The Internet has radically changed human life, and has had an equally significant impact on business. It is widely used in organizations for marketing and promotion of products and services. The Internet is also used to deliver customer support, share information and provide training to employees.
However, successful Internet Marketing (without AI) is becoming harder to achieve. Competition (especially for user / customer attention) has been intensifying while complexity has been increasing as new channels and capabilities are constantly being added. Traditional (Descriptive) Analytics and SEO (Search Engine Optimization) tools and strategies are no longer meeting the needs of marketing professionals.
What are key Internet Marketing challenges?
Companies wanting to leverage Internet Marketing to grow their business are having challenges crossing The Internet Marketing Chasm (much like the chasm faced by companies introducing new products / technology). In the case of Internet Marketing, for many companies, visitors and engagement is often limited to founder / employee contacts and a few early users who are interested in the company, and its products or content.
In the early days of the Internet, early adopters were able to use SEO Tools to identify important keywords and include them frequently in content (also called keyword stuffing) and drive significant organic search related traffic to their website. Competition / content available on the Internet was also limited so success on the Internet came easily.
Fast forward to the present Google Search has gotten far more sophisticated, and is now able to automatically identify techniques such as keyword stuffing and penalize such attempts. Furthermore, competition is now fiercer and a lot more content is available on the Internet. Companies using relatively simple strategies, and attempting to drive organic traffic from day 1 are failing.
Other approaches such as Advertising, Public Relations & Press Placement are relatively more expensive (also more discretionary). While some companies simply can't afford it, others hold off until additional funding is available. Furthermore, the planned deprecation of 3rd party cookies will make it hard for some forms of advertising to target specific audiences and attribute accurately; making Internet Marketing even more challenging.
Why does Internet Marketing need AI?
Internet Marketing differs significantly from traditional marketing and sales, in its ability to execute multiple parallel targeted campaigns and customize content and bid for ads real-time; cost effectively. For example, it is possible to programmatically run a very effective campaign on Google Ads (an Internet Marketing tool) at US $500 per month. Semantic Brain AI (includes Natural Language Processing) is capable of continually augmenting Marketers ability to optimize keywords, ad titles, ad body, content and bid price on Google Ads to deliver superior results. Initial studies have demonstrated ability to increase click rates by more than 4x.
Semantic Brain AI continuously optimizes Website, Content and Organic Search results (e.g. SEO). However, the biggest gains are realized when performing Ad and Public Relations & Press Placement. Organic search generally adds the most value during the later stages, but should be optimized from day 1. This type of continuous analysis and frequent optimization along multiple dimensions is not possible by human alone. Semantic Brain provides a solution where humans set Internet Marketing strategy and AI makes continuous adjustments.
For example, Marketers may come up with 1 or 2 ads for a single product covering broader geographic areas, and set budget and maximum bid price per click based on analysis performed on a spreadsheet. Data maybe analyzed weekly and changes made to ads on a monthly basis.
Semantic Brain AI on the other hand would breakdown these ads into multiple smaller ads by geography and figure out optimum sets of keywords, ad titles, ad body, max bid price and budget per location. More data can be analyzed on a daily basis (as opposed to weekly), and ads can be better optimized on a weekly basis deliver better click rates, engagement, goal completion and customer conversion / revenue. The data can also be used to prove business scalability to secure additional funding and higher business valuation.
Solution Approach & Architecture
Driving more Traffic & Engagement
Semantic Brain AI includes NLP (Natural Language Processing) on unstructured data, and quantitative analysis of tabular and time series data. We leveraged Domain-Expertise-Centric AI (defined below) to more accurately deliver
Diagnostic Analytics: More precisely identifies what is happening and why it is happening. In contrast most tools only support Descriptive Analytics which only identifies what is happening.
Predictive Analytics: More accurately predicts current trends, and Ad / PR / Press Placement outcomes.
Prescriptive Analytics: Recommends keywords, content changes, ad title, ad body, max bid and ad budget across a portfolio to optimize Lead Generation, Engagement, Goal Completion and Conversion over time.
