Calculating the Projected ROI of ChatGPT Visibility
A step-by-step model for 12-month forecasting, with every monthly figure shown.
Richard RowleyCount only the clicks and ChatGPT visibility loses money in year one. Count the answers and it returns 61%. This model shows both numbers and every step of the working.
SEO forecasts start with keyword volume. ChatGPT publishes no prompt volume, so this forecast starts with the share of answers that name your brand. We call that the presence rate. Raise it and revenue follows by two routes. Some people click the link in the answer. Others see your name, leave, and search for you later.
The method forecasts the percentage ROI of AI visibility work across the first 12 months. It uses the ratios from our forecast of the channel switch from search to AI. Treat every ratio as a starting point and replace it with your own data.
Step 1: Identify and Categorise AI Visibility Initiatives
Start by listing the work and when its effects begin. Our example has three initiatives.
- Answer-ready pages. Rewrite your key pages so each one states its facts plainly: a direct answer at the top, then prices, specifications and comparisons. Effects begin in month 2.
- Brand facts. Make your name, products and claims consistent everywhere a model reads them: your site, profiles, directories and product feeds. Effects begin in month 3.
- Third-party sources. Earn coverage on the sites ChatGPT cites for your category: reviews, trade press, comparison pieces and forums. Effects begin in month 4.
A fourth line runs all year: tracking. A fixed set of prompts, run every month, gives you the presence rate. That is your baseline and your result.
Step 2: Estimate Costs and Incremental Benefits
Costs. Price each initiative and add the tracking tool. We use rocketblue. Our example totals £18,400.
| Item | Cost |
|---|---|
| Answer-ready pages | £5,000 |
| Brand facts | £3,000 |
| Third-party sources | £8,000 |
| rocketblue tracking, £200 a month for 12 months | £2,400 |
| Total | £18,400 |
Incremental benefits. Work out what one point of presence is worth. Our example is a retailer in a category with 500,000 UK searches a month.
- Prompt volume. Take 20% of search volume. That gives 100,000 prompts a month.
- Presence. The brand appears in 10% of answers today. The three initiatives add 15 points: 4 from answer-ready pages, 3 from brand facts and 8 from third-party sources. The target is 25%.
- Answers. One point of presence is 1,000 extra answers a month that name the brand.
- Clicks. 40% of those answers carry a link and 7% of the links get clicked. That is 28 visits.
- Later brand visits. 2% of people who see the answer search for the brand later. That is 20 visits.
- Orders. The 48 visits convert at 4%, twice the organic rate of 2%. That is 1.92 orders.
- Revenue. At an average order of £125, that is £240.
One point of presence is worth £240 a month. Everything after this is multiplication.
The 2% in line 5 is the softest number in the model. Step 5 tests it.
Step 3: Timeline for Benefit Realisation
Each initiative ramps over four months: 25% of its effect in the first month, then 50%, 75% and 100%. Models refresh their sources at their own pace, so gains arrive gradually.
- Answer-ready pages start in month 2 and reach their full 4 points in month 5.
- Brand facts start in month 3 and reach their full 3 points in month 6.
- Third-party sources start in month 4 and reach their full 8 points in month 7.
From month 7 the brand holds all 15 points.
Step 4: Calculate Monthly Returns
Add the points gained each month and multiply by £240. The 12 months return £29,640.
| Month | Answer-ready pages | Brand facts | Third-party sources | Points gained | Extra answers | Return |
|---|---|---|---|---|---|---|
| 1 | 0 | 0 | 0 | 0 | 0 | £0 |
| 2 | 1 | 0 | 0 | 1 | 1,000 | £240 |
| 3 | 2 | 0.75 | 0 | 2.75 | 2,750 | £660 |
| 4 | 3 | 1.5 | 2 | 6.5 | 6,500 | £1,560 |
| 5 | 4 | 2.25 | 4 | 10.25 | 10,250 | £2,460 |
| 6 | 4 | 3 | 6 | 13 | 13,000 | £3,120 |
| 7 to 12, each month | 4 | 3 | 8 | 15 | 15,000 | £3,600 |
| 12-month total | 123.5 | 123,500 | £29,640 |
Months 1 to 6 return £8,040. Months 7 to 12 return £21,600.
Step 5: Calculate Total Return and ROI
Total return. £29,640 across 12 months.
ROI. ROI = (Total Return - Total Cost) / Total Cost x 100. That is (£29,640 - £18,400) / £18,400 x 100, or 61%. Returns pass costs in month 9.
Now test the soft number. Change the later brand visit rate and the answer swings from a loss to a clear win.
| Scenario | Later brand visits | Value of one point | 12-month return | ROI |
|---|---|---|---|---|
| Clicks only | 0% | £140 | £17,290 | -6% |
| Central | 2% | £240 | £29,640 | 61% |
| High | 4% | £340 | £41,990 | 128% |
The spread is the lesson. Judge ChatGPT on referral clicks and this project loses 6% in year one. The people who see your name and search for you later decide the case. Measure them.
Year two is stronger. The build costs are spent. A brand that holds its 15 points earns £43,200 across the year against £2,400 of rocketblue tracking.
The example uses revenue, as our SEO ROI model does. Use gross margin for a stricter test.
Replace Our Numbers With Yours
Seven assumptions drive the model. Each one has a place where your own figure lives.
| Assumption | Example value | Where to find your own |
|---|---|---|
| Prompt volume | 20% of search volume | Derived. ChatGPT publishes no prompt data, so keep this one cautious. |
| Starting presence | 10% of answers | The first run of your tracked prompt set |
| Answers with a link | 40% | Your tracker |
| Clicks per linked answer | 7% | GA4 referrals from chatgpt.com, set against your tracked citations |
| Later brand visits | 2% of answers | Branded search impressions in Search Console, before and after presence rises |
| Conversion rate | 4% | GA4, AI referral sessions |
| Average order | £125 | Your own sales data |
Run the model with your figures. If the clicks-only case pays, start the work. If the case rests on later brand visits, track branded search from day one.
Want the model run on your numbers? Get in touch.