the industry average.
- Modelled improvement, expressed in percentage points rather than percent. Going from 1.8% to 2.2% is +0.4 points, not “+22% conversion”, and points are much harder to inflate.
- Average customer value, and whether it is a one-off purchase or a contract with a lifetime.
- Gross margin. This is the input that makes the difference between a model and a sales pitch.
- Full cost: your fee, plus the client’s internal time, plus any tooling or development they will have to pay for.
From those five you can produce the four outputs that actually get budget approved.
The four outputs that get a yes
Net benefit over a fixed horizon — usually 12 months, on margin, not revenue.
Payback period. How many months until the project has returned what it cost. Anything under six months is easy to approve at most companies; anything over twelve needs a strategic reason as well as a financial one.
Break-even conversion lift. This is the most persuasive single figure in the whole case, and almost nobody presents it. It answers: how small an improvement does this need to produce before it stops losing money? When the honest answer is “an increase of 0.08 percentage points”, the risk conversation ends. You are no longer asking the client to believe an ambitious forecast — you are showing them how much has to go right for the project to merely break even.
A sensitivity range. Recalculate the case with the conversion lift halved. If the project still pays back within the horizon, say so explicitly. That single sentence does more for trust than any case study.
How to run the conversation
Do not present the finished number. Build it live, with the client’s own figures, in front of them.
Two things happen when you do this. First, the client corrects your inputs — which means they are now defending the model rather than doubting it. Second, the assumptions become theirs. A number a client typed is a number a client will repeat to their own board.
This is also why gated ROI calculators underperform: the visitor who already knows their baseline conversion rate is your most qualified prospect, and they are exactly the person who will not trade an email address for a number they suspect is inflated. Show the result first, ask for contact details second.
Do it with a tool that shows its workings
The NOVA ROI calculator was built for this exact conversation. It models eight assumptions — traffic, customer value, current and target conversion rate, gross margin, horizon, recurring cost, upfront investment — and returns net benefit, ROI, payback period and break-even lift with the formulas visible rather than hidden. You can switch between conservative, balanced and ambitious scenarios in one click, export the scenario to JSON or CSV, and print a clean one-page report to attach to the proposal.
It is a live demo with nothing gated, and it runs entirely in the browser: no inputs are sent anywhere, which matters when the numbers on screen belong to your client. If you want to put it on your own site with your branding, the editable kit is a single set of HTML, CSS and JavaScript files with no framework, no build step and no subscription.
For the reporting side of a retainer — dashboards, budget trackers, client-facing summaries — see Business & Marketing Tools. For the landing pages and lead-capture pieces around the pitch, see Website & Digital Product Kits.
FAQ
Should I promise a specific ranking or traffic number?
No. Model the conversion and revenue mechanics, state the range, and put the delay in writing. Search results are not something you control, and a promise you cannot keep costs the renewal.
What if the client doesn’t know their gross margin?
Ask for a band rather than a figure, and model the bottom of the band. A case that works at the pessimistic end of the range is far easier to approve than one that needs the optimistic end.
How long before an SEO project should be judged?
Set the review point in the proposal — typically leading indicators (impressions, indexed pages, rankings for target queries) at 90 days, and revenue impact at six months. Agreeing the measurement window in advance prevents the project being declared a failure at week six.
Is CRO easier to justify than SEO?
Usually yes, because the traffic already exists and the improvement is measurable within weeks. If a client is nervous about a long SEO commitment, sell the conversion work first and let the measured result fund the search work.