The table below provides additional details on how Traditional Marketing and Traditional Sales differ from Internet Marketing. Scope of Internet Marketing here is limited to approaches used by companies that make / provide knowledge oriented (as opposed to commodity) products and services. Upon review of this table it should be clearly evident how automated continuous adjustment of inputs is both feasible and beneficial in Internet Marketing.
Traditional Marketing | Traditional Sales | Internet Marketing |
Media: - Radio - TV | Media: - In Person - Phone | Media: - Website (incl. Blog, Forum) - Google Ads - LinkedIn / Twitter / YouTube |
Ability to target varies greatly, and is generally limited | Targeted process | Can customize and target (by topic, location, demographics, time, etc.) small groups and can also be personalized for individuals |
Content is created to emotionally appeal to a broader audience. Content creation requires relatively high investment. | Human centric and requires personal touch. Generic marketing content is often used to aid sales process. | Content is more likely to be informational, and can be composed to be appeal to broad or narrow (even personalized) audiences. |
Ability to measure results is limited. | Results are measurable | Very data rich environment |
Campaigns are relatively more expensive, and media placement prices are relatively static. | Scaling generally requires additional headcount. | Google Ads / LinkedIn campaigns can be successfully executed for as little as $500 per month. Real-time bidding and dynamic content creation are supported. |
Note: Facebook and Instagram are more popular social media for promoting commodity products / services, but are not covered for reasons explained below.
Value (or ROI potential) of Internet Marketing + AI will differ by industry and company. However, it can be generally assumed that broader the product and solution offerings and bigger the geographic footprint; the greater the savings that can be realized per lead generated, conversion and / or dollar of revenue. This can be achieved by leveraging AI's capability to cost effectively customize content and optimize bidding over geography and time.
Companies can maximize the value of Internet Marketing + AI by executing the following steps over time:
1) Assess website and social media structure and content;
2) Understand current activity by analyzing Audience, Acquisition, Behaviour and Conversions
3) Measure brand equity using SEO Tools and evaluate search terms (i.e., keywords), backlinks and referral traffic
4) Initiate $500 / month ad campaign with Google Ads (optionally add LinkedIn Ads)
5a) Augment existing content and create new ones
5b) Increase ad spend over time
5c) Engage in more paid Public Relations & Press Placement
Notes:
AI needs to be used for content optimization, not just for content / ad placement
Attributing Lead Generation / Conversions / Revenue to Artificial Intelligence: Let's assume there are two scenarios. Scenario #1 is completely Human driven, has a $2,000 Ad budget, runs over 1 months and has 5 ad campaigns, and produces 500 clicks. Scenario #2 is Human + AI driven, also has the ad budget of $2,000 and duration of1 month; however, leveraging AI, 20 ad campaigns are run and campaigns are updated weekly without additional staffing resulting in 1000 clicks. In Scenario #2, 500 clicks would be attributed to Internet Marketing; and 500 clicks would be attributed to AI.
Internet Marketing (e.g. website, LinkedIn presence) during the initial stages of a company (i.e. startup phase) adds credibility to its marketing and sales efforts. Furthermore, the data obtained can be leveraged to secure funding. The above chart does not capture these additional value propositions.
Build more Internet Brand Equity
We are increasingly living in a knowledge economy where business growth is dependent on the quantity, quality, and accessibility of the information available, rather than the means of production.
Internet Marketing is a key capability of the knowledge economy which enables companies to increase revenues faster, market and sell more efficiently, and achieve higher ROI. Broadly speaking, companies executing Internet Marketing fall into two categories
1) Companies that are involved in re-selling commodity products and services. Many retail outlets and e-Commerce companies fall into this category.
2) Companies that make / provide knowledge oriented products and services.
Semantic Brain's Internet Marketing product (Asperios Marketing) is more focused on second category of companies. These companies have significant know-how that can be converted into strategic content, to drive organic search traffic and to build communities and forums that facilitate exponential growth.
