I just received a paper copy of the Norwegian Research Council’s annual R&D statistics for 2012, which includes success rates (or conversion rates) for certain R&D programs. I was curious, as there have lately been extensive discussions about the use of time on project acquisition at Norwegian universities (see for example www.difi.no/filearchive/difi-rapport-2012-14-ute-av-kontroll.pdf), and it is of interest to understand the relationship between time spent and realized conversion rates. I also frequently get questions about what constitutes an acceptable conversion probability, for a specific opportunity or channel, both from R&D organizations and commercial organizations. In this blog I will present the results of some informal research, including an endeavor to provide robust estimates of conversion rates from microeconomic fundamentals.
Here are some examples of reported conversion rates, from a number of industries and channels and spanning four orders of magnitude (which explains the logarithmic scale):
Note the following:
- The data above come from a variety of sources, from hard statistics provided by governmental organizations to rough estimates based on anecdotal evidence obtained through informal discussions with industry sources. The general picture if however believed to be accurate.
- While then end point of a sales process is generally well defined (= financial commitment), the starting point for such process is frequently less well defined, especially for proactive situations. For example, does a proactive sales process start in first open-ended sales meeting or when sending the first proposal?
- Some processes involve significant guidance from the customer-side during the process, so that for example research grant applications will not be submitted unless expectation of reasonable success rate.
- The list above contains a mix of vendor-side and customer-side data. The two data sets may not be fully comparable.
Let us first try to understand at a qualitative level the significant variability of conversion rates:
- The list covers typical proactive situations (the potential vendor contacts the potential customer) as well as reactive situations (the potential customer contacts a number of potential vendors).
- The list above covers the full range of rights-based situations (Skattefunn, which should in principle give a potential conversion probability of 100%), through soft budgetary constraints (e.g., typical insight-based consulting engagement), to hard budgetary constraints (e.g., purchasing of auditing services, and from one and only one vendor).
- The marginal cost of submitting a proposal may vary significantly across situations (from zero cost for mass email to tens or hundreds of millions of USD for compiling a proposal from a defence contractor, say for a fighter aircraft programme).
- The constraint on capacity for writing a good proposal may also vary significantly, from no constraint on mass email to some constraint for an under-utilized professor at a mid-tier university to hard constraints for a commercial software organizations.
My hypotheses, when starting on this research, was that most companies are in practice faced with opportunities with a range of conversion probabilities, from 0% to 100% and depending on pricing and effort invested, and they organize their sales activities so that on the margin (= the least attractive deal actually pursued) the marginal deal is marginally attractive, in microeconomic terms. If true, it means that predictions of marginal conversion probabilities can be constructed based on microeconomic fundamentals.
We will focus on the following simple case: A range of a priori conversion probabilities, no proposal preparation capacity constraint, and no dependency of conversion probability on pricing.
The four microeconomic variables of interest here are: cost of opportunity COO per opportunity (including for example effort, travel costs, lead list costs, and non-refundable tender material fees); the contribution margin CM or price P less variable costs VC per opportunity; the proposal preparation effort E spent per opportunity; and the conversion probability p(E, P) per opportunity. In addition, denote by C the proposal preparation capacity, and by N the number of opportunities to pursue.
In this simple case of no proposal preparation capacity constraint, and no dependency of conversion probability on pricing, microeconomic theory tells us to pursue the N opportunities with conversion probability p(i) ≥ COO (i) / CM (i), skip others.
(In the more realistic case of proposal preparation capacity constraints, one must adjust the formula above to account for the detrimental effect of spreading effort too thinly, and also to optimize on effort per opportunity.)
Average conversion probability is Σ p(i) / N, which is higher than marginal conversion probability, but for mathematical reasons they tend to be fairly close. Here are some examples:
|Scenario||Marginal conv. rate||Average from chart above||Assumption||Comment|
|Defence contractor bidding on large project||27%||30%||NOK 400m in proposal preparation cost, NOK 1,5b in contribution margin||Source: 4% bid cost from training material Hyman Silver|
|Research organization applying for governmental fund from NRC||16%||28%||20 days work at NOK 1 500 per hour, NOK 1,5m grant|
|Auditing services||24%||25%||40 hours work at NOK 1 250 per hour, contribution margin of NOK 75 000 per year, 2.75 years||Source: personal communication auditor, typical case|
|Training provider selling small-class 2-day engineering course by cold-calling using third-party list||7,7%||8,2%||NOK 500 in lead cost and telephoning cost, NOK 6 500 in contribution margin per person|
|Small Indian provider of web design and web services, selling through mass email in email||0,25-0,50%||0,12%||NOK 50-100 in lead cost, NOK 20 000 in contribution margin||This is the only strange case, with estimated cut-off higher than estimated average. I suspect the mass email is ineffective, but that this is not understood by vendor|
In conclusion, it seems that we can for any industry or channel and across multiple orders of magnitude robustly predict typical conversion rates from microeconomic fundamentals, and the underlying model robustly explains the significant variability of conversion rates across industries and channels.
The exception is really research organizations, for which we predict an optimum marginal conversion probability which is significantly lower than what is observed historically (that is, too few opportunities are pursued). There could be three reasons for this anomaly: i) I don’t have accurate estimates of microeconomic variables; ii) research scientists use another concept of value to the organization than contribution margin, which is probably true; or iii) Norwegian research scientists submit too few (not too many) applications for research grants.
Here are the implications for your organization:
- Measure your marginal conversion rate based on historical values, either for all opportunities or for opportunities at the margin.
- Compare this value with what COO / CM, and with industry averages, and assess whether it is too low or too high.
- If materially higher than COO / CM, you should probably increase your ambition level in terms of # opportunities pursued (or adjust pricing upwards), as you are leaving money on the table.
- If materially lower:
- Either, reduce # opportunities pursued.
- Or, improve the quality of your proposal preparation process.
- Or, reduce pricing.
- In general, try to attach a subjective conversion probability to all future opportunities, and decide on whether to pursue based on COO / CM.
- Do (5) across a set of pricing scenarios. But, estimating conversion probability as function of price (and effort) is notoriously difficult, and pricing is besides over-valued as differentiator for technology SMEs.
That said and with reference to (5), many organizations may justify spending time on selected low-probability deals with great learning potential, that is, they are pursued not to be won, but as small business experiments.