Semantic Brain delivers Internet Brand Equity boost by integrating its Ad and PR / Press Placement Optimization AI with
1) SEO: Semantic Brain's AI (including NLP) also analyses Content, Search Terms (Keywords) and Backlinks that are not a part of the new Content / Ads. This information is used to optimize and organize search traffic on an ongoing basis.
2) Website Performance Improvement: Google plans to make page experience an official Google ranking factor. Semantic Brain incorporates this in its analysis, and makes website performance recommendations.
Note: Marketing & Sales Spend is not Ad spend, it is total Marketing & Sales spend. On the Internet, as brand equity increases over time, Ads may account for a small fraction (e.g. < 1%) of Internet traffic.
Case for Domain-Expertise-Centric AI
Our Research & Development efforts have helped us realize that Domain-Expertise-Centric AI delivers superior results when developing business (marketing and Finance) solutions. Furthermore, Domain-Expertise-Centric AI is particularly valuable in smaller or younger (Start-ups and Scale-ups) organizations which have limited data and infrastructure.
For example, using EDA (Exploratory Data Analysis), marketers can relatively easily understand Cause & Effect in marketing data. However, performing EDA on digital images or video is unlikely to provide significant insight. Once a Marketer understands cause and effect s/he can formulate Business Hypotheses, perform Feature Engineering, derive Explainable Architectures (these reflect cause & effect) and deliver better results.
Semantic Brain uses BizML (A Machine Learning framework which is both process and software) which is based on Domain-Expertise-Centric AI and is used to develop AI Microservices for business applications. BizML is presently being used by Semantic Brain internally to develop marketing and finance solutions for its customers and internal use. The tool will be made available for direct customer use as additional software building blocks become available and as the user interface matures.
Domain-Expertise-Centric AI in comparison to Model-Centric / Data-Centric AI:
Delivers 5% to 20% increase in accuracy
Requires 50% less time
Requires 70% less compute
Model-Centric AI | Data-Centric AI | Domain-Expertise-Centric AI |
Many applications | Many applications | Limited to quantitative data (tabular and time series) and NLP |
Best At: Computer Vision, Speech Recognition, Natural Language Processing | Even Better At: Computer Vision for Manufacturing applications | Best At: Business applications - especially marketing and finance |
Big data driven | Requires less data. Greater emphasis on data quality as opposed to quantity. | Requires least amount of data. Domain knowledge and Rules supplement data. |
Focus on Model Architecture and Hyper Parameter tuning, and automating these aspects. | Focusing changing or improving the data. For example, use higher equality images with more accurate labeling. Automate data quality improvement. | Start with Exploratory Data Analysis and understanding Cause & Effect. Achieve superior results using Feature Engineering and Explainable Architectures. |
Best at solving complex problems that can't be solved using conventional programming | Similar to Model-Centric AI | Best for improving augmenting decision making |
Delivers biggest break-throughs in AI | Delivers State-of-the-Art in Industry (especially Computer Vision) | Delivers State-of-the-Art in Industry (especially Business Applications) |
Conclusion & Recommendations
Internet Marketing presents significant opportunities to companies, however, maximizing those opportunities requires better tooling. In addition, what is required is an effective Human and AI partnership and feedback loop. Companies making / providing knowledge oriented products and services, can leverage Internet Marketing + AI to substantially increase Traffic, Engagement, Conversions and Internet Brand Equity over time.
We recommend starting out by performing automated Site Audit (e.g., Performance, Errors, Meta-data) and Internet Brand Equity (Backlinks, Referral, Traffic Trends) assessment. Subsequently:
Augment content and improve website to boost Google ranking
Manage Backlinks and Referring Domains
Experiment with Google Ads with a $500 per month budget
Optionally experiment with social media ads (LinkedIn is top choice) with a $500 per month budget
Repeat 1 & 2, and increase ad budget to run multiple campaigns possibly over many media platforms
Leverage learnings and apply it to Public Relations and Press Placement
Capitalize on Internet Brand Equity
Comentarios